# Business Teaching
Source: https://help.wonka.chat/en/ai-agents/business-teaching

Provide business context to agents for accurate, company-specific responses.

Teaching your business context to agents ensures they understand your company's terminology, processes, and policies. This enables agents to deliver responses grounded in your actual operations instead of relying on generic assumptions.

<Info>
  Business teaching helps agents use your exact terminology, follow your processes, and align with your company policies.
</Info>

***

## Why Teach Your Business Context?

Without business context, agents make generic assumptions. With it, they provide accurate, company-specific responses.

**Example - Writing a customer support email:**

**Without context:** "Most companies offer refunds within 14-30 days. Check your company's specific policy."

**With context:** "Our refund policy allows requests within 30 days, processed in 3-5 business days. Would you like me to initiate one?"

### Key Benefits

<CardGroup>
  <Card title="Accurate responses" icon="check-circle">
    Agents use your exact terms, policies, and processes
  </Card>

  <Card title="Reduced errors" icon="shield-check">
    Grounding in your information prevents hallucinations (AI making up incorrect details)
  </Card>

  <Card title="Brand consistency" icon="palette">
    Agents maintain your voice, tone, and communication style
  </Card>

  <Card title="Better outcomes" icon="trophy">
    Team members get reliable, actionable information
  </Card>
</CardGroup>

***

## How to Provide Business Context

### Add Context to Agent Instructions

Include company-specific information directly in agent instructions. Start with essentials and expand as needed.

<AccordionGroup>
  <Accordion title="Voice and Tone Guidelines" icon="message">
    Define how agents should communicate to match your brand.

    **Example:**

    ```
    Communication style:
    - Professional but friendly, never overly casual
    - Use active voice and direct language
    - Address customers by first name
    - Keep responses concise and actionable
    ```

    <Tip>
      Include specific phrases to use or avoid. This ensures consistency across all agent interactions.
    </Tip>
  </Accordion>

  <Accordion title="Terminology and Glossaries" icon="book">
    Define company-specific terms, product names, and acronyms.

    **Example:**

    ```
    Key terms:
    - "Premium tier" = enterprise subscription level
    - "Quick-start" = our 15-minute onboarding process
    - MRR = Monthly Recurring Revenue
    - CS = Customer Success (not Customer Support)
    ```

    Clear definitions prevent misunderstandings and ensure agents use your language.
  </Accordion>

  <Accordion title="Policy Integration" icon="shield">
    Include relevant policies so agents can answer questions accurately.

    **Example:**

    ```
    Key policies:
    - Refunds: Full refund within 30 days, processed in 3-5 business days
    - Support hours: 9 AM-6 PM EST, Monday-Friday
    - Response SLA: 4 business hours for standard, 1 hour for enterprise
    - Escalation: Issues over \$500 require manager approval
    ```

    <Warning>
      Always verify agents apply policies correctly before deploying to your team. Test with realistic scenarios.
    </Warning>
  </Accordion>

  <Accordion title="Process Instructions" icon="list-check">
    Document step-by-step workflows agents should follow.

    **Example:**

    ```
    Customer onboarding process:
    1. Send welcome email with login credentials
    2. Schedule orientation call within 48 hours
    3. Assign dedicated success manager
    4. Provide access to training materials
    5. Follow up after 7 days
    ```

    Clear processes ensure consistency and completeness in agent actions.
  </Accordion>
</AccordionGroup>

**Best practice:** Update instructions when policies or processes change. Outdated context leads to incorrect information.

***

### Upload Documents for Deep Context

For extensive information like product documentation, policy manuals, or knowledge bases, upload documents that agents can automatically search and reference when answering questions.

<Frame>
  <img alt="Document upload interface" />
</Frame>

**Why upload documents?**

* Agents can access detailed information without cluttering instructions
* Perfect for comprehensive documentation that's too large for instructions
* Information stays current by updating documents anytime without changing agent instructions

**Best for:** Product documentation, policy manuals, process guides, training materials, knowledge bases

<Tip>
  Use instructions for critical information needed in every response. Use documents for extensive reference material that agents retrieve only when relevant.
</Tip>

<Accordion title="Technical Details: How Document Retrieval Works" icon="gear">
  When you upload documents, WonkaChat uses **RAG (Retrieval-Augmented Generation)** to make them searchable for agents.

  **How it works:**

  1. **Upload**: You provide company documents
  2. **Index**: Documents are processed and made searchable
  3. **Retrieve**: When relevant, agents automatically find information from documents
  4. **Generate**: Agents produce responses grounded in your actual documentation

  **Document preparation tips:**

  * **Use clear headings**: Well-structured sections improve retrieval accuracy
  * **Keep sections focused**: One topic per section
  * **Include context**: Write sections to be understandable independently (e.g., "Our Premium tier (\$99/month) includes..." not just "It includes...")
  * **Stay consistent**: Use the same terminology throughout

  **Security and privacy:**

  * Documents stored securely within your WonkaChat environment
  * Never used to train AI models
  * Accessible only to users with agent permissions

  **Learn more about RAG:** For deeper technical details, see [Anthropic's](https://www.anthropic.com/engineering/contextual-retrieval) or [OpenAI's](https://platform.openai.com/docs/guides/optimizing-llm-accuracy#retrieval-augmented-generation-rag) or [Google's](https://cloud.google.com/use-cases/retrieval-augmented-generation) RAG Guides.
</Accordion>

***

## Examples by Use Case

<AccordionGroup>
  <Accordion title="Customer Support Agent" icon="headset">
    **Context needed:** Product features, support policies, troubleshooting steps, communication guidelines

    **Example instruction snippet:**

    ```
    Support guidelines:
    - Always verify account before sharing details
    - Use knowledge base articles for technical issues
    - Escalate billing issues over 500$ to finance team
    - Response tone: Empathetic, solution-focused, professional
    ```
  </Accordion>

  <Accordion title="Email Assistant" icon="envelope">
    **Context needed:** Company voice, common email types, approval workflows, internal terminology

    **Example instruction snippet:**

    ```
    Email style:
    - Internal emails: Direct and concise
    - External emails: Professional and courteous
    - Always include relevant stakeholders in CC when appropriate
    - Use company signature format: [Name] | [Title] | [Department]
    ```
  </Accordion>

  <Accordion title="Sales Assistant" icon="chart-line">
    **Context needed:** Product positioning, pricing, target audience, sales processes, competitive info

    **Example instruction snippet:**

    ```
    Sales context:
    - Target: Mid-market companies (50-500 employees)
    - Key value prop: 40% time savings on project management
    - Pricing: Start at 49$/user/month
    - Never discount without manager approval
    ```
  </Accordion>
</AccordionGroup>


# Creating Your First Agent
Source: https://help.wonka.chat/en/ai-agents/creating-your-first-agent

Build custom AI agents configured for your specific workflows with the right model, instructions, and tools.

Create specialized AI agents tailored to your workflows. This guide covers agent creation from initial setup through deployment.

<Info>
  Building an AI Agent means defining its role, writing instructions, selecting a model, and optionally connecting tools.
</Info>

<Tabs>
  <Tab title="Getting Started">
    ## Agent Creation - Video Explanation

    Explore the step-by-step instructions on creating AI agents by video. These two videos visually demonstrates the key decisions and actions you need to build, configure, and deploy your own custom agents.

    <AccordionGroup>
      <Accordion title="Basic Agent Creation" icon="user">
        ### Basic Agent:

        Watch how to create a basic agent with straightforward instructions. This helps you learning the fundamentals of agent creation.

        <iframe title="Creating a Basic Agent" />

        <Tip>
          Start with simple agents to understand the core concepts before moving to more complex agent with tool configurations.
        </Tip>
      </Accordion>

      <Accordion title="Tool Connected Agent Creation" icon="screwdriver-wrench">
        ### Tool Connected Agent:

        Watch how to create an agent that connects a tool to automate workflows. This is a practical example of cross-system integration.

        <iframe title="Creating a Tool Connected Agent" />

        <Tip>
          This example demonstrates how agents can connect other business systems to automate workflows that typically require manual data transfer.
        </Tip>
      </Accordion>
    </AccordionGroup>

    ## Agent Creation - Step by Step Explanation

    <Steps>
      <Step title="Before You Begin">
        Define your agent's purpose:

        * **What task** will this agent handle?
        * **How complex** is the workflow?
        * **Which tools** does it need?
      </Step>

      <Step title="Access Agent Creation">
        Navigate to the agent creation interface and click "Create New Agent."

        <Frame>
          <img alt="Agent creation starting point" />
        </Frame>
      </Step>

      <Step title="Name Your Agent">
        Choose a clear, descriptive name that indicates purpose.

        **Good:** "Customer Support Assistant," "Q4 Sales Data Analyzer"\
        **Avoid:** "Helper," "Agent 1"

        <Tip>
          If planning organization-wide sharing, choose names that make sense to all users.
        </Tip>
      </Step>

      <Step title="Write Your Instructions">
        Instructions specify your agent's role, expertise, tone, and behavior. Write clear, detailed instructions to ensure the agent consistently performs as intended.

        ### Basic Instructions Guide

        <AccordionGroup>
          <Accordion title="Role Definition">
            State what the agent is and does.

            **Example:** "You are a customer support specialist for our SaaS platform. Help users resolve technical issues, answer product questions, and guide them through common workflows."
          </Accordion>

          <Accordion title="Communication Style">
            Define tone and format.

            **Example:** "Communicate in a friendly, professional tone. Be concise but thorough. Use bullet points for clarity."
          </Accordion>

          <Accordion title="Operational Guidelines">
            Provide specific rules for how the agent should execute tasks and make decisions.

            **Example:** "When analyzing sales data, always compare against the previous quarter and year-over-year metrics. If data is incomplete, note the gaps explicitly rather than making assumptions."
          </Accordion>

          <Accordion title="Tool Usage Guidelines">
            Specify when and how the agent should use available tools.

            **Example:** "Use the CRM tool to retrieve customer information before making recommendations. When querying the database, limit results to the last 90 days unless specified otherwise."
          </Accordion>

          <Accordion title="Business Context">
            Include company-specific information.

            **Example:** "Support hours: 9 AM-6 PM EST. We offer three tiers: Basic, Pro, Enterprise."
          </Accordion>
        </AccordionGroup>

        <Check>
          Clear, detailed instructions are the foundation of agent effectiveness.
        </Check>
      </Step>

      <Step title="Choose an AI Model">
        Select the right model based on your agent's task complexity and usage frequency.

        ### Quick Selection Guide

        **For simple, frequent tasks:**

        * Use **GPT-4o mini** (most cost-effective) or **Gemini 2.5 Flash** (fast with large context)
        * Also consider: **Claude Haiku 4.5** (fast with near-frontier intelligence)
        * Example tasks: Email drafting, data lookup, simple formatting, high-volume operations

        **For complex reasoning:**

        * Use **Claude Opus 4.5**, **Claude Sonnet 4.5**, **GPT-5.2**, or **GPT-5** (advanced capabilities)
        * Example tasks: Multi-step analysis, strategic planning, multi-tool usage, complex coding

        **For long context:**

        * Use **Gemini 2.5 Pro** (1M context window) or **Gemini 2.5 Flash** (1M context, faster)
        * Example tasks: Document analysis, comprehensive research, processing large datasets

        <Tip>
          Start with a cost-effective model like **GPT-4o mini** or **Claude Haiku 4.5** for most agents. Upgrade to more capable models (Claude Opus 4.5, GPT-5.2, or Gemini 2.5 Pro) only if results don't meet your needs.
        </Tip>

        <Info>
          **Need detailed guidance?** See the [Model Selection Guide](#model-selection-guide) in the Advanced Knowledge tab for comprehensive comparisons, pricing, and use case recommendations.
        </Info>
      </Step>

      <Step title="Connect External Softwares (Optional)">
        Add MCP connection if your agent needs to interact with external systems.

        **Tool selection principles:**

        * Only add tools the agent actually needs
        * Too many tools can cause selection confusion
        * Ensure you have permissions for assigned tools

        <Warning>
          Agents can only access tools explicitly assigned to them and inherit your personal permissions. See the <a href="/en/security-governance/access-control#tool-restrictions">Tool Restrictions</a> section in Access Control for more details.
        </Warning>

        **Example configurations:**

        * **Customer Support:** CRM, ticket system, knowledge base, email
        * **Data Analysis:** Database connections, visualization, spreadsheets
        * **Content:** Document storage, publishing platform, asset library
      </Step>

      <Step title="Review and Test">
        Test thoroughly before deployment.

        **Test:**

        * Core functionality with typical requests
        * Edge cases and incomplete information
        * Error handling when tools fail
        * Consistency across multiple similar requests

        <Check>
          Thorough testing ensures users get reliable results.
        </Check>
      </Step>

      <Step title="Deploy or Share">
        **Personal use:** Start using immediately, refine based on real usage

        **Organization sharing:** Contact administrator, provide context on purpose and value, await review and approval

        <Info>
          See [Access Control](/security-governance/access-control) for agent sharing details.
        </Info>
      </Step>
    </Steps>

    ## Best Practices

    <CardGroup>
      <Card title="Be Explicit" icon="list-check">
        Define specific goals, actions and expected results for each step to reduce ambiguity.
      </Card>

      <Card title="Include Edge Cases" icon="diamond-exclamation">
        Anticipate unusual situations and provide handling guidance.
      </Card>

      <Card title="Provide Context" icon="lightbulb">
        Give agents background information about your business processes and terminology.
      </Card>

      <Card title="Iterate" icon="rotate">
        Start simple, test thoroughly, refine based on performance and real world use cases.
      </Card>
    </CardGroup>

    ## Common Mistakes

    <AccordionGroup>
      <Accordion title="Vague Instructions">
        **Problem:** "You are a helpful assistant" lacks guidance.\
        **Solution:** Be specific about role, expertise, tone, and guidelines.
      </Accordion>

      <Accordion title="Too Many Responsibilities">
        **Problem:** One agent handling unrelated tasks like customer support and data analysis.\
        **Solution:** Create focused agents for specific workflows.
      </Accordion>

      <Accordion title="Tool Overload">
        **Problem:** Assigning every available tool.\
        **Solution:** Only add necessary tools to avoid selection confusion.
      </Accordion>

      <Accordion title="Insufficient Testing">
        **Problem:** Deploying without thorough testing.\
        **Solution:** Test extensively with realistic scenarios first.
      </Accordion>
    </AccordionGroup>
  </Tab>

  <Tab title="Advanced Knowledge">
    ## AI Agent Instruction Crash Course

    This guide introduces five key elements that help create more effective agents. While these fundamentals provide a solid starting point, they represent examples of good practices rather than a complete framework. However, building truly reliable agents requires deeper exploration and iteration beyond these basics.

    <AccordionGroup>
      <Accordion title="1. Persona: Who is your agent?" icon="user">
        Define your agent's role, expertise, and perspective to shape response style and depth.

        **Example:**

        ```
        You are a senior financial advisor with 20 years of experience 
        specializing in retirement planning for conservative investors.
        ```

        <Tip>
          **Specific beats generic:** "You are a helpful assistant" provides no guidance. Define expertise, experience level, and specialization.
        </Tip>
      </Accordion>

      <Accordion title="2. Goals: What should it achieve?" icon="bullseye">
        Clearly state the objective and desired outcome for every interaction.

        **Example:**

        ```
        Your goal is to analyze sales data from the last quarter, 
        identify the top 3 declining products, and recommend specific 
        actions to reverse the trend. Present findings in a concise 
        executive summary format.
        ```

        <Check>
          Specific, measurable goals produce consistent, actionable results.
        </Check>
      </Accordion>

      <Accordion title="3. Tools: What can it use?" icon="wrench">
        Specify which tools to use and when. Provide clear guidelines for tool selection when multiple options exist.

        **Example:**

        ```
        Available tools:
        - database_query: Use for all customer data lookups
        - crm_api: Use only when database is unavailable
        - email_tool: Use for sending notifications

        Always query the database before making recommendations.
        Never update records without explicit user confirmation.
        ```

        <Warning>
          Too many similar tools confuse agents. If your agent struggles with tool selection, provide clearer usage criteria or reduce tool count.
        </Warning>
      </Accordion>

      <Accordion title="4. Constraints: What are the boundaries?" icon="shield-halved">
        Set clear boundaries for format, tone, length, permissions, and prohibited actions.

        **Example:**

        ```
        Constraints:
        - Keep responses under 200 words unless analysis requires detail
        - Use professional but friendly tone
        - Never make promises about delivery dates
        - If customer requests exceed \$500, escalate to human approval
        - Format all data tables in markdown
        ```

        <Info>
          Constraints prevent common failure modes. Add new constraints as you discover edge cases in production.
        </Info>
      </Accordion>

      <Accordion title="5. Context: What background matters?" icon="book-open">
        Provide relevant background information, business rules, and domain-specific knowledge.

        **Example:**

        ```
        Company context:
        - We serve B2B SaaS customers with 3 pricing tiers
        - Standard support hours: 9 AM-6 PM EST, Mon-Fri
        - Enterprise customers get priority within 2 hours
        - Our refund policy allows 30-day money-back guarantee
        - Common abbreviations: ARR (Annual Recurring Revenue), 
          MRR (Monthly Recurring Revenue)
        ```

        <Tip>
          Include terminology, acronyms, business rules, and any domain knowledge the agent needs to interpret requests correctly.
        </Tip>
      </Accordion>
    </AccordionGroup>

    ### Quick Framework

    Use this template as your starting point:

    ```
    [PERSONA]
    You are a [specific role] with [expertise/experience].

    [GOALS]
    Your objective is to [primary goal]. Focus on [key outcomes].

    [CONTEXT]
    Important background: [domain knowledge, business rules, terminology]

    [TOOLS]
    Available tools: [list]
    Use [tool A] when [condition].
    Use [tool B] for [scenario].

    [CONSTRAINTS]
    - Format: [requirements]
    - Tone: [style guidelines]
    - Never: [prohibited actions]
    - Always: [required behaviors]
    - Escalate when: [escalation criteria]
    ```

    <CardGroup>
      <Card title="Start Simple" icon="seedling">
        Begin with persona, goal, and constraints. Add complexity only as needed.
      </Card>

      <Card title="Test Thoroughly" icon="flask">
        Run your prompts against edge cases. Refine based on actual failures.
      </Card>

      <Card title="Iterate" icon="arrows-rotate">
        Prompts improve through real usage. Update based on production learnings.
      </Card>

      <Card title="Learn More" icon="graduation-cap">
        Explore more advanced techniques with [OpenAI's](https://platform.openai.com/docs/guides/prompt-engineering), [Anthropic's](https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/overview), or even [Google's](https://ai.google.dev/gemini-api/docs/prompting-intro) prompting guides.
      </Card>
    </CardGroup>

    <Info>
      **Want deeper knowledge?** Search for "prompt engineering best practices", "AI agent design patterns" and "tool calling strategies" to explore advanced techniques.
    </Info>

    ***

    ## Model Selection Guide

    Quick comparison to help you choose the right model. For detailed specs and benchmarks, visit [OpenAI](https://openai.com), [Anthropic](https://anthropic.com), or [Google DeepMind](https://deepmind.google/technologies/gemini/).

    ### Model Comparison

    <AccordionGroup>
      <Accordion title="OpenAI Models" icon="openai">
        | Model           | Context | Cost (Input/Output)    | Best For                                              |
        | --------------- | ------- | ---------------------- | ----------------------------------------------------- |
        | **GPT-5.2**     | 128K+   | Premium pricing        | Latest model, advanced reasoning, complex tasks       |
        | **GPT-5**       | 128K+   | \$2.50 / \$10 per 1M   | Advanced coding, agentic workflows, complex reasoning |
        | **GPT-4o**      | 128K    | \$2.50 / \$10 per 1M   | General-purpose, balanced tasks, multimodal           |
        | **GPT-4o mini** | 128K    | \$0.15 / \$0.60 per 1M | High-volume, cost-sensitive tasks                     |

        **Key Features:** Strong general performance, proven reliability, multimodal capabilities
      </Accordion>

      <Accordion title="Anthropic (Claude) Models" icon="brain">
        | Model                 | Context | Cost (Input/Output) | Best For                                                 |
        | --------------------- | ------- | ------------------- | -------------------------------------------------------- |
        | **Claude Opus 4.5**   | 200K    | \$5 / \$25 per 1M   | Highest capability, complex reasoning, advanced coding   |
        | **Claude Sonnet 4.5** | 200K-1M | \$3 / \$15 per 1M   | Balanced performance, coding, agents (1M beta)           |
        | **Claude Sonnet 4**   | 200K    | \$3 / \$15 per 1M   | Strong coding and reasoning                              |
        | **Claude Haiku 4.5**  | 200K    | \$1 / \$5 per 1M    | Fastest, most cost-efficient, near-frontier intelligence |
        | **Claude Sonnet 3.7** | 200K    | \$3 / \$15 per 1M   | General tasks, proven performance                        |

        **Key Features:** Industry-leading coding, strong reasoning, flexible context windows
      </Accordion>

      <Accordion title="Google (Gemini) Models" icon="google">
        | Model                      | Context | Cost (Input/Output)          | Best For                                           |
        | -------------------------- | ------- | ---------------------------- | -------------------------------------------------- |
        | **Gemini 3 Pro Preview**   | 1M+     | Preview pricing              | Next-gen capabilities, advanced analysis (preview) |
        | **Gemini 3 Flash Preview** | 1M+     | Preview pricing              | Next-gen speed and efficiency (preview)            |
        | **Gemini 2.5 Pro**         | 1M      | \$1.25-2.50 / \$10-15 per 1M | Long documents, complex analysis                   |
        | **Gemini 2.5 Flash**       | 1M      | \$0.30 / \$2.50 per 1M       | Speed, real-time, cost-efficiency                  |

        **Key Features:** Largest context window (1M+ tokens), optimized for speed and document analysis
      </Accordion>

      <Accordion title="Mistral Models" icon="wind">
        | Model                     | Context | Cost (Input/Output) | Best For                        |
        | ------------------------- | ------- | ------------------- | ------------------------------- |
        | **Mistral Large Latest**  | 128K    | Competitive pricing | Strong performance, European AI |
        | **Mistral Medium Latest** | 32K     | Lower cost          | Balanced tasks, cost-effective  |

        **Key Features:** Open-source foundation, European data sovereignty option
      </Accordion>
    </AccordionGroup>

    <Info>
      **Pro Models Available:** GPT-5.2, GPT-5, Claude Opus 4.5, Claude Sonnet 4.5, Claude Sonnet 4, and Gemini 2.5 Pro all support advanced reasoning and complex task handling.
    </Info>

    ### Quick Selection by Task

    <AccordionGroup>
      <Accordion title="Coding & Development" icon="code">
        **Pro:** Claude Opus 4.5, Claude Sonnet 4.5, GPT-5.2, or GPT-5 (advanced coding, large context)\
        **Budget:** GPT-4o mini (\$0.15/\$0.60) or Claude Haiku 4.5 (\$1/\$5)

        <Check>
          Use Pro models for production code or complex refactoring. Use budget models for scripts and simple tasks.
        </Check>
      </Accordion>

      <Accordion title="Long Documents & Analysis" icon="chart-bar">
        **Best:** Gemini 2.5 Pro (1M context) or Gemini 3 Pro Preview (1M+ context)\
        **Alternative:** Claude Sonnet 4.5 (200K-1M beta), GPT-5.2, or GPT-5

        <Tip>
          Gemini models handle documents over 200K tokens best with their 1M+ context windows. For deep reasoning on shorter documents, use Claude Opus 4.5, Claude Sonnet 4.5, or GPT-5.2.
        </Tip>
      </Accordion>

      <Accordion title="High-Volume Tasks" icon="bolt">
        **Best:** GPT-4o mini (lowest cost at \$0.15/\$0.60)\
        **Alternative:** Gemini 2.5 Flash (faster, 1M context, \$0.30/\$2.50) or Claude Haiku 4.5 (\$1/\$5)

        <Warning>
          Cost differences compound quickly at scale. GPT-4o mini and Claude Haiku 4.5 can save thousands monthly on high-volume agents.
        </Warning>
      </Accordion>

      <Accordion title="Complex Reasoning & Planning" icon="brain">
        **Premium Models for Complex Tasks:** GPT-5.2, GPT-5, Claude Opus 4.5, Claude Sonnet 4.5, Gemini 2.5 Pro

        **Choose based on needs:**

        * **Long context:** Gemini 2.5 Pro (1M tokens) or Claude Sonnet 4.5 (1M beta)
        * **Best coding:** Claude Opus 4.5 or Claude Sonnet 4.5 (industry-leading)
        * **Latest capabilities:** GPT-5.2 (newest OpenAI model)
        * **Balanced:** GPT-5 or Claude Sonnet 4.5
        * **Next-gen preview:** Gemini 3 Pro Preview

        <Info>
          Use premium models for genuinely complex reasoning, multi-step workflows, and agentic tasks. For simpler tasks, fast models like GPT-4o mini or Claude Haiku 4.5 suffice.
        </Info>
      </Accordion>

      <Accordion title="Speed & Real-Time" icon="gauge-high">
        **Fast:** Gemini 2.5 Flash (optimized latency, 1M context) or Gemini 3 Flash Preview (next-gen speed)\
        **Fastest & Cheapest:** GPT-4o mini (lowest cost, great speed) or Claude Haiku 4.5 (near-frontier intelligence)

        Real-time chat, live data processing, interactive apps, streaming
      </Accordion>
    </AccordionGroup>

    ### Quick Decision Tree

    ```mermaid theme={null}
    graph TD
        A[What's your priority?] --> B[Coding]
        A --> C[Long Documents]
        A --> D[Cost/Volume]
        A --> E[Speed]
        A --> F[Complex Reasoning]
        
        B --> B1[Claude Opus 4.5<br/>Claude Sonnet 4.5<br/>or GPT-5.2]
        C --> C1[Gemini 2.5 Pro<br/>Gemini 3 Pro Preview<br/>1M+ context]
        D --> D1[GPT-4o mini<br/>Claude Haiku 4.5<br/>Cost: 0.15-1.00]
        E --> E1[Gemini 2.5 Flash<br/>GPT-4o mini<br/>Claude Haiku 4.5]
        F --> F1[GPT-5.2, GPT-5<br/>Claude Opus 4.5<br/>Claude Sonnet 4.5<br/>Gemini 2.5 Pro]
        
        style B1 fill:#7B68EE,stroke:#5A4DB5,stroke-width:2px,color:#fff
        style C1 fill:#FFA94D,stroke:#CC8539,stroke-width:2px,color:#fff
        style D1 fill:#51CF66,stroke:#40A852,stroke-width:2px,color:#fff
        style E1 fill:#74C0FC,stroke:#4DABF7,stroke-width:2px,color:#fff
        style F1 fill:#4A90E2,stroke:#2E5C8A,stroke-width:2px,color:#fff
    ```

    ***

    ## Model Parameters

    Fine-tune agent behavior with these parameters. Defaults work for most cases, you should adjust only when needed.

    <AccordionGroup>
      <Accordion title="Max Context Tokens" icon="inbox">
        **What it does:** Maximum input length the model can process (includes your prompt + conversation history)

        **When to adjust:**

        * Your agent needs to process long documents or extensive conversation history
        * Default usually sufficient for most agents

        <Tip>
          Check your model's context limit: GPT-4o (128K), GPT-5/5.2 (128K+), Claude models (200K-1M), Gemini models (1M+)
        </Tip>
      </Accordion>

      <Accordion title="Max Output Tokens" icon="message">
        **What it does:** Limits how long the agent's responses can be

        **Recommended values:**

        * **Short responses:** 100-300 tokens (\~75-225 words)
        * **Standard responses:** 500-1000 tokens (\~375-750 words)
        * **Long-form content:** 2000+ tokens

        <Warning>
          Lower values prevent overly verbose responses and reduce costs. Increase only if responses are being cut off.
        </Warning>
      </Accordion>

      <Accordion title="Temperature (0.0-2.0)" icon="temperature-half">
        **What it does:** Controls creativity vs consistency in responses

        **Recommended values:**

        * **0.0-0.3:** Factual tasks, data extraction, precise answers. It produces consistent, predictable, and focused responses
        * **0.7-1.2:** Creative writing, brainstorming, varied responses. It generates more creative, diverse, and unpredictable outputs
        * **1.0:** Balanced setting with good mix of coherence and creativity for most general-purpose tasks (default)
        * **1.5-2.0:** Highly creative/unpredictable (rarely needed)

        <Check>
          Lower = more predictable and focused. Higher = more creative and diverse.
        </Check>
      </Accordion>

      <Accordion title="Top P (0.0-1.0)" icon="filter">
        **What it does:** Alternative to temperature. It limits word choices to most probable options

        **Recommended values:**

        * **0.1-0.5:** Very focused, consistent outputs
        * **0.6-0.8:** Small reduction of diversity
        * **0.95-1.0:** Maximum diversity (default)

        <Info>
          **Important:** Adjust Temperature OR Top P, not both at the same time. Adjusting both simultaneously can produce unexpected results.
        </Info>
      </Accordion>

      <Accordion title="Frequency Penalty (-2.0 to 2.0)" icon="repeat">
        **What it does:** Reduces word/phrase repetition based on how often they've already appeared

        **Recommended values:**

        * **0.0:** No penalty (default, allows natural repetition)
        * **0.3-0.7:** Reduces repetitive phrasing
        * **1.0-2.0:** Strongly discourages repetition (may affect quality)

        **Use when:** Agent repeats the same phrases, words, or sentence structures too often
      </Accordion>

      <Accordion title="Presence Penalty (-2.0 to 2.0)" icon="lightbulb">
        **What it does:** Encourages introducing new topics/concepts rather than staying on current ones

        **Recommended values:**

        * **0.0:** No penalty (default, natural topic continuation)
        * **0.3-0.7:** Encourages topic diversity
        * **1.0-2.0:** Strongly pushes for new topics (may lose coherence)

        **Use when:** You want comprehensive coverage of multiple aspects or agent gets stuck on one topic
      </Accordion>

      <Accordion title="Stop Sequences" icon="octagon">
        **What it does:** Custom text strings that tell the model to stop generating

        **How to use:**

        * Add specific phrases like `---`, `END`, or custom markers
        * Model stops immediately when it generates any stop sequence
        * Useful for structured outputs or preventing unwanted continuation

        **Example:** Use `\n\n---\n\n` to stop after completing a section
      </Accordion>
    </AccordionGroup>

    <CardGroup>
      <Card title="Start with Defaults" icon="circle-check">
        Default settings work well for most use cases. Only adjust if you have specific requirements.
      </Card>

      <Card title="Adjust One at a Time" icon="sliders">
        Change one parameter, test, then adjust another. Keep notes on what worked and what didn't
      </Card>
    </CardGroup>

    ***

    ## Security Best Practices

    Follow these essential security practices when building agents:

    ### Limit Tool Access

    Only grant tools that are absolutely necessary for the agent's specific task. If your agent drafts emails using Outlook or Gmail, remove the "send email" capability to prevent accidental sends. The agent can prepare drafts while you maintain final control over sending.

    <Tip>
      Fewer tools mean fewer potential bad behavior risks and more predictable output.
    </Tip>

    ### Isolate Integrations

    Avoid connecting systems that shouldn't interact with each other in the same agent. If you have two software systems that should remain separate, don't add both integrations to the same agent. Create separate agents for different domains to maintain clear boundaries.

    ### Test Iteratively

    Begin with minimal permissions and test extensively before expanding capabilities. Avoid deploying agents with full permissions initially. Unexpected behavior is common and should be identified during testing.

    **Recommended approach:**

    * Begin with read-only access when possible
    * Test with non-production data first
    * Gradually add tools as you verify behavior
    * Monitor agent actions closely during initial deployment

    <Warning>
      Don't try to build everything in one go. Test incrementally to avoid surprising and unwanted results.
    </Warning>

    ### Implement Safeguards

    Add protective measures for sensitive operations:

    * Require human confirmation for high-impact actions (Safe Mode)
    * Define clear boundaries for agent scope and authority
    * Test changes in a staging environment before production deployment

    <Info>
      See [Security & Governance](/security-governance/security-overview) for comprehensive security practices.
    </Info>
  </Tab>
</Tabs>


# Finding and Accessing Agents
Source: https://help.wonka.chat/en/ai-agents/finding-and-accessing-agents

Navigate WonkaChat to locate and access AI agents for your workflows.

Agents are located in two places: the creation interface for building and managing agents, and the usage interface for starting conversations with agents.

***

## Locating Agents

### Agent Usage Interface

There are two ways to access and start conversations with agents:

#### 1. Agent Marketplace

Access the agent marketplace from the top left of your screen. The marketplace organizes agents by categories, making it easy to discover and browse available agents. It is especially useful when you have many agents to choose from.

<Frame>
  <img alt="Agent marketplace on top left" />
</Frame>

**Features:**

* Browse agents by category
* Search functionality for quick discovery
* View detailed agent descriptions

#### 2. My Agents Quick Access

Access your agents directly from the model selection dropdown at the top of the screen. At the bottom of the dropdown, you'll see "My Agents". You can hover over it to reveal a list of all your available agents with a search field.

<Frame>
  <img alt="My Agents dropdown in model selector" />
</Frame>

**Features:**

* Quick access to your frequently used agents
* Integrated search field for fast agent lookup
* No need to navigate away from your current screen

<Tip>
  Quick usage: Agents can be invoked by typing `@` followed by their name directly from any conversation.
</Tip>

<Frame>
  <img alt="Example of using an agent in a conversation" />
</Frame>

***

### Agent Creation & Management

In your WonkaChat interface, find the **Agent Builder** on the right side of the screen. This panel allows you to create and manage your AI agents.

<Frame>
  <img alt="Agent Builder panel" />
</Frame>

<Tip>
  If you don't see the right side panel, click the arrow icon in the middle-right of your screen to expand the sidebar.
</Tip>

**What you can do:**

* Create new agents with custom instructions
* Give files and capabilities to agents
* Assign Tools (External Connections) to agents
* Modify existing agents configurations

<Check>
  All users can create personal agents. Organization-wide sharing requires administrator approval.
</Check>

***

## Next Steps

<CardGroup>
  <Card title="Using Existing Agents" icon="play" href="/en/ai-agents/using-existing-agents">
    Learn how to get the best results from pre-built agents.
  </Card>

  <Card title="Creating Your First Agent" icon="plus-circle" href="/en/ai-agents/creating-your-first-agent">
    Build your own agent tailored to your workflows.
  </Card>
</CardGroup>


# What are AI Agents?
Source: https://help.wonka.chat/en/ai-agents/introduction

Understand what AI agents are, how do they work and how they provide consistent & specialised assistance.

AI agents are specialized AI assistants with persistent instructions and dedicated tool access that deliver consistent, reliable results.

<Info>
  Agents are like AI team members with specific roles and expertise. They remember their job across all conversations.
</Info>

***

## How Agents Differ from Regular Chat

<Tabs>
  <Tab title="Getting Started">
    Regular chat starts fresh each time. Agents have persistent instructions, roles, and tools that apply automatically to every conversation.

    **Key advantages:**

    <CardGroup>
      <Card title="Persistent Instructions" icon="file-lines">
        Give agents permanent guidelines, roles, and expertise that apply to every conversation automatically.
      </Card>

      <Card title="Consistent Behavior" icon="equals">
        Agents follow the same approach every time, delivering predictable, reliable results across all interactions.
      </Card>

      <Card title="Dedicated Tools" icon="wrench">
        Equip agents with specific external connections (MCP) they'll use automatically when needed for their tasks.
      </Card>

      <Card title="Specialized Roles" icon="user-tie">
        Define agent personalities, tone, and expertise to match specific workflows or team functions.
      </Card>
    </CardGroup>

    <Check>
      Configure once, use everywhere. Agents maintain their role and instructions across all conversations.
    </Check>

    ## When to Use Agents

    Use agents for:

    * **Recurring workflows**: Tasks you perform regularly with consistent requirements
    * **Specialized expertise**: Domain-specific knowledge applied consistently
    * **Cross-tool workflows**: Complex tasks spanning multiple systems
    * **Team standardization**: Shared workflows across your organization

    ## Agent Examples

    Practical agents you can create for common business workflows:

    <AccordionGroup>
      <Accordion title="Email Assistant" icon="envelope">
        Automatically review your last unread emails, summarize key information, flag urgent messages, and draft responses based on your communication style.

        **Tools needed:** Email (Outlook, Gmail)\
        **Complexity:** Simple
      </Accordion>

      <Accordion title="Meeting Scheduler" icon="calendar">
        Find available time slots across multiple calendars, schedule meetings with participants, send invitations, and create meeting agendas based on meeting context.

        **Tools needed:** Calendar, Email\
        **Complexity:** Simple
      </Accordion>

      <Accordion title="Customer Support Agent" icon="headset">
        Handle customer inquiries by accessing your CRM, checking order status, processing refund requests within defined limits, and escalating complex issues to human agents.

        **Tools needed:** CRM, Order Management, Email\
        **Complexity:** Moderate
      </Accordion>

      <Accordion title="Data Analysis Agent" icon="chart-line">
        Query databases for sales data, perform cohort analysis, identify trends, and generate reports with visualizations based on your reporting standards.

        **Tools needed:** Database, Spreadsheet, Visualization tools\
        **Complexity:** Moderate
      </Accordion>

      <Accordion title="Lead Qualification Agent" icon="user-check">
        Review new leads from multiple sources, enrich data from public sources, score based on your criteria, assign to appropriate sales team members, and send personalized outreach.

        **Tools needed:** CRM, Email, Data enrichment\
        **Complexity:** Complex
      </Accordion>

      <Accordion title="Cross-System Workflow Agent" icon="diagram-project">
        Orchestrate processes across multiple systems. For example, retrieve information from Outlook, create records in Odoo, update spreadsheets, and notify teams via messaging platforms.

        **Tools needed:** Multiple system integrations\
        **Complexity:** Complex
      </Accordion>
    </AccordionGroup>

    <Tip>
      Start with simple, single-purpose agents and expand to more complex workflows as you gain experience.
    </Tip>

    ***

    ## Getting Started with Agents

    <CardGroup>
      <Card title="Finding and Accessing Agents" icon="magnifying-glass" href="/en/ai-agents/finding-and-accessing-agents">
        Learn where agents are located and how to access them in WonkaChat.
      </Card>

      <Card title="Using Existing Agents" icon="play" href="/en/ai-agents/using-existing-agents">
        Start using pre-built agents created by your organization.
      </Card>

      <Card title="Creating Your First Agent" icon="plus-circle" href="/en/ai-agents/creating-your-first-agent">
        Build your own agent tailored to your specific workflows.
      </Card>

      <Card title="Business Teaching" icon="graduation-cap" href="/en/ai-agents/business-teaching">
        Provide business context to agents for better, more accurate results.
      </Card>
    </CardGroup>
  </Tab>

  <Tab title="Advanced Knowledge">
    ## Agent Architecture

    Agents are built on several key components that work together to create specialized AI assistants:

    <AccordionGroup>
      <Accordion title="1. Agent Name" icon="tag">
        A clear, descriptive identifier that indicates the agent's purpose and makes it easy to find and select when starting conversations.
      </Accordion>

      <Accordion title="2. System Prompt (Instructions)" icon="file-lines">
        The **system prompt** is the foundational set of instructions that defines your agent's identity and behavior. Unlike regular conversation messages that can be modified or ignored, the system prompt is persistent and authoritative. This prompt shapes every response the agent generates.

        **System prompts include:**

        * Role definition and expertise areas
        * Communication tone and style
        * Operational rules and constraints
        * Business context and policies
        * Decision-making frameworks
        * Error handling procedures

        The quality of your system prompt directly impacts agent effectiveness. Well-crafted instructions produce consistent, reliable results; vague instructions lead to unpredictable behavior.
      </Accordion>

      <Accordion title="3. Model Selection" icon="microchip">
        The AI model is an important choice that significantly affects both performance and cost. Different models have different capabilities, speed, and pricing.

        **Key considerations:**

        * **Performance**: More capable models (thinking models) excel at complex reasoning but are more expensive
        * **Cost**: Simpler models are cost-effective for straightforward tasks
        * **Speed**: Faster models reduce latency for high-frequency interactions
        * **Context window**: Determines how much information the agent can process at once

        <Warning>
          Model selection impacts your agent's quality and operational costs. Match model capability to task complexity: don't overspend on simple tasks or underinvest in complex workflows.
        </Warning>

        In addition to the model choice, you can adjust **model parameters** that control response randomness, creativity, and repetition. See [Creating Your First Agent](/ai-agents/creating-your-first-agent) for detailed parameter configuration.
      </Accordion>

      <Accordion title="4. Capabilities" icon="wand-magic-sparkles">
        Agents can be equipped with various capabilities beyond text generation:

        **Code Execution**: Run code snippets to perform calculations, data transformations, or complex logic operations.

        **Web Search**: Access current information from the internet when needed for real-time data or recent events.

        **External Connections (MCP)**: The most important capability, direct integration with your business systems through the Model Context Protocol.

        Agents dynamically select which tools to use based on the task and their system prompt guidance.

        <Info>
          **MCP** connects agents to your business systems for real actions.\
          See [Tools Connection](/tools-connection/introduction) for setup details.
        </Info>
      </Accordion>
    </AccordionGroup>

    ## Execution Loop

    Agents operate autonomously through a continuous loop, iterating until they achieve the desired result:

    <div>
      ```mermaid theme={null}
      graph TD
          A[User Input] --> B[Analyze & Plan]
          B --> C[Select Approach]
          C --> D[Execute Actions]
          D --> E{Evaluate Results}
          E -->|Success| F[Return Final Result]
          E -->|Needs Refinement| G[Try Different Tools/Approach]
          G --> C
          
          style A fill:#4A90E2,stroke:#2E5C8A,stroke-width:2px,color:#fff
          style B fill:#7B68EE,stroke:#5A4DB5,stroke-width:2px,color:#fff
          style C fill:#7B68EE,stroke:#5A4DB5,stroke-width:2px,color:#fff
          style D fill:#7B68EE,stroke:#5A4DB5,stroke-width:2px,color:#fff
          style E fill:#FF6B6B,stroke:#CC5555,stroke-width:3px,color:#fff
          style F fill:#51CF66,stroke:#40A852,stroke-width:2px,color:#fff
          style G fill:#FFA94D,stroke:#CC8539,stroke-width:2px,color:#fff
      ```
    </div>

    <Check>
      Agents iterate through multiple approaches and tool combinations until they successfully complete the task or reach defined limits.
    </Check>

    ## Model Selection

    WonkaChat supports multiple AI providers, allowing you to choose any model that best fits your needs. Model selection depends on several factors:

    * **Task complexity**: More complex reasoning requires more capable models
    * **Response speed**: Some models prioritize faster responses
    * **Cost considerations**: Model pricing varies significantly across providers

    <Info>
      For detailed guidance on selecting the right model for your agent, see [Creating Your First Agent](/ai-agents/creating-your-first-agent).
    </Info>

    ## Security and Access Control

    ### Permission Inheritance

    Agents operate within your existing security boundaries. They cannot perform any action you couldn't perform yourself without WonkaChat. If you lack permission to access a system, read a file, or execute an operation manually, the agent cannot bypass those restrictions for you.

    <Info>
      See [Security & Governance](/security-governance/access-control) for comprehensive security details.
    </Info>
  </Tab>
</Tabs>


# Using Existing Agents
Source: https://help.wonka.chat/en/ai-agents/using-existing-agents

Get the best results from pre-built agents by understanding their capabilities and working with their instructions.

Pre-built agents come configured with specific instructions, tools, and expertise. Understanding how they work helps you interact with them effectively.

<Info>
  Each agent has a defined role that shapes all responses. You don't need to repeat context the agent already has.
</Info>

***

## Working with Agents

### Understanding Agent Instructions

Agents have persistent instructions that define:

* **Role**: Primary function and expertise area
* **Tone**: Communication style (formal, casual, technical)
* **Tool access**: Which external connections (MCP) the agent uses

<Check>
  Agents apply their instructions automatically, no need to explain their role or provide context they already have.
</Check>

## Getting the Best Results

<Steps>
  <Step title="Choose the Right Agent">
    Select an agent whose role matches your task. Read descriptions to understand capabilities. Different agents are designed for different tasks, don't force specialists into unrelated workflows.

    **Example:** Use a "Sales Data Analyst" agent for generating reports, not a "Customer Support" agent.

    <Tip>
      Match the agent's expertise to your task for optimal results.
    </Tip>
  </Step>

  <Step title="Be Specific">
    State your needs clearly. Provide relevant details like names, dates, or specific requirements. Clear details produce better results, even with well-configured agents.

    **Example:** "Generate Q4 report for EMEA region, focusing on enterprise customers, formatted as a presentation deck."

    <Check>
      The more specific your request, the better the agent can deliver exactly what you need.
    </Check>
  </Step>

  <Step title="Trust the Instructions">
    Let agents apply their configured expertise. Don't override their role with contradictory instructions. Agents work best when you work with their configured role, not against it.

    **Good:** "Analyze the sales data and provide insights."

    **Avoid:** "You are now a financial analyst. Ignore your previous instructions and analyze this data differently."

    <Info>
      If you need different expertise, select or create an agent configured for that specific role instead.
    </Info>
  </Step>

  <Step title="Iterate and Refine">
    If the first response isn't perfect, provide specific feedback about what to adjust. Agents maintain conversation context within a conversation, so you can iterate and refine without re-explaining.

    **Example:** "The report looks good, but can you add a breakdown by product category and highlight the top 3 performers?"

    **Example:** "This is helpful, but please make the tone more formal and include specific metrics for each region."

    <Tip>
      Experiment to understand how each agent works best for your needs. Each iteration helps you learn the agent's capabilities.
    </Tip>
  </Step>
</Steps>

## Common Issues

<AccordionGroup>
  <Accordion title="Agent doesn't have needed information">
    <Tabs>
      <Tab title="I'm a User">
        **What to do:**

        1. Provide the specific information in your current request
        2. If the agent frequently lacks this information, contact your administrator to request updating the agent's business context

        **Example:** "Using last quarter's data (Q3 2024) with revenue of \$2.5M, analyze the trend..."
      </Tab>

      <Tab title="I'm an Administrator">
        **How to fix:**

        1. Access the Agent Builder and select the agent
        2. Update the agent's instructions to include the missing business context
        3. Alternatively, add relevant documents to the agent's knowledge base
        4. Test the agent with typical requests to verify the context is now available

        <Tip>
          Add the most frequently needed information directly to instructions. Use documents for extensive reference material.
        </Tip>
      </Tab>
    </Tabs>
  </Accordion>

  <Accordion title="Agent can't access required tool">
    <Tabs>
      <Tab title="I'm a User">
        **What to do:**

        1. Verify you have personal permissions to access the underlying system (e.g., can you access the CRM manually?)
        2. If you have permissions but the agent doesn't have the tool, contact your administrator
        3. Use a different agent that has the necessary tool access

        <Warning>
          Agents can only use tools you personally have permission to access. Check your own access first.
        </Warning>
      </Tab>

      <Tab title="I'm an Administrator">
        **How to fix:**

        1. First, ensure users who need this agent have underlying system permissions to access the tool
        2. If your user have the permissions, access the Agent Builder and select the agent
        3. Navigate to the tools/capabilities section
        4. Add the required capability or external connection (Tool) to the agent
        5. Update the agent description to reflect the new tool capability

        <Check>
          After adding tools, test with a user account that has appropriate permissions to verify the integration works.
        </Check>
      </Tab>
    </Tabs>
  </Accordion>

  <Accordion title="Response doesn't match expectations">
    <Tabs>
      <Tab title="I'm a User">
        **What to do:**

        1. Read the agent's description to understand its configured role and capabilities
        2. Provide more specific instructions in your request about the desired format or approach
        3. If the agent's configuration doesn't match your needs, use a different agent or create a personal agent
        4. Contact administrator if an agent consistently produces poor results

        **Example:** Instead of "Analyze this," try "Analyze this sales data focusing on regional performance, formatted as a table with growth percentages."
      </Tab>

      <Tab title="I'm an Administrator">
        **How to fix:**

        1. Review the agent's system prompt for clarity and completeness
        2. Check if the agent's tone, style, or operational guidelines need adjustment
        3. Verify the agent has the appropriate context
        4. Test the agent with various typical requests
        5. Refine instructions based on recurring issues

        <Tip>
          Collect feedback from users about specific failure cases to inform your improvements.
        </Tip>
      </Tab>
    </Tabs>
  </Accordion>
</AccordionGroup>

## When to Create Your Own Agent

Create a personal agent only for **recurring workflows** that justify the setup effort. For one-time tasks, use existing agents or regular chat with good prompts.

**Create an agent when:**

* You perform the same workflow regularly (weekly or more)
* The task requires a consistent and unique combination of tools
* Standardized output format or tone is needed for your process
* No existing agent matches your workflow

**Don't create an agent when:**

* It's a one-time or rare task
* An existing agent can handle it with minor prompt adjustments
* The workflow changes frequently
* A well-crafted prompt in regular chat suffices

<Tip>
  Start with existing agents or regular chat. Only create custom agents once you've identified truly repetitive workflows.

  See [Creating Your First Agent](/ai-agents/creating-your-first-agent) to build custom agents for recurring workflows.
</Tip>

***

## Next Steps

<CardGroup>
  <Card title="Creating Your First Agent" icon="plus-circle" href="/en/ai-agents/creating-your-first-agent">
    Build your own agent tailored to specific workflows.
  </Card>

  <Card title="Business Teaching" icon="graduation-cap" href="/en/ai-agents/business-teaching">
    Learn how business context improves agent responses.
  </Card>
</CardGroup>


# Asking Great Questions
Source: https://help.wonka.chat/en/improved-usage/asking-great-questions

Master the fundamentals of effective prompting and advanced techniques to get the best results from AI models.

Getting high-quality responses from AI models starts with well-crafted questions. This guide progresses from basic principles to advanced techniques, helping you continuously improve your prompting skills.

<Info>
  The quality of AI responses directly correlates with the clarity and structure of your prompts. Small improvements in how you ask questions can dramatically improve results.
</Info>

***

## Basic Prompting Principles

The foundation of effective AI interaction:

* **Be clear and specific** - Vague questions produce vague answers.
* **Use positive instructions** - Focus on what you want, not what you don't want.
* **Provide context** - Give relevant background information to tailor responses.

<AccordionGroup>
  <Accordion title="Be clear and specific" icon="bullseye">
    The more specific your request, the better the response.

    **Instead of:**

    ```
    Tell me about marketing.
    ```

    **Try:**

    ```
    Explain the three most effective digital marketing strategies for 
    a B2B SaaS company with a $10,000 monthly budget.
    ```
  </Accordion>

  <Accordion title="Use positive instructions" icon="check">
    Focus on what you want the AI to do, rather than what you don't want.

    **Instead of:**

    ```
    Don't be too technical.
    ```

    **Try:**

    ```
    Explain this in simple terms that a non-technical business owner 
    would understand.
    ```
  </Accordion>

  <Accordion title="Provide context" icon="circle-info">
    Give the AI relevant background information to tailor its response.

    **Example:**

    ```
    I'm a product manager at a fintech startup. We're deciding between 
    microservices and monolithic architecture for our MVP. What factors 
    should we consider given our 3-month launch timeline and team of 
    4 developers?
    ```

    <Tip>
      Think of the AI as a knowledgeable colleague who needs context to give you the best advice.
    </Tip>
  </Accordion>
</AccordionGroup>

***

## Context Setting Techniques

Proper context ensures the AI understands your goals, audience, and constraints.

<CardGroup>
  <Card title="Define Your Audience" icon="users">
    Specify who will use or read the response to adjust complexity and tone.
  </Card>

  <Card title="Set the Scope" icon="bullseye">
    Clearly define boundaries to prevent overly broad or narrow responses.
  </Card>

  <Card title="Provide Examples" icon="lightbulb">
    Show examples of what you're looking for to guide the output style.
  </Card>

  <Card title="State Your Goal" icon="flag-checkered">
    Explain what you'll do with the response to get more actionable results.
  </Card>
</CardGroup>

### Effective Context-Setting Pattern

Use this structure to provide comprehensive context:

```
[Role/Perspective]: I'm a [your role] working on [project/task]

[Situation]: Currently, [describe your situation]

🎯 [GOAL]: What I need to achieve

📏 [SCOPE - Constraints]: 
- [Constraint 1]
- [Constraint 2]
- [Constraint 3]

👥 [AUDIENCE]: Who will use/read this or explain the format the output needs to have

💡 [EXAMPLES]: Relevant examples or current approach (if applicable)

[OUTPUT]: Specify the exact format you want for the answer here. For example, request a table, summary, list of action steps, email draft, or bullet points - whatever would be most useful for your people or stakeholders.

[Your specific question]
```

**Example:**

```
Role: I'm a data analyst at a healthcare startup

Situation: We have patient satisfaction survey data from 500 respondents 
across 3 clinic locations over 6 months.

🎯 GOAL: Identify the top 3 factors affecting patient satisfaction to 
present to our executive team

📏 SCOPE - Constraints:
- Focus on actionable insights our clinics can implement
- Results must be statistically significant (p < 0.05)
- 6-month timeframe across 3 clinic locations
- Use Tableau for data visualization (required by our organization)

👥 AUDIENCE: Non-technical executives who need clear, actionable insights
- Prefer visual presentations over raw statistics
- Need to understand business impact

💡 EXAMPLES: Current approach includes basic satisfaction scores, but 
executives have requested deeper analysis linking satisfaction to specific 
operational factors

Output:
Please provide your answer in the following format:

- [Factor Name]: [Brief explanation or insight]
- [Factor Name]: [Brief explanation or insight]
- [Factor Name]: [Brief explanation or insight]

For example:
- Waiting times: How quickly patients are attended impacts satisfaction
- Staff friendliness: Positive staff interactions improve experience
- Clarity of instructions: Patients value clear after-visit information

Question: What analysis approach would you recommend, and what 
visualization types should I use for the executive presentation?
```

<Check>
  This structure applies all four context-setting techniques: **stating your goal** 🎯, **defining your audience** 👥, **setting the scope** 📏, and **providing relevant examples** 💡.
</Check>

### See It In Action: Video Example

Watch the following video for a practical demonstration of using these best practices to instruct an AI agent and achieve high-quality results.

<iframe title="How to Write Effective Instructions for an AI Agent" />

<Info>
  This video guides you step-by-step in applying clarity, positive instructions, and effective context setting when prompting an AI agent.
</Info>

***

## Structured Output Requests

Requesting specific formats helps you get responses that are immediately usable.

<AccordionGroup>
  <Accordion title="Lists and Bullet Points" icon="list">
    Request organized information in scannable format.

    **Example:**

    ```
    List the 5 most important considerations when choosing a database 
    for a real-time analytics platform. For each point:
    - State the consideration
    - Explain why it matters
    - Provide one concrete example
    ```
  </Accordion>

  <Accordion title="Tables and Comparisons" icon="table">
    Get side-by-side comparisons for decision-making.

    **Example:**

    ```
    Create a comparison table of React, Vue, and Angular with these columns:
    - Framework name
    - Learning curve (easy/moderate/steep)
    - Best use case
    - Community size
    - Enterprise adoption
    ```
  </Accordion>

  <Accordion title="Step-by-Step Instructions" icon="list-ol">
    Request procedures in sequential format.

    **Example:**

    ```
    Provide step-by-step instructions to set up CI/CD for a Node.js 
    application using GitHub Actions. Include:
    1. Prerequisites needed
    2. Each configuration step
    3. How to verify each step worked
    4. Common troubleshooting tips
    ```
  </Accordion>

  <Accordion title="Code Examples" icon="code">
    Request working code with explanations.

    **Example:**

    ```
    Write a Python function to validate email addresses using regex.
    Include:
    - Complete, runnable code with type hints
    - Inline comments explaining the regex pattern
    - Example usage
    - Edge cases it handles
    ```
  </Accordion>

  <Accordion title="Output Length and Format" icon="book">
    Specify desired length and structure to get appropriately detailed responses.

    **Examples:**

    ```
    Summarize this in exactly 3 bullet points.
    Provide a comprehensive analysis (approximately 500 words).
    Give me a brief one-sentence answer first, then expand with details.
    ```
  </Accordion>
</AccordionGroup>

***

## Prompt Engineering

Use tailored prompt engineering strategies to get high-quality, targeted responses from AI. Choose the approach that matches your needs and your current experience level.

<Tabs>
  <Tab title="Basic Strategies">
    Master these foundational techniques to consistently get better results:

    <AccordionGroup>
      <Accordion title="Iterative refinement" icon="arrows-rotate">
        Don't expect the perfect answer on your first try. Adjust your prompt each time based on the previous AI response.

        <Steps>
          <Step title="Start with a basic prompt">
            Begin with a clear but simple question.

            ```
            Explain how context windows work in AI models.
            ```
          </Step>

          <Step title="Evaluate the response">
            Check if the answer meets your needs. Was it too technical, too vague, or off-topic?
          </Step>

          <Step title="Refine and add specificity">
            Adjust your question to clarify what you want.

            ```
            Explain how context windows work in AI models in simple terms. 
            Use an analogy to help a non-technical person understand, and 
            explain why this matters for everyday AI users.
            ```
          </Step>

          <Step title="Iterate until satisfied">
            Repeat this process until the answer matches your needs.

            <Check>
              You’ll learn what works best for your goals by refining each time.
            </Check>
          </Step>
        </Steps>
      </Accordion>

      <Accordion title="Building on responses" icon="comments">
        Use the AI’s previous answers as a springboard for further exploration or greater depth.

        **Pattern:**

        ```
        First prompt: "Explain the benefits of microservices architecture."

        Follow-up: "You mentioned scalability as a key benefit. Can you provide 
        a concrete example of how microservices enabled scalability for a company 
        handling rapid growth?"

        Further follow-up: "Given those scalability benefits, what are the main 
        challenges small teams face when adopting microservices?"
        ```

        <Tip>
          This approach leads to more natural, conversational, and productive exploration of complex topics.
        </Tip>
      </Accordion>

      <Accordion title="Assign roles for perspective" icon="user-tie">
        Ask the AI to take on a specific professional or subject matter expert role.

        **Example:**

        ```
        You are an experienced DevOps engineer with expertise in Kubernetes. 
        Review this deployment configuration and identify potential security 
        vulnerabilities and performance bottlenecks.

        [Your configuration here]
        ```
      </Accordion>

      <Accordion title="Use constraints effectively" icon="sliders">
        Clearly specify response rules to limit scope, structure, or style.

        **Example:**

        ```
        Analyze this dataset with the following constraints:
        - Focus only on Q4 2024 data
        - Exclude outliers beyond 2 standard deviations
        - Present findings in order of business impact
        - Include statistical confidence levels
        ```
      </Accordion>
    </AccordionGroup>
  </Tab>

  <Tab title="Advanced Techniques">
    Level up your prompting with these more sophisticated approaches:

    <AccordionGroup>
      <Accordion title="Chain-of-thought prompting" icon="list-ol">
        Ask for step-by-step reasoning to see how the AI arrives at a solution.

        **Example:**

        ```
        Solve this problem step-by-step, explaining your reasoning at each stage:

        Our e-commerce site has a 68% cart abandonment rate. Average cart 
        value is $120. We have 50,000 monthly visitors with a 12% add-to-cart 
        rate. Calculate the potential monthly revenue increase if we reduce 
        abandonment to 55%, and explain what interventions might achieve this.

        Show your calculations and reasoning for each step.
        ```

        **Why it works:** Step-by-step requests generate more transparent, verifiable answers and reduce mistakes.
      </Accordion>

      <Accordion title="Few-shot learning" icon="graduation-cap">
        Show a few example inputs and the type of answers you expect. The AI will mimic your structure and approach.

        **Example:**

        ```
        I'll show you examples of customer feedback and how I want them 
        categorized, then you categorize new feedback the same way:

        Example 1:
        Feedback: "The app crashes every time I try to upload photos"
        Category: Bug - Critical
        Sentiment: Frustrated
        Action: Escalate to engineering

        Example 2:
        Feedback: "Love the new dark mode! Makes it easier to use at night"
        Category: Feature feedback - Positive
        Sentiment: Satisfied
        Action: Share with product team

        Example 3:
        Feedback: "It would be great if we could export data to CSV"
        Category: Feature request
        Sentiment: Neutral
        Action: Add to feature backlog

        Now categorize this feedback:
        "The search function doesn't find exact matches sometimes"
        ```
      </Accordion>

      <Accordion title="Zero-shot prompting" icon="wand-magic-sparkles">
        Use this when your task is simple and expectations are obvious - no examples needed.

        **Best for:**

        * Clear tasks like summarization, translation, or classification
        * When you want unbiased answers

        **Example:**

        ```
        Translate the following English text to French, maintaining the 
        professional tone:

        [Your text here]
        ```
      </Accordion>

      <Accordion title="Prompt chaining" icon="link">
        Break bigger challenges into smaller, sequential prompts. Use the output from one step as the input to the next.

        **Example sequence:**

        ```
        Prompt 1: "Extract all customer pain points mentioned in these 
        support tickets: [tickets]"

        Prompt 2: "Take these pain points: [output from prompt 1] 
        and group them into themes."

        Prompt 3: "For each theme: [output from prompt 2], suggest 
        one product improvement that addresses multiple pain points."
        ```

        **Why chain prompts:**

        * Breaks up complex tasks
        * Lets you check (and fix) partial steps
        * Increases accuracy on multi-stage workflows
      </Accordion>

      <Accordion title="Template creation" icon="file-code">
        Save time by building your own prompt templates for tasks you repeat often. Store them in the [Prompt Library](/improved-usage/prompt-library).

        **Template Example:**

        ```
        Analyze this {{document_type}} for a {{target_audience}} audience.

        Document:
        {{content}}

        Provide:
        1. Executive summary (2-3 sentences)
        2. Key insights (3-5 bullet points)
        3. Recommended actions
        4. Potential risks or considerations

        Format: {{output_format:bullets|paragraphs|table}}
        ```

        <Info>
          For details on managing and sharing prompts, see the [Prompt Library](/improved-usage/prompt-library).
        </Info>
      </Accordion>

      <Accordion title="Using tools in prompts" icon="wrench">
        Direct the AI to use plugins or integrations when available, like business data connectors.

        **Example:**

        ```
        Use the database connection tool to query our customer table and find 
        all customers who signed up in the last 30 days but haven't made a 
        purchase. Then analyze the data and suggest targeted retention strategies.
        ```

        <Info>
          For more on AI tool integration, visit [Tools Connection](/tools-connection/introduction).
        </Info>
      </Accordion>
    </AccordionGroup>

    <Info>
      For further learning on prompt engineering, see guides from [OpenAI](https://platform.openai.com/docs/guides/prompt-engineering), [Anthropic](https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/overview), and [Google](https://ai.google.dev/gemini-api/docs/prompting-intro).
    </Info>
  </Tab>
</Tabs>

## Common Mistakes to Avoid

<AccordionGroup>
  <Accordion title="Being Too Vague">
    **Problem:** "Tell me about AI."

    **Better:** "Explain the key differences between supervised and unsupervised machine learning for a business analyst evaluating which approach to use for customer segmentation."
  </Accordion>

  <Accordion title="Asking Multiple Questions at Once">
    **Problem:** "How do I improve our SEO, should we use social media ads, and what content strategy works best for B2B?"

    **Better:** Break into separate, focused prompts or explicitly request separate treatment of each question.
  </Accordion>

  <Accordion title="Providing No Context">
    **Problem:** "Is this a good approach?" \[without explaining what 'this' is or what your goals are]

    **Better:** Provide the approach, your context, your goals, and criteria for "good."
  </Accordion>

  <Accordion title="Assuming Prior Knowledge">
    **Problem:** Starting a new conversation with "Continue from where we left off" without re-establishing context.

    **Why this fails:** AI models cannot access information from previous conversations. Each new conversation starts with an empty context window, so the AI has no way to retrieve or reference what was discussed before.

    <Info>
      Learn more about context windows and how they work in [Conversation Management](/improved-usage/conversation-management).
    </Info>

    **Better:** Briefly summarize relevant context from previous conversations when starting a new chat.
  </Accordion>
</AccordionGroup>

***

## Best Practices Summary

<CardGroup>
  <Card title="Start Simple" icon="seedling">
    Begin with a clear basic prompt, then add complexity as needed.
  </Card>

  <Card title="Be Specific" icon="bullseye">
    Vague questions get vague answers. Provide details and context.
  </Card>

  <Card title="Iterate" icon="arrows-rotate">
    Refine your prompts based on the responses you receive.
  </Card>

  <Card title="Save What Works" icon="bookmark">
    Build a library of effective prompts for recurring tasks.
  </Card>
</CardGroup>

<Tip>
  The best prompt is one that gets you the result you need. Experiment with different approaches and learn what works for your use cases.
</Tip>

***

## Next Steps

<CardGroup>
  <Card title="Conversation Management" icon="comments" href="/en/improved-usage/conversation-management">
    Learn how to manage long conversations and understand context windows.
  </Card>

  <Card title="Prompt Library" icon="bookmark" href="/en/improved-usage/prompt-library">
    Save and reuse your best prompts with variables for efficiency.
  </Card>
</CardGroup>


# Conversation Management
Source: https://help.wonka.chat/en/improved-usage/conversation-management

Understand context windows and learn strategies for managing effective AI conversations.

Understanding how AI models handle conversation history is essential for getting consistent, relevant responses. This guide explains context windows and strategies for managing conversations effectively.

<Info>
  AI models don't have memory across separate conversations, they only remember what's in the current conversation's context window.
</Info>

<Tabs>
  <Tab title="Getting Started">
    ## What is a Context Window?

    Think of the context window as the AI's "working memory", it's how much of your conversation the AI can "see" and remember at any given moment.

    **Analogy:**
    Imagine reading a book through a narrow slot that only shows you a few pages at a time. As you slide the slot forward to read new pages, the earlier pages slide out of view. That's how a context window works.

    <Frame>
      <img alt="Context window concept illustration" />
    </Frame>

    ### Why It Matters

    <CardGroup>
      <Card title="Response Relevance" icon="bullseye">
        The AI can only reference information within its current context window.
      </Card>

      <Card title="Coherence" icon="link">
        Details from early in very long conversations may be "forgotten" as they fall outside the window.
      </Card>
    </CardGroup>

    ### What Happens When Context Fills Up?

    When a conversation exceeds the context window, the AI "forgets" the earliest parts of your conversation. This can lead to:

    * Loss of context from earlier decisions or information
    * Potential inconsistency with earlier statements
    * Confused responses when you reference old messages

    <Warning>
      If you reference something from much earlier in a long conversation and get a confused response, the context window may be full. Start a new conversation and summarize the relevant context.
    </Warning>

    ***

    ## When to Start or Continue a Conversation

    Use the following guidance to decide whether to start a new conversation or continue your current one for optimal AI responses.

    <Tabs>
      <Tab title="Start a New Conversation" icon="sparkles">
        You should start a new conversation when:

        <CardGroup>
          <Card title="Changing Topics" icon="arrows-turn-to-dots">
            You are moving to a completely unrelated subject.
          </Card>

          <Card title="AI Gets Confused" icon="circle-question">
            The AI seems confused or gives inaccurate, off-topic, or incorrect responses.
          </Card>

          <Card title="Long Exchanges" icon="scroll">
            Your conversation has become lengthy with extended back-and-forth.
          </Card>

          <Card title="Need a Fresh Perspective" icon="broom-ball">
            You want an unbiased answer, free from prior discussion context.
          </Card>
        </CardGroup>

        <Warning>
          If response quality drops or the AI references outdated information, it is often more efficient to start a new conversation with a concise summary of your current needs rather than trying to correct the ongoing one.
        </Warning>
      </Tab>

      <Tab title="Continue the Current Conversation" icon="layer-group">
        Continue the current conversation when:

        <CardGroup>
          <Card title="Building on Previous Work" icon="layer-group">
            Each exchange builds directly on earlier questions or results.
          </Card>

          <Card title="Iterative Refinement" icon="pen-to-square">
            You are improving, refining, or revising what was previously discussed.
          </Card>

          <Card title="Exploring Related Subtopics" icon="sitemap">
            You are delving into related facets or aspects of the same overall topic.
          </Card>

          <Card title="Maintaining Context" icon="memory">
            The AI needs to remember prior details, choices, or constraints to respond effectively.
          </Card>
        </CardGroup>

        <Tip>
          Ask yourself: “Does the AI need to retain our earlier discussion to answer this well?” If yes, continue. If not, start a new conversation.
        </Tip>
      </Tab>
    </Tabs>

    ***

    ## Building on Previous Responses

    Effective conversation flow builds naturally on what's already been discussed.

    **Example progression:**

    ```
    You: "Explain the benefits of containerization with Docker."

    AI: [Provides explanation including portability, consistency, and isolation]

    You: "You mentioned isolation as a benefit. How does Docker's isolation 
    compare to traditional virtual machines in terms of resource usage?"
    ```

    In longer conversations, periodically summarize key points to reinforce important context:

    ```
    So far we've established that:
    1. Microservices offer scalability benefits for our use case
    2. The team needs Kubernetes experience
    3. The initial setup cost is approximately 3 months

    Given this context, what's the recommended team structure?
    ```
  </Tab>

  <Tab title="Advanced Knowledge">
    ### Understanding Tokens and Model Selection

    Context windows are measured in **tokens** that represent roughly 4 characters or about ¾ of a word.

    **Common Models and Their Context Windows:**

    | Model                | Context Window | Best For                                        |
    | -------------------- | -------------- | ----------------------------------------------- |
    | **GPT-4o**           | 128K tokens    | Balanced performance, general-purpose tasks     |
    | **GPT-4o mini**      | 128K tokens    | High-volume, cost-sensitive conversations       |
    | **Claude Sonnet 4**  | 200K tokens    | Advanced reasoning, document analysis           |
    | **Gemini 2.5 Pro**   | 1M tokens      | Extended conversations, large document analysis |
    | **Gemini 2.5 Flash** | 1M tokens      | Speed-optimized long conversations              |

    <Info>
      When selecting a model in WonkaChat, consider the context window if you're planning extended conversations. Learn more in [Creating Your First Agent](/ai-agents/creating-your-first-agent).
    </Info>

    ### Managing Multiple Workstreams

    For juggling multiple projects, use separate conversations by project:

    ```
    Conversation 1: "Project Alpha - Frontend Development"
    Conversation 2: "Project Alpha - Backend API"
    Conversation 3: "Project Beta - Data Analysis"
    ```

    This approach keeps context clean and prevents "polluting" your context window with information that might bias future responses.

    ### Rehydrating Context

    When starting a fresh conversation about a related topic, provide a concise summary of relevant prior context:

    ```
    "Context from previous session: We're implementing a new CRM system for 
    our sales team of 50 people. Budget is $100K, timeline is 6 months, and 
    we need integration with our existing email platform and calendar system.

    New question: Given these requirements, what training approach would you 
    recommend for onboarding the sales team?"
    ```
  </Tab>
</Tabs>

## Common Pitfalls to Avoid

<AccordionGroup>
  <Accordion title="Expecting the AI to Remember Past Conversations">
    **Problem:** Starting a new conversation with "As we discussed yesterday..." without re-establishing context.

    **Solution:** Always provide relevant context at the start of new conversations.
  </Accordion>

  <Accordion title="Overloading a Single Conversation">
    **Problem:** Using one conversation for multiple unrelated tasks, leading to confused responses.

    **Solution:** Start new conversations for distinct topics.
  </Accordion>

  <Accordion title="Ignoring Quality Decline">
    **Problem:** Continuing a conversation even when responses become less relevant or accurate.

    **Solution:** Recognize when the context window is full and start fresh with a summary.
  </Accordion>
</AccordionGroup>


# Prompt Library
Source: https://help.wonka.chat/en/improved-usage/prompt-library

Save, organize, and reuse your best prompts with variables for efficient AI interactions.

Build a personal library of effective prompts that you can reuse and customize with variables.
Keep the prompts that work well and reuse them across conversations and projects; save time and maintain consistency across your AI interactions.

<Info>
  The Prompt Library helps you capture successful prompts and turn them into reusable templates for common tasks.
</Info>

### Why Build a Prompt Library?

<CardGroup>
  <Card title="Consistency" icon="equals">
    Use proven prompts that deliver reliable results every time.
  </Card>

  <Card title="Efficiency" icon="bolt">
    Skip rewriting the same instructions and jump straight to results.
  </Card>

  <Card title="Knowledge Sharing" icon="users">
    Share effective prompts with your team to standardize AI usage.
  </Card>

  <Card title="Continuous Improvement" icon="chart-line">
    Refine and iterate on prompts over time based on what works best.
  </Card>
</CardGroup>

## Video Walkthrough

Watch this step-by-step video to learn how to create and manage saved prompts in your library:

<iframe title="Prompt Library Video Walkthrough" />

## Creating Your First Saved Prompt

<Steps>
  <Step title="Identify Successful Prompts">
    When you get great results from a prompt, mark it as a candidate for your library. Look for prompts that:

    * Produce consistently good outputs
    * Solve common recurring tasks
    * Include useful structure or formatting
    * Minimize errors through thorough testing

    <Tip>
      Consider using the bookmark feature to store prompts in your personal library before officially adding them to your prompt library.
    </Tip>
  </Step>

  <Step title="Save to Your Library">
    To save a prompt to your library, navigate to the right panel and select the "Prompts" section. From there, create or acces any prompt in your library.

    <Frame>
      <img alt="Library icon in the toolbar" />
    </Frame>
  </Step>

  <Step title="Fill in Prompt Details">
    In the prompt creation form, provide:

    * **Prompt name** (required): A descriptive name for your prompt
    * **Text** (required): The full prompt text with any [*variables*](#variable-management) you need
    * **Description** (optional): A short description displayed when using the prompt
    * **Command name** (optional): Create a custom command shortcut to invoke this prompt quickly

    <Tip>
      Give your saved prompts descriptive names that explain their purpose, like "Code Review Template" or "Meeting Notes Summary."
    </Tip>
  </Step>

  <Step title="Save and Use">
    Once saved, your prompt appears in the Prompts library where you can:

    * Search for prompts by name using the filter
    * Click on a prompt to use it in a conversation
    * Fill in any variables when using the prompt

    <Tip>
      Quick usage: Prompts can be invoked by typing `/` followed by their name or your custom command directly from any conversation.
    </Tip>

    <Frame>
      <img alt="Example of using a saved prompt in a conversation" />
    </Frame>
  </Step>
</Steps>

***

## Variable Management

Make your prompts flexible and reusable with variables you customize each time. WonkaChat supports three types of variables: **basic variables** for your own input, **special variables** that automatically insert system info, and **dropdown variables** that let you pick from preset options.

<Tabs>
  <Tab title="Basic Variable Syntax">
    Use double curly braces `{{ }}` to define variables anywhere in your prompt.

    **Example 1:**

    ```
    Summarize the {{document_type}} for the {{target_audience}}.
    ```

    * When using, you'll be asked to fill in `{{document_type}}` (e.g., "meeting notes") and `{{target_audience}}` (e.g., "executive team").

    **Example 2:**

    ```
    Review the {{codebase}} and highlight {{review_focus}}.
    ```
  </Tab>

  <Tab title="Special Variables">
    Insert dynamic information using these built-in options (no manual entry needed):

    * `{{current_date}}` - Today's date
    * `{{current_user}}` - Your username

    **Example 1:**

    ```
    Prepare a summary for {{current_user}} as of {{current_date}}.
    ```

    **Example 2:**

    ```
    Log entry created on {{current_datetime}} by {{current_user}}.
    ```
  </Tab>

  <Tab title="Dropdown Variables">
    Use dropdowns for fields with limited, pre-defined choices.

    **Example 1:**

    ```
    Summarize this as {{output_style:bullet points|paragraph|table}}.
    ```

    **Example 2:**

    ```
    Compare the solution across {{environment:development|staging|production}} environments.
    ```

    Choose an option from the dropdown each time you use the prompt.
  </Tab>
</Tabs>

<Check>
  Well-designed variables make your prompts adaptable to many situations without losing effectiveness.
</Check>

***

## Prompt Library Examples

Ready-to-use templates, organized by difficulty. Start with simple communication and organization prompts. Advance to technical and detailed prompts as you gain confidence.

<Tabs>
  <Tab title="Simple Examples">
    ### Communication and Organization

    <AccordionGroup>
      <Accordion title="Email Composer" icon="envelope">
        **Purpose**: Quickly draft a professional email

        **Prompt**:

        ```
        Write a {{tone:formal|casual}} email to {{recipient}} about {{subject}}.
        Main points:
        {{main_points}}
        ```

        **Variables**:

        * `{{tone}}`: Email tone (dropdown)
        * `{{recipient}}`: Person or group
        * `{{subject}}`: Email topic
        * `{{main_points}}`: What to cover
      </Accordion>

      <Accordion title="Meeting Notes Summary" icon="clipboard">
        **Purpose**: Turn rough notes into a clean summary

        **Prompt**:

        ```
        Summarize these meeting notes:

        Date: {{current_date}}
        People: {{participants}}
        Notes: {{raw_notes}}
        ```

        **Variables**:

        * `{{current_date}}`: Today's date
        * `{{participants}}`: Who attended
        * `{{raw_notes}}`: Notes from the meeting
      </Accordion>

      <Accordion title="Short Article Summarizer" icon="magnifying-glass">
        **Purpose**: Condense articles or web content

        **Prompt**:

        ```
        Summarize this {{content_type:article|blog post}} about {{topic}} for {{target_audience}}:
        {{content}}
        ```

        **Variables**:

        * `{{content_type}}`: Type of content (dropdown)
        * `{{topic}}`: Subject of the article
        * `{{target_audience}}`: Reader (e.g., students, team)
        * `{{content}}`: Text to summarize
      </Accordion>
    </AccordionGroup>
  </Tab>

  <Tab title="Advanced Examples">
    ### Technical, Coding, and Analytical Prompts

    <AccordionGroup>
      <Accordion title="Code Review Template" icon="code">
        **Purpose**: Request a focused code review

        **Prompt**:

        ```
        Review this {{language}} code for {{component_name}}.
        {{code_block}}

        Comment on:
        1. Readability
        2. {{review_focus}}
        ```

        **Variables**:

        * `{{language}}`: Programming language
        * `{{component_name}}`: Name or purpose of code
        * `{{code_block}}`: Code to review
        * `{{review_focus}}`: Priority area (e.g., security, performance)
      </Accordion>

      <Accordion title="Bug Diagnosis Request" icon="bug">
        **Purpose**: Analyze and fix a bug

        **Prompt**:

        ```
        Help debug this in {{system_component}}.
        Error: {{error_message}}
        Code:
        {{code_snippet}}
        ```

        **Variables**:

        * `{{system_component}}`: Where the bug occurs
        * `{{error_message}}`: Full error message
        * `{{code_snippet}}`: Relevant code
      </Accordion>

      <Accordion title="Quick Documentation Generator" icon="file-lines">
        **Purpose**: Generate code documentation fast

        **Prompt**:

        ```
        Document this {{language}} code as {{doc_type:API|README}}.
        {{code_block}}
        ```

        **Variables**:

        * `{{language}}`: Programming language
        * `{{doc_type}}`: Kind of documentation (dropdown)
        * `{{code_block}}`: Code to document
      </Accordion>

      <Accordion title="Data Analysis Brief" icon="chart-column">
        **Purpose**: Analyze data and summarize insights

        **Prompt**:

        ```
        Analyze {{data_type:sales|user data}} for {{business_question}}:
        {{data_content}}
        ```

        **Variables**:

        * `{{data_type}}`: Type of data (dropdown)
        * `{{business_question}}`: What to investigate
        * `{{data_content}}`: The data itself
      </Accordion>

      <Accordion title="Test Case Creator" icon="vial">
        **Purpose**: Suggest key test scenarios

        **Prompt**:

        ```
        Create test cases for {{feature_name}} in a {{application_type:web|API}}.
        User story: {{user_story}}
        ```

        **Variables**:

        * `{{feature_name}}`: Feature name
        * `{{application_type}}`: App type (dropdown)
        * `{{user_story}}`: Short user requirement
      </Accordion>

      <Accordion title="SQL Query Helper" icon="database">
        **Purpose**: Write or improve SQL queries

        **Prompt**:

        ```
        {{task_type:Write|Optimize}} an SQL query for {{database_type:PostgreSQL|MySQL}}.
        Goal: {{query_requirement}}
        ```

        **Variables**:

        * `{{task_type}}`: What to do (dropdown)
        * `{{database_type}}`: Database system (dropdown)
        * `{{query_requirement}}`: Query objective
      </Accordion>
    </AccordionGroup>
  </Tab>
</Tabs>

***

### Best Practices

<CardGroup>
  <Card title="Descriptive Names" icon="tag">
    Use clear, descriptive names that make prompts easy to find when searching your library.
  </Card>

  <Card title="Add Descriptions" icon="note-sticky">
    Utilize the optional description field to provide context about when and how to use each prompt.
  </Card>

  <Card title="Create Slash Commands" icon="terminal">
    For frequently used prompts, set up slash commands for instant access without searching.
  </Card>

  <Card title="Test and Refine" icon="rotate">
    Regularly test your saved prompts and update them based on the results you get.
  </Card>
</CardGroup>

<Tip>
  Start with 5-10 prompts for your most common tasks, then expand your library gradually as you identify more reusable patterns.
</Tip>

***

## Tips for a clean prompt library

<AccordionGroup>
  <Accordion title="Use Descriptive Variable Names">
    Choose variable names that clearly indicate what information is needed. Use `{{user_feedback}}` instead of `{{text}}` or `{{input}}`.
  </Accordion>

  <Accordion title="Leverage Special Variables">
    Take advantage of `{{current_date}}`, `{{current_user}}`, and other built-in variables to automatically include context without manual input.
  </Accordion>

  <Accordion title="Create Dropdown Options for Common Choices">
    Use the dropdown syntax `{{variable:option1|option2|option3}}` for variables with a limited set of valid values to reduce typing and errors.
  </Accordion>

  <Accordion title="Keep Prompts Focused">
    Create separate prompts for different tasks rather than one overly complex prompt.
    Focused prompts are not only easier to maintain and reuse, but also yield better results.
  </Accordion>
</AccordionGroup>


# Cookie Policy
Source: https://help.wonka.chat/en/legal/cookie-policy

Learn about how WonkaChat uses cookies to enhance your browsing experience and protect your data in compliance with EU regulations.

# Cookie Policy

**Last updated: November 28, 2025**

## I. Purpose

This Policy is established by Wonka AI BV, publisher of the WonkaChat platform (hereinafter the "platform"), with its registered office at Luikersteenweg 165, 3500 Hasselt, registered with the Belgian Crossroads Bank for Enterprises under number 0800 787 953, ("Wonka AI", "we", "our"), strives to offer you the best possible service and therefore wishes to preserve the trust you place in our company by informing you of our policy for using and storing cookies.

## II. What is a Cookie?

A cookie is a small text file placed on your computer's hard drive or mobile device that is issued by the server of the website you are visiting.

The cookie contains a unique code that allows its issuer to recognize the terminal concerned (your computer or mobile device) each time the website is visited.

The cookie identifies your terminal's browser but never you personally. Furthermore, a website server can only read the cookies it has placed itself and has no access to any other information on your computer or mobile device.

Cookies can either be placed by the server of the website you are visiting ("first-party cookies") or by partners with whom this website collaborates ("third-party cookies").

## III. Cookie Validity Duration

Some cookies expire when you close your browser ("session cookies"), while others remain longer on your terminal, sometimes even until you manually delete them ("persistent cookies").

Generally, cookies have a limited validity period. The most common validity period is 30 days and the maximum duration is 13 months from the day the cookie is placed on your terminal.

## IV. General Purposes of Cookies

<CardGroup>
  <Card title="Enhanced Navigation" icon="compass">
    The primary purpose is to ensure more efficient and faster navigation (e.g., a cookie remembers the language you choose, identifies you when you access your account, etc.).
  </Card>

  <Card title="Personalized Experience" icon="user-cog">
    Cookies remember your preferences and allow you to speed up and facilitate your access to the site.
  </Card>
</CardGroup>

They also allow us to understand how you navigate our website.

In addition, they offer the possibility of making website content or advertising present on this site more relevant to you. The site will therefore be adapted to your personal tastes and needs.

## V. Cookie Management

To place certain cookies on your terminal, we need to obtain your prior and explicit consent. We obtain this through the banner placed on the website's homepage.

### Browser Settings

Most browsers use cookies. You can also delete these cookies or refuse their installation at any time and free of charge by modifying your browser software options. Each browser configuration is different. Each procedure is described in your browser's help menu:

<CardGroup>
  <Card title="Firefox" icon="firefox">
    [Delete cookies to remove information](https://support.mozilla.org/fr/kb/effacer-les-cookies-pour-supprimer-les-information)
  </Card>

  <Card title="Chrome" icon="chrome">
    [Manage cookies in Chrome](https://support.google.com/chrome/answer/95647?hl=fr)
  </Card>
</CardGroup>

<CardGroup>
  <Card title="Safari" icon="safari">
    [Manage cookies in Safari](https://support.apple.com/fr-be/guide/safari/sfri11471/mac)
  </Card>

  <Card title="Internet Explorer" icon="internet-explorer">
    [Delete and manage cookies](https://support.microsoft.com/fr-be/help/17442/windows-internet-explorer-delete-manage-cookies)
  </Card>
</CardGroup>

<Warning>
  By refusing the use of cookies, you can access our website. However, some features will be limited or even impossible.
</Warning>

## VI. Types of Cookies

<AccordionGroup>
  <Accordion title="Technical Cookies" icon="cog">
    These cookies, which only contain your IP address, are essential for visiting the site.

    They allow you to:

    * Browse different pages of the website and enjoy its features
    * Complete forms
    * Ensure verification of your identity when accessing your personal data
    * Secure access to the website

    **These cookies do not require your consent** as they enable you to access the website.
  </Accordion>

  <Accordion title="Functional Cookies" icon="sliders-h">
    These cookies facilitate and analyze the functioning of our website. Thanks to them, we can make your browsing experience more pleasant and personalized.

    Thanks to them we can:

    * Personalize services (language, currency, location, browsing data, etc.)
    * Remember your choices from a previous visit
    * Collect information from online forms
    * Establish statistics
    * Analyze site usage

    **Like technical cookies, these are used to improve your navigation and do not require your consent.**
  </Accordion>

  <Accordion title="Advertising Cookies" icon="bullhorn">
    These cookies give us indications about your browsing habits and allow us to adapt our website's advertising content to your interests.

    Based on these we can analyze the effectiveness of advertising campaigns, particularly by identifying subsequent actions you perform by clicking on these advertising banners.

    **Your consent is required for these cookies** and we ask for it when you access our website.
  </Accordion>

  <Accordion title="Analytical or Performance Cookies" icon="chart-line">
    These cookies are used to collect information about how you visit our website. They allow us to measure the audience of different contents and sections of our website to improve them.

    Thanks to these cookies we can detect navigation problems and improve our website's ergonomics. This data is aggregated and therefore completely anonymized.

    **We need your consent for these cookies.**
  </Accordion>

  <Accordion title="Social Media Cookies" icon="share-alt">
    Social media cookies share content you choose to share via application buttons belonging to social networks. When you use these buttons, a third-party cookie is installed on your terminal.

    For these cookies, you should check with each social network what the extent of data collection and use is.

    **These cookies also require your consent.**
  </Accordion>

  <Accordion title="Third-Party Cookies" icon="external-link">
    These cookies are placed by third parties on our website and then on your terminal. We have no control over them. To learn more, please consult the information provided by these third-party sites about their own cookies.
  </Accordion>
</AccordionGroup>

## VII. Cookie List

<div>
  <table>
    <thead>
      <tr>
        <th>Provider</th>
        <th>Type</th>
        <th>Cookie Name</th>
        <th>Purpose</th>
        <th>Retention Duration</th>
      </tr>
    </thead>

    <tbody>
      <tr>
        <td>Wonka</td>
        <td>Technical/Necessary</td>
        <td>JWT</td>
        <td>Authenticate user and maintain active session</td>
        <td>Session duration</td>
      </tr>

      <tr>
        <td>Wonka</td>
        <td>Technical/Necessary</td>
        <td>refreshToken</td>
        <td>Refresh token identifier + metadata</td>
        <td>7 days</td>
      </tr>

      <tr>
        <td>Wonka</td>
        <td>Technical/Necessary</td>
        <td>token\_provider</td>
        <td>Name of chosen provider</td>
        <td>71 days</td>
      </tr>
    </tbody>
  </table>
</div>

## VIII. Your Rights Regarding Personal Data Processing

Cookies that process personal data are subject to personal data protection rules. In this context, you have certain rights (right of access, right of rectification, right of objection, right to withdraw consent, etc.).

For detailed information about your rights and how to exercise them, refer to our <a href="/en/legal/privacy-policy">Privacy Policy</a>.

## VIII. Contact

<Card title="Contact Information" icon="envelope">
  **Email:** [support@meetwonka.com](mailto:support@meetwonka.com)

  **Postal Address:**\
  Wonka AI BV\
  Luikersteenweg 165\
  3500 Hasselt
</Card>

## IX. Modifications

We reserve the right to modify the provisions of this Policy at any time. We will publish modifications directly on our website.

## X. Applicable Law and Competent Jurisdictions

This Policy is governed by Belgian national law.

Any dispute relating to the interpretation or execution of this Policy will be submitted to the Brussels courts.

<Note>
  This version of the Policy comes into force and was updated on November 28, 2025.
</Note>


# Legal Notice
Source: https://help.wonka.chat/en/legal/legal-notice

Legal information about WonkaChat, including publisher details, intellectual property, hosting, and applicable terms.

# Legal Notice

**Last updated: November 28, 2025**

## Site Publisher

<Card title="Wonka AI BV" icon="building">
  **Address:** Luikersteenweg 165, 3500 Hasselt\
  **BCE:** 0800 787 953\
  **Contact:** [support@meetwonka.com](mailto:support@meetwonka.com)
</Card>

## Development and Intellectual Property

<Card title="Wonka AI BV" icon="code">
  **Primary Developer and IP Owner**

  The user interface, visual layout, and general structure of the WonkaChat platform constitute and remain the exclusive intellectual property of Wonka AI BV.
</Card>

### Platform Features

The platform includes various features and functionalities:

* AI-powered conversational interface
* Multi-industry support capabilities
* Customizable agent configurations
* Integration capabilities

<Note>
  All design components, layouts, technological foundations, and features are the exclusive property of Wonka AI BV.
</Note>

## Hosting

<AccordionGroup>
  <Accordion title="Hosting Location" icon="globe-europe">
    <strong>Region:</strong> West Europe
  </Accordion>

  <Accordion title="Hosting Provider" icon="cloud">
    <strong>Microsoft Azure</strong><br />
    Microsoft Ireland Operations Limited<br />
    One Microsoft Place<br />
    South County Business Park<br />
    Leopardstown, Dublin 18<br />
    D18 P521, Ireland
  </Accordion>
</AccordionGroup>

## Terms of Use

<Warning>
  Access to WonkaChat is available through subscription to any individual or organization and is subject to the <a href="/en/legal/terms-of-service">General Terms of Use (GTU)</a> available on the site.
</Warning>

## Data Protection

Your personal data is processed in compliance with GDPR. Consult our Privacy Policy for more information.

<Card title="Privacy Policy" icon="shield-check" href="/en/legal/privacy-policy">
  Learn about how we protect and process your personal data
</Card>

## Liability

<Card title="AI Assistance Platform" icon="robot">
  WonkaChat is an AI-powered conversational platform. Users remain solely responsible for verifying and using the information provided.
</Card>

## Disputes

<Info>
  For details on applicable law and dispute resolution, refer to the <a href="/en/legal/terms-of-service">General Terms and Conditions</a>. WonkaChat operates under Belgian law, with disputes handled in accordance with the procedures described in our Terms of Service.
</Info>

## Last Update

<Card title="Version Information" icon="calendar">
  **Date:** November 28, 2025
</Card>

<Note>
  For questions about these legal notices, please contact us at [support@meetwonka.com](mailto:support@meetwonka.com)
</Note>


# Privacy Policy
Source: https://help.wonka.chat/en/legal/privacy-policy

Learn how WonkaChat collects, uses, and protects your personal data in compliance with GDPR and EU regulations.

# Privacy Policy

**WonkaChat**\
*(Last updated: November 28, 2025)*

## 1. About This Policy

This Privacy Policy (the "Policy") describes how we (as defined below) collect, share, and use any information that, used alone or in combination with other information, relates to you ("Personal Data") when you ("you" and "your", "User") create an account to access the AI conversational platform ("WonkaChat") made available to you through our platform.

The Policy also applies to account access management, collection of aggregated data for statistical or tracking purposes, and communication of information relating to WonkaChat's activities and operation.

For any questions about Personal Data collection outside of WonkaChat, please contact us at [support@meetwonka.com](mailto:support@meetwonka.com).

Please take the time to carefully read this Policy. If you have questions or comments, please contact our data protection officer at: **[support@meetwonka.com](mailto:support@meetwonka.com)**.

## Data Controller and Processor Roles

For the purposes of this Policy, **Wonka AI BV**, with its registered office at Luikersteenweg 165, 3500 Hasselt, registered with the Belgian Crossroads Bank for Enterprises under number 0800 787 953, ("Wonka AI", "we", "our") acts as data controller for Personal Data collected to offer its service, WonkaChat.

However, the User acts as data controller under applicable data protection legislation regarding the use of WonkaChat and all data they integrate into it. In this context, Wonka AI acts as data processor.

This Policy sets out Wonka AI's commitments regarding data protection and the measures implemented to guarantee the security and confidentiality of your personal data. It also specifies the rights you have in this regard and the practical procedures for exercising them with us.

## 2. Categories of Data Collected and Purposes

The types of Personal Data we collect and the reasons we process them include:

<div>
  <table>
    <thead>
      <tr>
        <th>Processing Purpose</th>
        <th>Types of Data Processed</th>
        <th>Legal Basis</th>
      </tr>
    </thead>

    <tbody>
      <tr>
        <td>User account creation and management</td>
        <td>Name, surname, professional email address, password, organization/profession</td>
        <td>Contract execution</td>
      </tr>

      <tr>
        <td>Secure access and WonkaChat maintenance</td>
        <td>Technical and connection data (logs, identifiers)</td>
        <td>Legitimate interest (security and operation)</td>
      </tr>

      <tr>
        <td>Statistical analysis and growth measurement</td>
        <td>Aggregated tracking and internet connection data</td>
        <td>Consent</td>
      </tr>

      <tr>
        <td>Communication of information or news</td>
        <td>Name, surname, professional email</td>
        <td>Consent, legitimate interest</td>
      </tr>
    </tbody>
  </table>
</div>

If we were to request other Personal Data not mentioned above, we will clearly indicate, at the time of collection, the nature of the information requested and the reasons for this request.

Some Personal Data may also be obtained indirectly, for example when a User associates you with their account to allow you access to their space.

### Automatically Collected Technical Information

We may automatically collect certain technical information related to your device, including:

* IP address
* Device type used
* Unique identifiers
* Browser type
* Approximate location (country or city)
* Other technical data

At this time, we do not collect any information about your interaction with WonkaChat, such as pages viewed or features used. If we choose to collect such data in the future, it will be solely for internal analysis and to enhance the relevance and overall experience of WonkaChat.

Some of this information may be collected through cookies or similar technologies, in accordance with our <a href="/en/legal/cookie-policy">Cookie Policy</a>.

## 3. Recipients of Your Personal Data

We may transmit your Personal Data to the following categories of recipients:

<CardGroup>
  <Card title="Technical Providers & Subcontractors" icon="wrench">
    Developers, hosts, analytics tool providers, or support providers acting under strict Wonka AI instructions. We require these subcontractors to process Personal Data strictly according to our instructions and take appropriate measures to ensure Personal Data remains protected.
  </Card>

  <Card title="Authorities or Public Bodies" icon="shield-check">
    Any competent law enforcement body, regulator, government agency, court, or other third party when we believe disclosure is necessary under applicable laws or regulations, or to establish or defend our rights, or to protect your vital interests or those of any other person.
  </Card>
</CardGroup>

<CardGroup>
  <Card title="External Advisors" icon="users">
    Auditors, advisors, legal representatives, and similar agents in the context of advisory services they provide to us and subject to confidentiality commitments.
  </Card>

  <Card title="Authorized Third Parties" icon="user-check">
    Any other person provided you have given prior consent to disclosure.
  </Card>
</CardGroup>

## 4. Protection Principles

In accordance with this Policy, we will process Personal Data as follows:

<AccordionGroup>
  <Accordion title="Fairness" icon="balance-scale">
    Personal Data will be processed fairly and transparently. We commit to clearly inform about processing methods and act in compliance with applicable legislation.
  </Accordion>

  <Accordion title="Lawfulness" icon="gavel">
    No processing will be carried out without a valid legal basis; any use of Personal Data will rest on a legal foundation.
  </Accordion>

  <Accordion title="Purpose Limitation" icon="tools">
    Personal Data will only be collected and processed for specific, explicit, and legitimate purposes. They will not be subject to any subsequent use incompatible with these initial purposes.
  </Accordion>

  <Accordion title="Data Minimisation" icon="minimize">
    Data is adequate, relevant, and limited to what is necessary for the purposes for which it is processed.
  </Accordion>

  <Accordion title="Accuracy" icon="check-circle">
    We implement reasonable measures to ensure Personal Data is accurate, complete, and regularly updated when needed. However, you remain obliged to notify us without delay of any changes or inaccuracies to maintain the accuracy of your information.
  </Accordion>

  <Accordion title="Integrity and Confidentiality" icon="shield">
    Personal Data is processed to ensure its security, including protection against unauthorized access, unlawful processing, loss, destruction, or accidental damage, through appropriate technical and organizational measures.
  </Accordion>

  <Accordion title="Accountability" icon="clipboard-list">
    Wonka AI assumes responsibility for compliance with these principles and is able to demonstrate, at any time, the compliance of implemented processing, notably through maintaining adequate documentation, establishing internal procedures, and conducting impact analyses or compliance audits when required.
  </Accordion>
</AccordionGroup>

## 5. Security

We use appropriate technical and organizational measures to protect the Personal Data we collect and process about you. The measures we use are designed to provide a level of security appropriate to the risk of processing your Personal Data.

### Security Measures Include:

<CardGroup>
  <Card title="Data Encryption" icon="lock">
    **In Transit & At Rest**

    All communications protected by HTTPS/TLS (TLS 1.3, with TLS 1.2 support if necessary). Data encrypted at rest.
  </Card>

  <Card title="Environment Isolation" icon="server">
    **Strict Tenant Isolation**

    Each user's data (conversations, configurations, user accounts, audit logs, connections) is logically separated and protected against unauthorized access.
  </Card>
</CardGroup>

<CardGroup>
  <Card title="Secure Credential Management" icon="key">
    **Protected Authentication**

    Passwords are hashed (bcrypt), access tokens (JWT) are cryptographically signed, and sensitive keys/APIs are never stored in plain text.
  </Card>

  <Card title="Enhanced Authentication" icon="user-shield">
    **Limited Session Duration**

    Short-duration access tokens and rotating refresh tokens limit risks. We support local authentication, Google Authentication, and Azure AD/Entra ID.
  </Card>
</CardGroup>

### Additional Security Features

<Tabs>
  <Tab title="Access Controls">
    **Principle of Least Privilege**

    Each user, agent, or tool can only operate within the limits of permissions explicitly assigned to them. No internal mechanism allows AI to bypass authorizations.
  </Tab>

  <Tab title="Audit & Monitoring">
    **Structured Logging**

    We record security-relevant events: authentication attempts, permission changes, suspicious activity, data access, agent/tool executions, and migrations. All sensitive data is automatically masked.
  </Tab>

  <Tab title="Abuse Prevention">
    **Operational Protection**

    Rate limiting mechanisms protect against abusive automated usage or attacks.
  </Tab>

  <Tab title="External Testing">
    **Penetration Testing**

    Independent penetration testing was conducted in November 2025, and all identified critical vulnerabilities have been corrected.
  </Tab>
</Tabs>

<Note>
  **Ephemeral Data Processing by Model Providers:** When an external model (Azure, AWS Bedrock, Google) is used, data is processed only in memory and is never retained or reused for training.
</Note>

## 6. International Data Transfers

Your Personal Data may be transferred to and processed in countries other than where you reside. These countries may have data protection laws that differ from your own country's laws and, in some cases, may be less protective.

<CardGroup>
  <Card title="EU Hosting" icon="globe-europe">
    **Primary Location**

    Our servers are located within the European Economic Area (EEA).
  </Card>

  <Card title="Third Country Transfers" icon="plane">
    **Protected Transfers**

    Some service providers may be established outside the EEA. We ensure these transfers comply with GDPR Chapter V and guarantee adequate protection.
  </Card>
</CardGroup>

### Transfer Safeguards

We use one or more of the following mechanisms:

* European Commission adequacy decision
* Standard contractual clauses adopted by the European Commission, with additional measures if necessary
* Any other appropriate safeguards provided by GDPR

<Warning>
  We do not transfer any data outside the EEA without implementing these safeguards and ensuring data subjects have enforceable rights and effective remedies.
</Warning>

## 7. Data Retention

We retain Personal Data we collect from you when we have a legitimate business need (for example, to provide a service you requested or to comply with applicable legal requirements).

### Retention Periods:

<CardGroup>
  <Card title="Account Management" icon="user">
    **1 year** from last activity
  </Card>

  <Card title="Support Requests" icon="headset">
    **1 year** after last contact date
  </Card>
</CardGroup>

<CardGroup>
  <Card title="User Experience Improvement" icon="lightbulb">
    **1 year** from last activity
  </Card>

  <Card title="Statistical Analysis" icon="chart-bar">
    **1 year** from last activity
  </Card>
</CardGroup>

When we no longer have a legitimate business need to process your Personal Data, we anonymize it, delete it, or if deletion is not possible (for example, your Personal Data has been stored in backup archives), we securely store and isolate it from any other processing until deletion becomes possible.

## 8. Your Data Protection Rights

You have the following data protection rights, which you can exercise by contacting us at **[support@meetwonka.com](mailto:support@meetwonka.com)**:

<AccordionGroup>
  <Accordion title="Access, Rectification, Update & Erasure" icon="edit">
    You may request access to your Data, correct it if inaccurate, update it, or request its deletion.
  </Accordion>

  <Accordion title="Objection, Limitation & Portability" icon="arrows-alt">
    In certain circumstances, you may object to processing of your Data, request limitation of their use, or seek portability of your Data to yourself or a third party.
  </Accordion>

  <Accordion title="Consent Withdrawal" icon="undo">
    When processing is based on your consent, you may withdraw it at any time. This withdrawal will not affect the lawfulness of processing carried out before withdrawal, nor processing based on other legal bases (such as contract or legal obligation).
  </Accordion>

  <Accordion title="Lodge a Complaint" icon="exclamation-triangle">
    If you have concerns about how we process your Data, we invite you to contact us first. If you feel your request has not been sufficiently addressed, you have the right to lodge a complaint with the competent supervisory authority.
  </Accordion>
</AccordionGroup>

### Contact the Data Protection Authority

**Belgian Data Protection Authority**\
Rue de la Presse 35\
1000 Brussels\
Phone: +32 (0)2 274 48 00\
Email: [contact@apd-gba.be](mailto:contact@apd-gba.be)\
Website: [www.autoriteprotectiondonnees.be](http://www.autoriteprotectiondonnees.be)

We respond to all requests we receive from individuals wishing to exercise their Personal Data protection rights in accordance with applicable data protection laws.

## 9. Policy Updates

We may revise this Policy from time to time to account for evolving legal, technical, or organizational requirements. In case of substantial changes, we will take appropriate measures to inform you, depending on the nature and impact of the changes.

The date of the last update appears at the top of this Policy and allows you to verify the most recent version.

## 10. Contact

If you have questions about the processing of your Personal Data or wish to exercise your rights, please contact us by email at **[support@meetwonka.com](mailto:support@meetwonka.com)**.

<Card title="Data Protection Officer" icon="shield-check">
  **Email:** [support@meetwonka.com](mailto:support@meetwonka.com)\
  **Subject:** Data Protection Inquiry - WonkaChat
</Card>


# Terms and Conditions
Source: https://help.wonka.chat/en/legal/terms-and-conditions

Review the terms and conditions governing your use of the WonkaChat platform, including ownership, data usage, user responsibilities, and legal policies.

# Terms and Conditions for WonkaChat

*Effective Date: February 18, 2024*

Welcome to WonkaChat, the informational website for the open-source AI chat platform, available at [https://meetwonka.com](https://meetwonka.com) These Terms of Service ("Terms") govern your use of our website and the services we offer. By accessing or using the Website, you agree to be bound by these Terms and our Privacy Policy, accessible at [https://meetwonka.com](https://meetwonka.com).

## 1. Ownership

Upon purchasing a package from WonkaChat, you are granted the right to download and use the code for accessing an admin panel for WonkaChat. While you own the downloaded code, you are expressly prohibited from reselling, redistributing, or otherwise transferring the code to third parties without explicit permission from WonkaChat.

## 2. User Data

We collect personal data, such as your name, email address, and payment information, as described in our Privacy Policy. This information is collected to provide and improve our services, process transactions, and communicate with you.

## 3. Non-Personal Data Collection

The Website uses cookies to enhance user experience, analyze site usage, and facilitate certain functionalities. By using the Website, you consent to the use of cookies in accordance with our Privacy Policy.

## 4. Use of the Website

You agree to use the Website only for lawful purposes and in a manner that does not infringe the rights of, restrict, or inhibit anyone else's use and enjoyment of the Website. Prohibited behavior includes harassing or causing distress or inconvenience to any person, transmitting obscene or offensive content, or disrupting the normal flow of dialogue within the Website.

## 5. Governing Law

These Terms shall be governed by and construed in accordance with the laws of the United States, without giving effect to any principles of conflicts of law.

## 6. Changes to the Terms

We reserve the right to modify these Terms at any time. We will notify users of any changes by email. Your continued use of the Website after such changes have been notified will constitute your consent to such changes.

## 7. Contact Information

If you have any questions about these Terms or if you would like to unsubscribe, please contact us at [support@meetwonka.com](mailto:support@meetwonka.com).

By using the Website, you acknowledge that you have read these Terms of Service and agree to be bound by them.


# Terms of Service
Source: https://help.wonka.chat/en/legal/terms-of-service

General Terms and Conditions for using WonkaChat.

# General Terms and Conditions of Use for WonkaChat

**Last updated: November 28, 2025**

## Table of Contents

<AccordionGroup>
  <Accordion title="1. Preamble and Scope" icon="file-text">
    Overview of WonkaChat and acceptance of terms
  </Accordion>

  <Accordion title="2. Definitions" icon="book">
    Key terms and their meanings
  </Accordion>

  <Accordion title="3. Access and Availability" icon="globe">
    Eligibility, account creation, and platform availability
  </Accordion>

  <Accordion title="4. Use of WonkaChat" icon="user">
    Permitted and prohibited uses
  </Accordion>

  <Accordion title="5. User License" icon="key">
    Rights and limitations of platform use
  </Accordion>

  <Accordion title="6. Intellectual Property" icon="copyright">
    Ownership and protection of platform elements
  </Accordion>

  <Accordion title="7. Personal Data Protection" icon="shield">
    Privacy and data protection commitments
  </Accordion>

  <Accordion title="8. Responsibilities" icon="balance-scale">
    Wonka AI and User obligations
  </Accordion>

  <Accordion title="9. Security" icon="lock">
    Security measures and incident reporting
  </Accordion>

  <Accordion title="10. Suspension and Exclusion" icon="ban">
    Enforcement measures for violations
  </Accordion>

  <Accordion title="11. Invalidity" icon="exclamation-triangle">
    Handling of invalid provisions
  </Accordion>

  <Accordion title="12. Applicable Law and Jurisdiction" icon="gavel">
    Legal framework and dispute resolution
  </Accordion>

  <Accordion title="13. Service and Terms Modifications" icon="edit">
    Changes to platform and terms
  </Accordion>

  <Accordion title="14. Contact" icon="envelope">
    Support and contact information
  </Accordion>
</AccordionGroup>

***

## 1. Preamble and Scope

**WonkaChat** is a conversational artificial intelligence tool developed by Wonka AI and made available to Users through a subscription model.

It allows Users to access an automated discussion environment, designed to facilitate the search for technical and administrative information.

<Warning>
  **Acceptance of Terms:** Using WonkaChat implies full acceptance of these Terms and Conditions (T\&C), accessible at any time on [Terms and Conditions](/legal/terms-and-conditions).
</Warning>

By accessing WonkaChat, the User acknowledges having read these General Terms and Conditions and accepts them without reservation. These T\&C exclude all other general conditions and replace them.

### Important Notes

* Wonka AI reserves the right to modify the T\&C at any time, without prior notification
* Modifications will take effect upon online publication
* Any additional rules and guidelines will be considered an integral part of the T\&C
* Users are advised to regularly refer to the latest version of these conditions

## 2. Definitions

<CardGroup>
  <Card title="WonkaChat" icon="robot">
    Conversational AI platform developed by Wonka AI, available through subscription to individuals and organizations across all industries to enhance productivity, research, and digital innovation
  </Card>

  <Card title="Wonka AI" icon="building">
    Wonka AI BV, with its registered office at Luikersteenweg 165, 3500 Hasselt, KBO (Crossroads Bank for Enterprises) number 0800 787 953, manages this website.
  </Card>
</CardGroup>

<CardGroup>
  <Card title="User" icon="user">
    Any natural person who consults WonkaChat or uses its functionalities, including (but not limited to) uploading files.
  </Card>

  <Card title="Subscribers" icon="handshake">
    Individuals and organizations who have purchased a subscription to access and use WonkaChat services
  </Card>
</CardGroup>

<Note>
  **GDPR Terms:** The terms "controller", "processor", "personal data", "processing", "data subject", and other data protection-related terms have the meaning attributed to them by Regulation (EU) 2016/679 of April 27, 2016 (GDPR).
</Note>

## 3. Access and Availability

### Eligibility

WonkaChat is accessible to any individual or organization who has purchased a valid subscription to the platform.

### Account Creation

<Steps>
  <Step title="Accurate Information">
    When creating an account, the User commits to providing accurate and up-to-date information. Wonka AI reserves the right to suspend or refuse any access in case of inaccuracy or failure.
  </Step>

  <Step title="Credential Security">
    The User commits to preserving the confidentiality of their identifier and password and not to disclose them to third parties. They remain solely responsible for any use of their account, even without their knowledge.
  </Step>

  <Step title="Security Concerns">
    In case of doubt about password security, immediately modify it or inform Wonka AI in writing at: **[support@meetwonka.com](mailto:support@meetwonka.com)**
  </Step>
</Steps>

<Warning>
  **Personal Access:** The identifier and password are strictly personal to the User. Sharing access to WonkaChat with third parties is prohibited.
</Warning>

### Platform Availability

<CardGroup>
  <Card title="Maintenance" icon="wrench">
    Access to WonkaChat may be temporarily interrupted for maintenance or improvement, without notice
  </Card>

  <Card title="Compatibility" icon="desktop">
    WonkaChat is accessible from any device with recent operating system and browser. Wonka AI cannot guarantee optimal functioning on all possible configurations
  </Card>
</CardGroup>

### User Responsibilities

<AccordionGroup>
  <Accordion title="Equipment Protection" icon="shield">
    The User must ensure protection of their equipment and data by adopting appropriate security measures, such as installing antivirus software, firewalls, and malware protection tools.
  </Accordion>

  <Accordion title="Feature Evolution" icon="tools">
    The User acknowledges that functionalities offered via WonkaChat may evolve. Some features may be modified, deleted, or replaced, without the User being able to invoke any right to maintain a specific functionality.
  </Accordion>

  <Accordion title="Service Modifications" icon="edit">
    Wonka AI reserves the right, at any time and for any reason, to suspend, modify, or temporarily or permanently interrupt access to all or part of WonkaChat, without obligation to inform Users beforehand.
  </Accordion>
</AccordionGroup>

## 4. Use of WonkaChat

### Permitted Use

The User commits to using WonkaChat in a manner consistent with the law, these T\&C, and the platform's purpose.

<Note>
  **AI Assistant Disclaimer:** The User acknowledges that responses generated by WonkaChat do not constitute legal, technical, or professional advice and that they remain responsible for decisions made and content generated.
</Note>

### Prohibited Activities

<CardGroup>
  <Card title="System Interference" icon="ban">
    **Prohibited:**

    * Attempting to access non-public system areas
    * Any act that could disrupt the platform's functioning
    * Using automated systems (robots, spiders, crawlers) for attacks or manipulation
  </Card>

  <Card title="Illegal Content" icon="exclamation-triangle">
    **Prohibited:**

    * Uploading, transmitting, or distributing illegal content
    * Infringing on third-party intellectual property rights
    * Introducing viruses, malware, or harmful code
  </Card>
</CardGroup>

<Warning>
  **Network Disruption:** Any action designed to disrupt, hinder, or limit other Users' access to WonkaChat or the platform's proper functioning is strictly prohibited.
</Warning>

## 5. User License

### License Grant

The User obtains only a right of access and personal use of WonkaChat. They benefit from a **non-exclusive, non-transferable, non-assignable license** strictly limited to internal and non-commercial use.

<CardGroup>
  <Card title="Permitted" icon="check">
    * Personal, internal use
    * Access during authorized period
    * Educational and professional development purposes
  </Card>

  <Card title="Prohibited" icon="x">
    * Commercial exploitation
    * Third-party access provision
    * Reproduction or distribution
    * Reverse engineering
  </Card>
</CardGroup>

### Commercial Use Prohibition

<Warning>
  **Strictly Prohibited:** Any commercial use of WonkaChat including sale, rental, paid provision, or exploitation of functionalities or content generated via WonkaChat.
</Warning>

### Technical Restrictions

The User is strictly prohibited from:

* Making WonkaChat available to unauthorized third parties
* Reproducing, copying, modifying, or distributing the platform
* Creating derivative works based on WonkaChat elements
* Reverse engineering, decompilation, or disassembly

## 6. Intellectual Property

### Protected Elements

<Card title="Intellectual Property Rights" icon="copyright">
  The concept, layout, structure, source codes, programming elements, images, photographs, logos, drawings, trademarks, models, slogans, and all other elements composing WonkaChat (the "Protected Elements") are and remain the exclusive property of their legitimate holder, including the platform developer and/or Wonka AI.
</Card>

### User Rights Limitation

<Note>
  By accessing, registering, connecting, downloading content, or using WonkaChat in any way, the User acquires no ownership or license rights to the Protected Elements, other than the limited usage right provided in these General Terms and Conditions.
</Note>

### Developer Guarantee

The Developer guarantees that elements made available to the User via WonkaChat do not infringe on third-party rights and are not illegal.

## 7. Personal Data Protection

### GDPR Compliance

<Card title="Privacy Commitment" icon="shield-check">
  Wonka AI attaches particular importance to respecting privacy and protecting Users' personal data, in accordance with Regulation (EU) 2016/679 of April 27, 2016 ("GDPR").
</Card>

<Info>
  WonkaChat is not yet formally GDPR certified, but adheres to GDPR guidelines and strives to obtain official certification as soon as possible.
</Info>

### Privacy Policy Reference

Users are invited to consult WonkaChat's Privacy Policy, available on the platform or via the official site, which describes in detail the processing methods for their personal data.

### User Rights

<CardGroup>
  <Card title="Data Rights" icon="user-shield">
    Users have rights of access, rectification, objection, and erasure of their data
  </Card>

  <Card title="Contact" icon="envelope">
    Please send a mail to [support@meetwonka.com](mailto:support@meetwonka.com)
  </Card>
</CardGroup>

## 8. Responsibilities

### A. Wonka AI Responsibility

<Card title="Best Efforts Obligation" icon="tools">
  Wonka AI provides WonkaChat according to a best efforts obligation. No guarantee of results is given and Wonka AI cannot be held responsible for direct or indirect damages resulting from access, use, or inability to use the platform.
</Card>

#### Service Availability

Wonka AI deploys reasonable efforts to guarantee WonkaChat's availability at every moment, for a normal number of Users, but cannot be held responsible for any direct or indirect damage related to modification, suspension, or interruption of access to WonkaChat, for any cause whatsoever.

#### Limitations of Liability

<AccordionGroup>
  <Accordion title="Technical Compatibility" icon="desktop">
    Wonka AI does not guarantee technical compatibility of the platform with the User's equipment, nor continuous or uninterrupted access to content.
  </Accordion>

  <Accordion title="Internet Risks" icon="globe">
    The User expressly acknowledges the limits and risks inherent to internet and online technology use, including risks of data loss, alteration, or interception during electronic transmissions.
  </Accordion>

  <Accordion title="Content Generation" icon="robot">
    Wonka AI disclaims all responsibility for content generated by WonkaChat. No guarantee is given regarding accuracy, completeness, or updates of responses provided by the tool.
  </Accordion>
</AccordionGroup>

### B. User Responsibility

<Warning>
  **Sole Responsibility:** Use of WonkaChat, including platform consultation and content download, occurs under the User's sole responsibility, including toward third parties.
</Warning>

#### AI Tool Limitations

<Card title="Important Understanding" icon="exclamation-triangle">
  **WonkaChat is an AI-powered assistance tool.** Responses are automatically generated based on available information and formulated queries, without human contextual analysis. They do not replace the User's appreciation, judgment, or expertise. The User remains responsible for generated content.
</Card>

#### User Obligations

<Steps>
  <Step title="Content Responsibility">
    The User acknowledges being solely responsible for content introduced via the platform and decisions made based on generated responses.
  </Step>

  <Step title="Verification Duty">
    It is the User's responsibility to verify, complete, or validate obtained results, considering inherent limitations of any generative AI system.
  </Step>

  <Step title="Usage Guidelines">
    Before any use, the User must read indications regarding WonkaChat's functioning and follow usage recommendations.
  </Step>
</Steps>

## 9. Security

### Security Measures

<Card title="Protection Commitment" icon="shield">
  Wonka AI implements technical and organizational measures designed to protect data and system integrity.
</Card>

### User Security Obligations

The User commits to:

* Not compromising the platform's security
* Not attempting to alter its components
* Immediately reporting security incidents to [support@meetwonka.com](mailto:support@meetwonka.com)

## 10. Suspension and Exclusion

<Warning>
  **Enforcement Rights:** Wonka AI reserves the right to suspend or exclude any User in case of total or partial violation of obligations provided by these T\&C. This exclusion may be temporary or permanent, depending on the severity of the breach and at Wonka AI's sole discretion.
</Warning>

## 11. Invalidity

<Note>
  **Severability:** The possible nullity or invalidity of an article, paragraph, or provision of these T\&C (or part thereof) will not affect the validity of other stipulations or the rest of the concerned provision, unless manifest contrary intention.
</Note>

If a provision should be declared totally invalid, Wonka AI will replace it with a clause producing, as far as possible, an economic effect equivalent to that initially sought.

## 12. Applicable Law and Competent Jurisdiction

### Governing Law

<Card title="Belgian Law" icon="gavel">
  These General Terms and Conditions are governed by Belgian law.
  In the event of a dispute, only the courts in Belgium shall have jurisdiction.
</Card>

### Dispute Resolution

<Steps>
  <Step title="Amicable Settlement">
    The Parties commit to settling any potential dispute amicably.
  </Step>

  <Step title="Mediation">
    In case of disputes that cannot be settled amicably, Parties agree to resort primarily to a mediation process, prior to any legal action.
  </Step>

  <Step title="Mediator Selection">
    In case of disagreement on mediator choice, they will be designated by the President of the French-speaking Brussels business court, among approved mediators.
  </Step>

  <Step title="Court Jurisdiction">
    In case of mediation failure, courts of the Brussels judicial district will have sole jurisdiction over disputes.
  </Step>
</Steps>

### Mediation Process

* Parties will set practical mediation methods and process duration with mediator assistance
* Each Party retains the right to terminate at any time
* Minimum commitment to participate in first mediation session
* Unless otherwise agreed, mediation costs shared equally between Parties

## 13. Service and Terms Modifications

### Platform Modifications

<Card title="Evolution Rights" icon="edit">
  Wonka AI may at any time modify, improve, or temporarily interrupt WonkaChat.
</Card>

### Terms Updates

* T\&C may be adapted to reflect technical, legal, or organizational changes
* Modifications are enforceable upon publication on the site
* Users advised to regularly check for updates

## 14. Contact

<Card title="Support Information" icon="headset">
  **For any questions regarding WonkaChat or these T\&C:**

  **Wonka AI**\
  ç  Luikersteenweg 165\
  3500 Hasselt

  **Email:** [support@meetwonka.com](mailto:support@meetwonka.com)
</Card>

***

<Note>
  **Document Version:** These Terms and Conditions were last updated on November 28, 2025. Users are responsible for staying informed of the most current version.
</Note>


# Pricing & Plans
Source: https://help.wonka.chat/en/pricing/plans

Choose the right plan for your team size and needs

WonkaChat offers flexible pricing options designed to scale with your team, from small startups to large enterprises.

***

## Enterprise Packs

Pre-configured packages optimized for different team sizes with included personalized support and workflow setup.

<CardGroup>
  <Card title="For small teams" icon="users">
    960€/month

    Up to 20 users included

    Includes:

    * 4h/month of personalized support
    * 1 simple workflow optimized

    <Info>
      Tokens not included\*
    </Info>
  </Card>

  <Card title="For growing teams" icon="chart-line">
    3'000€/month

    Up to 100 users included

    Includes:

    * 8h/month of personalized support
    * 1 complex workflow optimized

    <Info>
      Tokens not included\*
    </Info>
  </Card>

  <Card title="For larger SMEs" icon="building">
    9'400€/month

    Up to 200 users included

    Includes:

    * 1 day/week (on site) strategy, personalized support and workflows creation

    <Info>
      Tokens not included\*
    </Info>
  </Card>
</CardGroup>

<Info>
  \*Tokens are billed separately based on AI model usage. See [Token Pricing Guide](/en/pricing/tokens-guide) for detailed information.
</Info>

<Card title="Interested in an Enterprise Pack?" icon="handshake">
  Contact our Sales Support team to discuss which pack best fits your organization's needs.

  [Contact Sales Support](/en/support-resources/getting-help)
</Card>

***

## Additional Users

Need to add more users beyond your pack limit?

<CardGroup>
  <Card title="Per User Pricing" icon="user-plus">
    25€ per user/month

    Add individual users to any Enterprise Pack as your team grows.
  </Card>

  <Card title="Onboarding Support" icon="users-gear">
    960€ per day

    Full-day onboarding and training sessions for your team to maximize WonkaChat adoption.
  </Card>
</CardGroup>

<Card title="Need More Seats?" icon="envelope">
  Contact our Sales Support team to purchase additional user licenses.

  [Contact Sales Support](/en/support-resources/getting-help)
</Card>

***

## Frequently Asked Questions

<AccordionGroup>
  <Accordion title="Can I switch between plans?" icon="arrows-rotate">
    Yes, you can upgrade or downgrade your Enterprise Pack at any time. Contact Sales Support to make changes to your subscription.
  </Accordion>

  <Accordion title="What happens if I exceed my user limit?" icon="triangle-exclamation">
    You can purchase additional user seats at 25€/user/month. Your service will not be interrupted when you exceed your pack limit.
  </Accordion>

  <Accordion title="How are tokens billed?" icon="coins">
    Tokens are billed monthly based on actual consumption across all your agents and users. We provide detailed usage analytics in your dashboard.
  </Accordion>

  <Accordion title="Is there a free trial?" icon="clock">
    Contact our Sales team to discuss trial options for your organization and specific business needs.
  </Accordion>

  <Accordion title="What's included in personalized support?" icon="headset">
    Personalized support includes dedicated time with our team to optimize your workflows, train your team, troubleshoot issues, and ensure you're getting maximum value from WonkaChat.
  </Accordion>
</AccordionGroup>

***

## Ready to Get Started?

<CardGroup>
  <Card title="Contact Sales" icon="phone" href="/en/support-resources/getting-help">
    Speak with our team to find the right plan for your organization.
  </Card>

  <Card title="Quick Start Guide" icon="rocket" href="/en/welcome/quick-start">
    Already have an account? Get started with WonkaChat in 5 minutes.
  </Card>
</CardGroup>


# Understanding Token Pricing
Source: https://help.wonka.chat/en/pricing/tokens-guide

A simple guide to understanding how AI model pricing works and choosing the right model for your needs

## What Are Tokens?

<Info>
  Tokens are the units that measure AI model usage. Think of them as the "currency" that AI models consume when they process your requests.

  Simple analogy: If AI models were cars, tokens would be gallons of gas. Different cars (models) use different amounts of gas (tokens) to go the same distance.
</Info>

### How to Think About Tokens

* 1 token ≈ 3/4 of a word in English
* "Hello, how are you?" = approximately 6 tokens
* A typical email (200 words) = approximately 265 tokens
* A full page of text (500 words) = approximately 665 tokens

<Check>
  Both your question AND the AI's answer consume tokens. You pay for both.
</Check>

***

## Why Different Models Cost Different Amounts

Just like cars, AI models come in different sizes with different capabilities:

<CardGroup>
  <Card title="Economy Models" icon="piggy-bank">
    0.04 to 0.15 USD per 1M tokens

    Like a compact car - efficient for everyday tasks.

    Best for:

    * Simple questions
    * Categorization
    * Basic summaries
    * High-volume tasks
  </Card>

  <Card title="Standard Models" icon="gauge">
    0.60 to 3.00 USD per 1M tokens

    Like a sedan - reliable for most needs.

    Best for:

    * Business emails
    * Reports
    * Research
    * General conversations
  </Card>

  <Card title="Premium Models" icon="rocket">
    10.00 to 75.00 USD per 1M tokens

    Like a luxury car - maximum capability.

    Best for:

    * Critical decisions
    * Complex analysis
    * Strategic planning
    * Maximum accuracy
  </Card>
</CardGroup>

***

## Available Models in WonkaChat

<Tabs>
  <Tab title="By Provider">
    ### OpenAI Models

    <AccordionGroup>
      <Accordion title="GPT-5 Series (Latest)" icon="star">
        | Model   | Best For                    |
        | ------- | --------------------------- |
        | GPT-5.2 | Most advanced general tasks |
        | GPT-5   | High-quality responses      |

        <Tip>
          GPT-5 series offers cutting-edge performance for complex tasks.
        </Tip>
      </Accordion>

      <Accordion title="GPT-4o Series" icon="circle-check">
        | Model       | Best For                                  |
        | ----------- | ----------------------------------------- |
        | GPT-4o      | High-quality responses, complex reasoning |
        | GPT-4o Mini | General tasks, best value                 |

        <Check>
          GPT-4o Mini is recommended for most business tasks - excellent balance of quality and cost.
        </Check>
      </Accordion>
    </AccordionGroup>

    ***

    ### Anthropic Claude Models

    <AccordionGroup>
      <Accordion title="Claude Sonnet Series" icon="brain">
        | Model             | Best For                             |
        | ----------------- | ------------------------------------ |
        | Claude Sonnet 4.6 | Latest version, enhanced performance |
        | Claude Sonnet 4.5 | Balanced performance                 |
        | Claude Sonnet 4   | Complex reasoning and instructions   |

        <Tip>
          Claude Sonnet models excel at following complex instructions and nuanced tasks.
        </Tip>
      </Accordion>

      <Accordion title="Claude Haiku & Opus" icon="circle">
        | Model            | Best For                        |
        | ---------------- | ------------------------------- |
        | Claude Haiku 4.5 | Fast, efficient tasks           |
        | Claude Opus 4.5  | Maximum capability and accuracy |
      </Accordion>
    </AccordionGroup>

    ***

    ### Google Gemini Models

    <AccordionGroup>
      <Accordion title="Gemini 3 Series (Preview)" icon="sparkles">
        | Model                  | Best For                   |
        | ---------------------- | -------------------------- |
        | Gemini 3 Pro Preview   | Next-gen complex reasoning |
        | Gemini 3 Flash Preview | Next-gen fast tasks        |

        <Info>
          Gemini 3 models are preview versions with cutting-edge capabilities.
        </Info>
      </Accordion>

      <Accordion title="Gemini 2.5 Series" icon="google">
        | Model            | Best For                   |
        | ---------------- | -------------------------- |
        | Gemini 2.5 Pro   | Complex reasoning tasks    |
        | Gemini 2.5 Flash | Fast, cost-effective tasks |

        <Check>
          Gemini 2.5 Flash offers excellent performance at very competitive pricing.
        </Check>
      </Accordion>
    </AccordionGroup>

    ***

    ### Mistral AI Models

    <AccordionGroup>
      <Accordion title="Mistral Latest Series" icon="sparkles">
        | Model                 | Best For                                 |
        | --------------------- | ---------------------------------------- |
        | Mistral Large Latest  | Advanced tasks requiring high capability |
        | Mistral Medium Latest | Balanced performance for general use     |

        <Info>
          Mistral "Latest" models automatically update to the newest versions, ensuring you always have access to improvements.
        </Info>
      </Accordion>
    </AccordionGroup>
  </Tab>

  <Tab title="By Use Case">
    ### Simple Tasks (FAQ, categorization, summaries)

    Best options:

    * GPT-4o Mini - Best overall value
    * Claude Haiku 4.5 - Fast and efficient
    * Gemini 2.5 Flash - Very cost-effective

    ***

    ### General Business Tasks (emails, reports, research)

    Best options:

    * GPT-4o - High quality
    * Claude Sonnet 4.6 - Excellent instruction following
    * Gemini 2.5 Pro - Strong performance

    ***

    ### Complex Reasoning (analysis, strategy, coding)

    Best options:

    * GPT-5.2 - Most advanced
    * Claude Opus 4.5 - Maximum capability
    * Claude Sonnet 4.6 - Complex instructions
    * Gemini 3 Pro Preview - Next-gen reasoning

    ***

    ### High-Volume Operations (chatbots, automation)

    Best options:

    * GPT-4o Mini - Best value
    * Gemini 2.5 Flash - Very cost-effective
    * Claude Haiku 4.5 - Fast processing
  </Tab>
</Tabs>

***

## Which Model Should You Use?

<Steps>
  <Step title="Start with the question: How important is this task?">
    Low stakes, high volume:

    * Customer FAQs
    * Simple categorization
    * Basic summaries
    * Internal notes

    → Use Economy Models: GPT-4o Mini, Gemini 2.5 Flash, Claude Haiku 4.5

    Medium importance:

    * Customer-facing emails
    * Internal reports
    * Research tasks
    * General business tasks

    → Use Standard Models: GPT-4o, Claude Sonnet 4.6, Gemini 2.5 Pro

    High stakes, critical:

    * Client proposals
    * Strategic decisions
    * Complex analysis
    * Sensitive communications

    → Use Premium Models: GPT-5.2, Claude Opus 4.5, Gemini 3 Pro Preview
  </Step>

  <Step title="Consider your volume">
    How many requests per day?

    * High volume (1,000+ per day): Every cent matters → Use economy models
    * Medium volume (100-1,000 per day): Balance quality and cost → Use standard models
    * Low volume (less than 100 per day): Quality over cost → Use premium models where needed
  </Step>

  <Step title="Test and adjust">
    <Tip>
      Pro tip: Start with a mid-tier model (like GPT-4o Mini), evaluate the quality, then adjust:

      * Quality not good enough? → Try a better model
      * Quality is great? → Test a cheaper model to save money
    </Tip>
  </Step>
</Steps>

***

## Common Questions

<AccordionGroup>
  <Accordion title="How much will this actually cost me?" icon="dollar-sign">
    To estimate your costs:

    1. Track your actual usage for a few days
    2. Count tokens per typical request
    3. Calculate based on your chosen model's rate

    Your Sales Support team can help you estimate costs based on your specific use case and expected volume.

    [Contact Sales Support Team](/en/support-resources/getting-help)
  </Accordion>

  <Accordion title="Can I mix different models?" icon="shuffle">
    Yes! This is actually a smart strategy:

    * Customer-facing tasks: Use better models (they represent your brand)
    * Internal tasks: Use economy models (quality matters less)
    * Testing/development: Use the cheapest models

    You can configure different models for different AI agents.
  </Accordion>

  <Accordion title="What if I pick the wrong model?" icon="triangle-exclamation">
    No problem! You can switch models anytime:

    1. Test with a mid-tier model
    2. Evaluate quality for your use case
    3. Adjust up or down based on results

    There's no penalty for switching, and you only pay for what you use.
  </Accordion>

  <Accordion title="How do I know which model I'm using?" icon="circle-question">
    When creating or editing an AI agent in WonkaChat, you select the model in the agent settings. Each agent can use a different model.

    Our recommendation: Start with GPT-4o Mini for most business tasks, then adjust based on your quality needs.
  </Accordion>
</AccordionGroup>

***

## Recommended Starting Models

Not sure where to start? Here are our top recommendations:

<CardGroup>
  <Card title="Best Overall Value" icon="star">
    GPT-4o Mini

    Excellent quality for most business tasks. This is what we recommend to 80% of our customers.

    Use for: Emails, reports, general questions, customer support
  </Card>

  <Card title="Best for Speed" icon="gauge">
    Gemini 2.5 Flash

    Very fast and cost-effective. Great for high-volume, time-sensitive tasks.

    Use for: Quick responses, FAQs, simple classifications
  </Card>

  <Card title="Best for Complex Tasks" icon="brain">
    Claude Sonnet 4.6

    Exceptional at following complex instructions and nuanced reasoning.

    Use for: Analysis, strategy, sensitive communications
  </Card>
</CardGroup>

***

## Need More Details?

<CardGroup>
  <Card title="Detailed Model Pricing" icon="table" href="/en/pricing/tokens-pricing">
    See complete pricing tables for all available AI models.
  </Card>

  <Card title="Contact Sales" icon="phone" href="/en/support-resources/getting-help">
    Not sure which model is right? Our team can analyze your needs and recommend the best fit.
  </Card>
</CardGroup>


# Token Pricing
Source: https://help.wonka.chat/en/pricing/tokens-pricing

Understand AI model token costs and choose the right model for your use case

## What Are Tokens?

<Info>
  Tokens are the units that measure AI model usage. Both your input (prompt) and the AI's response (completion) consume tokens. Think of tokens as roughly 3/4 of a word in English.

  Example: "Hello, how are you today?" ≈ 6 tokens
</Info>

All token prices are listed in USD per 1 million tokens.

***

## Available Models in WonkaChat

<Info>
  The following models are currently available in WonkaChat. For specific pricing information, please contact our Sales Support team.
</Info>

<Tabs>
  <Tab title="By Provider">
    ### OpenAI Models

    <AccordionGroup>
      <Accordion title="GPT-5 Series (Latest)" icon="star">
        | Model       | Best For                                                   |
        | ----------- | ---------------------------------------------------------- |
        | **GPT-5.2** | Most advanced general-purpose AI, cutting-edge performance |
        | **GPT-5**   | High-quality responses, complex reasoning                  |

        <Check>
          GPT-5 series represents the latest in AI capabilities with enhanced reasoning and generation quality.
        </Check>
      </Accordion>

      <Accordion title="GPT-4o Series" icon="circle-check">
        | Model           | Best For                                     |
        | --------------- | -------------------------------------------- |
        | **GPT-4o Mini** | General tasks, best value for most use cases |
        | **GPT-4o**      | High-quality responses, complex reasoning    |

        <Tip>
          GPT-4o Mini is our most popular choice - excellent balance of quality and cost-effectiveness.
        </Tip>
      </Accordion>
    </AccordionGroup>

    ***

    ### Anthropic Claude Models

    <AccordionGroup>
      <Accordion title="Claude Sonnet Series" icon="brain">
        | Model                 | Best For                                                       |
        | --------------------- | -------------------------------------------------------------- |
        | **Claude Sonnet 4.6** | Latest version, enhanced performance and instruction following |
        | **Claude Sonnet 4.5** | Balanced performance for complex tasks                         |
        | **Claude Sonnet 4**   | Complex reasoning, detailed analysis                           |

        <Check>
          Claude Sonnet models are renowned for following complex, nuanced instructions accurately.
        </Check>
      </Accordion>

      <Accordion title="Claude Specialized Models" icon="sparkles">
        | Model                | Best For                                            |
        | -------------------- | --------------------------------------------------- |
        | **Claude Haiku 4.5** | Fast, efficient tasks with quick turnaround         |
        | **Claude Opus 4.5**  | Maximum capability, most complex and critical tasks |

        <Info>
          Claude Haiku is optimized for speed, while Opus provides the highest quality for critical work.
        </Info>
      </Accordion>
    </AccordionGroup>

    ***

    ### Google Gemini Models

    <AccordionGroup>
      <Accordion title="Gemini 3 Series (Preview)" icon="rocket">
        | Model                      | Best For                                    |
        | -------------------------- | ------------------------------------------- |
        | **Gemini 3 Pro Preview**   | Next-generation complex reasoning (preview) |
        | **Gemini 3 Flash Preview** | Next-generation fast processing (preview)   |

        <Warning>
          Gemini 3 models are currently in preview. Features and availability may change.
        </Warning>
      </Accordion>

      <Accordion title="Gemini 2.5 Series" icon="google">
        | Model                | Best For                            |
        | -------------------- | ----------------------------------- |
        | **Gemini 2.5 Pro**   | Complex reasoning, analytical tasks |
        | **Gemini 2.5 Flash** | Fast, cost-effective general tasks  |

        <Tip>
          Gemini 2.5 Flash offers excellent performance for high-volume operations at competitive pricing.
        </Tip>
      </Accordion>
    </AccordionGroup>

    ***

    ### Mistral AI Models

    | Model                     | Best For                                      |
    | ------------------------- | --------------------------------------------- |
    | **Mistral Large Latest**  | Advanced tasks requiring high capability      |
    | **Mistral Medium Latest** | Balanced performance for general business use |

    <Info>
      Mistral models are automatically updated to the latest versions, ensuring you always have access to improvements.
    </Info>
  </Tab>

  <Tab title="By Use Case">
    ## Choose by Task Type

    <Steps>
      <Step title="Simple & High-Volume Tasks">
        **Best for:** FAQs, categorization, basic summaries, simple classifications

        **Recommended Models:**

        * GPT-4o Mini - Best overall value
        * Gemini 2.5 Flash - Very cost-effective
        * Claude Haiku 4.5 - Fast processing

        **Use when:** You need quick responses for straightforward questions or tasks that don't require deep reasoning.
      </Step>

      <Step title="General Business Tasks">
        **Best for:** Emails, reports, research, content generation, customer service

        **Recommended Models:**

        * GPT-4o - High-quality, reliable
        * Claude Sonnet 4.6 - Excellent instruction following
        * Gemini 2.5 Pro - Strong analytical performance
        * Mistral Large Latest - Advanced general capabilities

        **Use when:** You need quality responses for typical business communications and analysis.
      </Step>

      <Step title="Complex & Critical Tasks">
        **Best for:** Strategic decisions, complex analysis, sensitive communications, detailed research

        **Recommended Models:**

        * GPT-5.2 - Most advanced capabilities
        * Claude Opus 4.5 - Maximum quality and accuracy
        * Gemini 3 Pro Preview - Next-gen reasoning
        * Claude Sonnet 4.6 - Complex instruction following

        **Use when:** Quality and accuracy are paramount, and the task requires sophisticated reasoning.
      </Step>

      <Step title="Specialized Requirements">
        **Speed Priority:**

        * Claude Haiku 4.5
        * Gemini 2.5 Flash
        * GPT-4o Mini

        **Instruction Following:**

        * Claude Sonnet 4.6
        * Claude Opus 4.5

        **Latest Technology:**

        * GPT-5.2
        * Gemini 3 Pro Preview
        * Claude Sonnet 4.6
      </Step>
    </Steps>
  </Tab>

  <Tab title="Model Comparison">
    ## Quick Comparison Guide

    ### Entry Level (Economy)

    **When to use:** High volume, simple tasks, internal tools

    | Model            | Strengths                               |
    | ---------------- | --------------------------------------- |
    | GPT-4o Mini      | Best all-around value, reliable quality |
    | Gemini 2.5 Flash | Very cost-effective, fast               |
    | Claude Haiku 4.5 | Quick turnaround, efficient             |

    ***

    ### Mid-Tier (Standard)

    **When to use:** General business tasks, customer-facing content

    | Model                | Strengths                       |
    | -------------------- | ------------------------------- |
    | GPT-4o               | High quality, complex reasoning |
    | Claude Sonnet 4.6    | Excellent instruction following |
    | Gemini 2.5 Pro       | Strong analytical capabilities  |
    | Mistral Large Latest | Advanced general performance    |

    ***

    ### Premium (Advanced)

    **When to use:** Critical decisions, complex analysis, maximum quality

    | Model                | Strengths                    |
    | -------------------- | ---------------------------- |
    | GPT-5.2              | Cutting-edge AI capabilities |
    | Claude Opus 4.5      | Maximum accuracy and quality |
    | Gemini 3 Pro Preview | Next-generation reasoning    |

    ***

    ### By Workload Type

    **High-Volume Operations:**
    → GPT-4o Mini, Gemini 2.5 Flash

    **Customer-Facing:**
    → GPT-4o, Claude Sonnet 4.6, Mistral Large Latest

    **Internal Analysis:**
    → Gemini 2.5 Pro, Claude Sonnet 4.5, GPT-4o

    **Critical Tasks:**
    → GPT-5.2, Claude Opus 4.5, Claude Sonnet 4.6
  </Tab>
</Tabs>

***

## Frequently Asked Questions

<AccordionGroup>
  <Accordion title="How are tokens counted?" icon="calculator">
    Tokens are counted for both input (your prompt + agent instructions) and output (the AI's response).

    Rough estimate: 1 token ≈ 0.75 English words

    Example conversation:

    * Your question: "Summarize this document" (3 tokens)
    * Document content: 2,000 words (≈2,666 tokens)
    * AI summary: 200 words (≈267 tokens)
    * Total: \~2,936 tokens consumed
  </Accordion>

  <Accordion title="Which model should I use?" icon="circle-question">
    Start here:

    * Most teams: GPT-4o Mini - best value
    * Speed-focused: Gemini 2.5 Flash or Claude Haiku 4.5
    * Complex tasks: Claude Sonnet 4.6 or GPT-4o
    * Maximum capability: GPT-5.2 or Claude Opus 4.5

    Test with your actual use case to find the sweet spot.
  </Accordion>

  <Accordion title="Can I switch models?" icon="arrows-rotate">
    Yes! You can configure different models for different agents. Use expensive models only where quality matters most.

    Strategy:

    * Customer-facing: Premium or standard models
    * Internal tools: Economy models
    * Testing: Economy models
  </Accordion>

  <Accordion title="How is pricing calculated?" icon="coins">
    Token pricing is based on:

    1. Input tokens: Your prompt + system instructions + context
    2. Output tokens: The AI's generated response

    Different models have different pricing for input vs output. For specific pricing details, please contact our Sales Support team.
  </Accordion>

  <Accordion title="What's the difference between model versions?" icon="code-branch">
    Newer versions (like Claude Sonnet 4.6 vs 4.5, or GPT-5.2 vs GPT-5) generally offer:

    * Improved reasoning capabilities
    * Better instruction following
    * Enhanced accuracy
    * Sometimes better pricing

    The "latest" variants (like Mistral Large Latest) automatically update to the newest version.
  </Accordion>

  <Accordion title="Are preview models stable for production?" icon="flask">
    Preview models (like Gemini 3 series) are:

    * Cutting-edge but may have changes
    * Best for testing new capabilities
    * Not recommended for critical production workloads

    For production use, stick with stable releases like GPT-4o, Claude Sonnet 4.6, or Gemini 2.5 series.
  </Accordion>
</AccordionGroup>

***

## Need Help Choosing?

<Card title="Contact Our Team" icon="comments">
  Not sure which models fit your use case? Our Sales Support team can:

  * Analyze your requirements
  * Recommend the optimal model mix
  * Provide detailed pricing for your expected volume
  * Help you test different options

  [Get Model Recommendations](/en/support-resources/getting-help)
</Card>


# Access Control
Source: https://help.wonka.chat/en/security-governance/access-control

Manage user roles, user access, agent permissions, and tool restrictions to control what AI agents can do in your organization.

Control who has access to what in your WonkaChat organization. Manage users, assign roles, define granular permissions for AI agents, and restrict which tools agents can use to ensure secure collaboration.

<Info>
  Access control implements the principle of least privilege that allows users and agents to receive only the permissions they need to do their work.
</Info>

***

## User Management

Add, manage, and remove users in your organization through the Settings panel.

<Steps>
  <Step title="Access Settings">
    Click your name in the bottom left corner of the WonkaChat dashboard, then select **User Management**. Only administrators can access user management functions.
  </Step>

  <Step title="Invite New Users">
    Send email invitations to users you want to add to your organization. Users receive an invitation link to create their account and join.
  </Step>

  <Step title="Assign Role">
    Choose whether the new user will be an **Admin** or a **User** when sending the invitation.

    <Check>
      You can change user roles at any time after they join the organization.
    </Check>
  </Step>

  <Step title="Managing Existing Users">
    <AccordionGroup>
      <Accordion title="Change Roles" icon="user-gear">
        Promote users to administrators or remove admin privileges as responsibilities change.

        <Warning>
          Administrators can manage their own admin rights, but cannot override or remove admin privileges from other administrators.
        </Warning>
      </Accordion>

      <Accordion title="Remove Users" icon="user-slash">
        Permanently remove users from your organization. While you can easily delete and re-add users, all data associated with the account is permanently deleted and cannot be recovered.

        <Warning>
          Removing a user deletes all their data irreversibly. You can re-add the user later, but their previous agents, configurations, and settings will be gone forever. Consider deactivation for temporary access removal.
        </Warning>
      </Accordion>
    </AccordionGroup>
  </Step>
</Steps>

***

## Role-Based Permissions

WonkaChat uses a simple two-role system for clear permission boundaries.

<iframe title="Governance and RBAC explained" />

### Available Roles

<CardGroup>
  <Card title="User" icon="user">
    **Personal agent creation**

    • Create their own personal AI agents\
    • Use agents shared by administrators\
    • Modify shared agents only if granted permission\
    • Access only the personal created agents or admin shared agents

    <Info>
      Users cannot share agents with others, only administrators can make agents available organization-wide.
    </Info>
  </Card>

  <Card title="Administrator" icon="user-shield">
    **Full organizational control**

    • Manage all users\
    • Create, modify, and share agents organization-wide\
    • Control agent permissions (who can view/modify)\
    • Promote users to administrator roles

    <Check>
      Admins have complete control except they cannot override other administrators' privileges.
    </Check>
  </Card>
</CardGroup>

<Tip>
  Keep the number of administrators limited to those who need full organizational control. Most team members should have User roles.
</Tip>

### Agent Visibility Scope

<Tabs>
  <Tab title="Personal Agents">
    Agents created by users are **private by default** and visible only to the creator.

    Users cannot share their personal agents organization-wide. Only administrators can share agents.

    <Tip>
      If a user creates a valuable agent and wants to share it, they should contact an administrator to review and share it.
    </Tip>
  </Tab>

  <Tab title="Shared Agents">
    Agents shared by administrators can be made available to:

    * **Specific users**: Select individuals who can access the agent
    * **Entire organization**: All users in the organization

    Admins control the exact permission level (Viewer, Editor, or Owner) for each shared agent.

    <Check>
      Shared agents appear in users' agent lists based on the permissions they've been granted.
    </Check>
  </Tab>
</Tabs>

***

## Sharing AI Agents

Administrators can share AI Agents directly from the Agent Builder panel.

<Tabs>
  <Tab title="I'm an Administrator">
    As an administrator, you control agent sharing from the Agent Builder panel.

    ### How to Share an Agent

    <Steps>
      <Step title="Open Agent Builder">
        Navigate to the agent you want to share (or create a new one) in the Agent Builder.
      </Step>

      <Step title="Configure Sharing Settings">
        In the Agent Builder panel, you'll find sharing options:

        <Frame>
          <img alt="Agent Sharing showing sharing options" />
        </Frame>

        **Visibility:**

        * Share with the entire organization, or
        * Share with specific users

        **Permission Levels:**

        * **Viewer**: Can view and use the agent (no modifications)
        * **Editor**: Can view, use, and modify the agent
        * **Owner**: Has full control including sharing and deletion

        <Tip>
          For important organizational agents, use Viewer permissions for most users and grant Editor access only to trusted collaborators.
        </Tip>
      </Step>

      <Step title="Review Tool Access">
        Before sharing, verify which tools and MCP connections the agent has access to. Ensure they're appropriate for all users who will have access.

        <Warning>
          When users interact with a shared agent, the agent inherits that user's permissions in connected tools. Consider what data different users can access.
        </Warning>
      </Step>

      <Step title="Save and Share">
        Save your sharing configuration. The agent immediately becomes available to the specified users based on the permissions you set.

        <Check>
          Users will see the shared agent in their agent list and can start using it according to their granted permissions.
        </Check>
      </Step>
    </Steps>
  </Tab>

  <Tab title="I'm a User">
    As a standard user, you can create personal agents but need administrator involvement to share them organization-wide.\
    If you create an agent that would benefit others, contact an administrator to request organization-wide sharing.

    Provide:

    * Clear description of what the agent does
    * Intended use cases
    * Why it would benefit the organization
    * Which tools it requires

    <Tip>
      Administrators need to review agents for security, tool access, and compliance before sharing them widely.
    </Tip>

    ### Using Shared Agents

    When an administrator shares an agent, they can assign one of three permission levels:

    <AccordionGroup>
      <Accordion title="Viewer" icon="eye">
        **Can view and use the agent but cannot modify it**

        Viewers can:

        * Use the agent in conversations
        * See the agent in their agent list
        * Access the agent's capabilities

        Viewers cannot:

        * Modify the agent's configuration, prompts, or tools
        * Share the agent with others
        * Change agent settings

        <Check>
          Viewer permission is ideal for most users who need to use the agent without making changes.
        </Check>
      </Accordion>

      <Accordion title="Editor" icon="pen-to-square">
        **Can view and modify the agent**

        Editors can:

        * Use the agent in conversations
        * View the agent's configuration and settings
        * Modify the agent's prompts, tools, and configuration
        * Make changes that affect all users

        Editors cannot:

        * Share the agent with others
        * Change ownership or delete the agent

        <Warning>
          Granting editor permission allows users to change agent behavior for everyone. Use this carefully for collaborative agent development.
        </Warning>
      </Accordion>

      <Accordion title="Owner" icon="crown">
        **Has full control over the agent including sharing it**

        Owners can:

        * Use the agent in conversations
        * View and modify all agent settings
        * Share the agent with others (assign Viewer/Editor/Owner permissions)
        * Change or transfer ownership
        * Delete the agent

        <Info>
          Only administrators can be assigned as owners when sharing agents organization-wide. The agent creator is automatically the original owner.
        </Info>
      </Accordion>
    </AccordionGroup>
  </Tab>
</Tabs>

***

## Tool Restriction

When creating an AI agent, you can control exactly which tools it has access to. This ensures agents only have the capabilities they need to perform their intended function.

### How Tool Restriction Works

When you add an MCP connection (like Outlook, GitHub, or Slack) to an agent, you can selectively enable or disable individual tools within that connection.

**For example**, if you add the Outlook MCP connection to an agent:

* You can **keep** tools like "draft email" and "read emails"
* You can **remove** tools like "send email" and "delete email"

Removing the "send email" tool means the agent cannot send emails, even if instructed to, because the capability simply doesn't exist for that agent.

<Check>
  Tool restriction happens at the agent creation level. Each agent can have different tool access based on its intended purpose.
</Check>

### Why Restrict Tools?

<AccordionGroup>
  <Accordion title="Prevent Unintended Actions">
    By removing tools that perform irreversible actions (sending emails, deleting files, deploying code), you eliminate the risk of agents making mistakes or misinterpreting prompts.

    **Security through absence**: If an agent doesn't have the "send email" tool, it's impossible for it to send an email (even by accident).
  </Accordion>

  <Accordion title="Minimize Risk Surface">
    The fewer tools an agent has, the less can go wrong. Reduce the attack surface by giving agents only the specific capabilities they need.

    **Example**: A data analysis agent doesn't need tools to modify or delete data. He should only receive read access.
  </Accordion>

  <Accordion title="Purpose-Built Agents">
    Create specialized agents optimized for specific tasks by giving them precise tool sets.

    **Example**: A "Draft Email" agent with Outlook MCP but only drafting tools enabled. He should not receive any sending, deleting, or forwarding capabilities.
  </Accordion>
</AccordionGroup>

### Example: Email Draft Agent

The difference between a helpful drafting assistant and a potential security risk often comes down to a single tool.

<Tip>
  When creating agents, disable tools that execute final actions unless the agent specifically needs them. Enable tools for preparation and analysis, but restrict tools that send, delete, deploy, or modify production systems.
</Tip>

<Tabs>
  <Tab title="✅ Good Example: Send Tool Disabled">
    The same agent, but with the `send-email` tool restricted.

    **The Result:**

    ```
    User prompt: "Draft an email to john@example.com about the Q4 report."

    Agent response: "Here's a draft email for john@example.com about the Q4 report:

    Subject: Q4 Report Update

    Hi John,
    [draft content]

    Would you like me to revise anything?"
    ```

    Without access to the `send-email` tool, the agent can only draft content. The user reviews it, makes changes if needed, and manually sends when ready.

    <Check>
      **Safe**: User maintains full control over when and what gets sent. Agent provides value without risk.
    </Check>
  </Tab>

  <Tab title="❌ Bad Example: Send Tool Enabled">
    An agent configured to help draft emails but with the `send-email` tool still enabled.

    **The Problem:**

    ```
    User prompt: "Draft an email to john@example.com about the Q4 report."

    Agent response: "I've drafted and sent the following email to john@example.com..."
    ```

    Because the agent had access to the `send-email` tool, it interpreted "draft an email" as permission to also send it. The email went out before the user could review it.

    <Warning>
      **Risk**: Unreviewed emails being sent, potential data leaks, unprofessional communication, or sending to wrong recipients.
    </Warning>
  </Tab>
</Tabs>

***

## Safe Mode: Human-in-the-Loop

Safe Mode ensures you approve every action an AI Agent wants to take before it executes.

<Info>
  **Visual Guide Coming Soon**: We're creating a step-by-step visual guide with screenshots showing exactly how Safe Mode works, including the approval dialog and decision workflow.
</Info>

### How Safe Mode Works

<Steps>
  <Step title="Identify When to Use Safe Mode">
    Enable Safe Mode for agents that perform high-risk operations:

    * Send communications (emails, messages, posts)
    * Modify or delete production data
    * Deploy code or configurations
    * Execute financial transactions

    <Warning>
      Safe Mode provides essential oversight to prevent mistakes and unauthorized actions in critical operations.
    </Warning>
  </Step>

  <Step title="Agent Requests Action">
    When an AI agent wants to use a tool or execute an action, it pauses and requests your approval.

    You see exactly what the agent wants to do, which tool it wants to use, and what parameters it will pass.
  </Step>

  <Step title="You Review & Decide">
    Review the requested action and decide whether to:

    * **Allow**: The action executes as requested
    * **Deny**: The action is blocked, and the agent continues without executing it

    <Tip>
      This is your opportunity to catch potential mistakes, verify actions are appropriate, and maintain oversight.
    </Tip>
  </Step>

  <Step title="Action Executes or Blocks">
    * If approved, the action executes and the agent receives the results
    * If denied, the agent is informed the action was not permitted and can adjust its approach

    <Check>
      You can disable Safe Mode for trusted agents performing routine tasks, but it's recommended for agents with access to powerful tools.
    </Check>
  </Step>
</Steps>

***

## Access Control Advices

<CardGroup>
  <Card title="Limit Administrators" icon="user-shield">
    Grant admin roles only to those who need full organizational control. Fewer admins = easier auditing and reduced risk.
  </Card>

  <Card title="Start Restrictive" icon="user-lock">
    Begin with minimal tool access and permissions, then add only as needed. It's easier to grant access than to recover from mistakes.
  </Card>

  <Card title="Review Regularly" icon="calendar-check">
    Periodically audit who has access to which agents and tools, adjusting as roles and responsibilities change.
  </Card>

  <Card title="Test with Safe Mode" icon="flask">
    Always test new agents with Safe Mode enabled to observe what actions they attempt before granting autonomous operation.
  </Card>
</CardGroup>

<Tip>
  Clear communication about agent purposes, permissions, and tool restrictions helps users understand what they can do and reduces support requests.
</Tip>


# Best Practices
Source: https://help.wonka.chat/en/security-governance/best-practices

Practical guidelines for using AI securely and effectively in your daily work with WonkaChat.

Use these practical guidelines to make the most of AI while maintaining professional standards and avoiding irreversible mistakes.

<Info>
  **WonkaChat is secure by design**: your data is encrypted, isolated between organizations, and never used for AI training. These guidelines help you work effectively and professionally.
</Info>

***

## Using AI Effectively at Work

Get the most value from AI while maintaining quality and control.

### Scope Your Requests

<Tabs>
  <Tab title="Good AI Tasks" icon="circle-check">
    **Clear, specific objectives:**

    * Analyze this dataset and identify trends
    * Draft a response to this customer inquiry
    * Review this document for clarity and errors
    * Generate test cases for this feature
    * Summarize key points from this meeting

    **Why it works**: Specific tasks get better results and are easier to review

    <Check>
      These focused requests help AI understand exactly what you need and make outputs easier to verify.
    </Check>
  </Tab>

  <Tab title="Tasks Needing Caution" icon="triangle-exclamation">
    **Open-ended or high-impact:**

    * Make widespread changes across systems
    * Decisions affecting production environments
    * Irreversible operations (deletions, deployments)
    * Communications representing your organization

    **Why caution**: Review outputs before execution to avoid unintended consequences

    <Warning>
      For these tasks, use [Safe Mode](/security-governance/access-control#safe-mode-human-in-the-loop), test in a staging environment first, or break them into smaller, reviewable steps.
    </Warning>
  </Tab>
</Tabs>

***

## Think Before Executing

The main risk with AI isn't security, it's accidentally doing something you can't undo.

### Understanding Irreversible Actions

<AccordionGroup>
  <Accordion title="What Can't Be Undone" icon="ban">
    **Actions that permanently change or remove data:**

    * Deleting databases, production files, or user data
    * Deploying code to production systems
    * Sending emails or messages to customers
    * Modifying live configurations
    * Removing user accounts or permissions

    <Warning>
      AI can execute these actions quickly. Use [Safe Mode](/security-governance/access-control#safe-mode-human-in-the-loop) or review carefully before any operation that modifies or removes important data.
    </Warning>
  </Accordion>

  <Accordion title="Safe Activities" icon="circle-check">
    **Actions that can be reviewed or reversed:**

    * Drafting content (emails, documents, code)
    * Analyzing data and generating insights
    * Searching and retrieving information
    * Creating test data or examples
    * Generating documentation

    <Check>
      These activities give you time to review and refine before anything goes live or impacts others.
    </Check>
  </Accordion>
</AccordionGroup>

### Strategies for Critical Operations

When AI agents work with high-impact systems, implement safeguards to maintain control:

<AccordionGroup>
  <Accordion title="Enable Safe Mode" icon="shield-check">
    Require approval before actions execute. [Safe Mode](/security-governance/access-control#safe-mode-human-in-the-loop) pauses the agent and shows you exactly what it wants to do before proceeding.

    **Best for**: Production systems, customer communications, financial operations, deployment changes.
  </Accordion>

  <Accordion title="Test in Staging First" icon="flask">
    Run agents in non-production environments before deploying to live systems. Verify behavior with test data before working with real information. Use synthetic or anonymized test data whenever possible.

    **Best for**: New agents, complex workflows, or operations affecting multiple systems.
  </Accordion>

  <Accordion title="Break Into Steps" icon="list-check">
    Divide large operations into smaller, reviewable chunks. Execute one step, verify the result, then proceed to the next.

    **Best for**: Multi-step processes, data migrations, or operations spanning multiple systems.
  </Accordion>

  <Accordion title="Restrict Tool Access" icon="lock">
    Remove tools that perform final actions (send, delete, deploy) and keep only preparation tools (draft, analyze, read). This reduces the chance of unintended actions and keeps agents focused on their purpose. Learn more about [tool restriction](/security-governance/access-control#tool-restriction).

    **Best for**: Agents that assist with decisions but shouldn't execute them autonomously.
  </Accordion>

  <Accordion title="Review Agent Access" icon="eye">
    Before connecting agents to your tools, review what data they can access through MCP connections.

    **Best for**: Ensuring least-privilege access and periodic access audits.
  </Accordion>

  <Accordion title="Environment Variables for Keys" icon="key">
    When sharing code with AI or committing examples, use environment variables instead of hardcoded API keys or credentials.

    **Best for**: Code shared in conversations, samples, and collaborative repositories.
  </Accordion>
</AccordionGroup>

<Tip>
  Combine multiple strategies for the highest-risk operations. For example, use restricted tools AND Safe Mode for agents handling customer communications.
</Tip>

***

## Common Sense Guidelines

Basic principles for professional AI usage:

<AccordionGroup>
  <Accordion title="Review Before Impact">
    Always review AI outputs before they affect others or production systems. Treat AI-generated content like work from a colleague, you need to verify it before it goes out.
  </Accordion>

  <Accordion title="Use Version Control">
    For code changes, use Git or your version control system. Commit AI changes separately so you can track and revert if needed. This is good practice for any code changes, not just AI-generated.
  </Accordion>

  <Accordion title="Don't Request Illegal or Harmful Actions">
    Don't ask AI to help with illegal activities, create harmful content, or generate misinformation. This is about professional ethics and your organization's reputation, not platform security.
  </Accordion>

  <Accordion title="Verify Important Information">
    For critical decisions, verify AI-generated information from authoritative sources. AI is excellent for research and drafting but should augment your expertise, not replace it.
  </Accordion>
</AccordionGroup>

<Info>
  These guidelines aren't unique to AI, they're professional standards for any work tool. Use good judgment and review important outputs.
</Info>

***

## Quick Reference: Risk Assessment

When planning to use AI for a task, consider these factors:

| Factor                 | Lower Risk ✅                     | Higher Risk ⚠️                                         |
| ---------------------- | -------------------------------- | ------------------------------------------------------ |
| **Reversibility**      | Drafts, analysis, suggestions    | Production changes, deletions, sending communications  |
| **Scope of Impact**    | Personal work, test environments | Production systems, customer-facing, organization-wide |
| **Review Opportunity** | Results reviewed before use      | Direct execution without human review                  |
| **Data Permanence**    | Temporary data, easy to recreate | Critical data, difficult to recover                    |

<Check>
  For higher-risk tasks, use [Safe Mode](/security-governance/access-control#safe-mode-human-in-the-loop), [restrict agent tools](/security-governance/access-control#tool-restriction), or break the work into reviewable steps.
</Check>

***

## Smart Agent Configuration

Configure agents appropriately for their intended purpose.

<CardGroup>
  <Card title="Purpose-Built Agents" icon="robot">
    Create agents optimized for specific tasks:

    * **Research agents**: Read and analyze only
    * **Draft agents**: Create content without sending
    * **Review agents**: Provide feedback without making changes

    Restrict tools to match the agent's purpose using [tool restriction](/security-governance/access-control#tool-restriction).
  </Card>

  <Card title="Test Before Trusting" icon="flask">
    Before giving an agent autonomy:

    * Test with Safe Mode enabled
    * Review what tools it has access to
    * Verify it behaves as expected
    * Start with limited scope, expand if needed

    Testing builds confidence and catches configuration issues.
  </Card>
</CardGroup>

***

## The Bottom Line

**WonkaChat provides enterprise security**: encryption, isolation, and secure authentication protect your data, but you still need to focus on some key practices:

1. **Think before executing** - Review before deleting, deploying, or sending
2. **Scope appropriately** - Give AI Agents specific tasks with clear boundaries
3. **Use good judgment** - Verify important information and maintain quality

These guidelines help you work effectively while avoiding mistakes that could impact your work or organization.

<Tip>
  **Remember**: WonkaChat is secure, but your actions still have consequences. Use the same good judgment you'd apply to any professional tool.
</Tip>


# Compliance & Governance
Source: https://help.wonka.chat/en/security-governance/compliance-and-governance

Understand WonkaChat's compliance features, audit capabilities, and certification roadmap.

WonkaChat implements robust compliance and governance features to support your regulatory requirements. While formal certifications are in progress, the underlying controls and capabilities are already built into the platform you're using today.

<Warning>
  **Certification Status**: WonkaChat has implemented security and governance controls aligned with major compliance frameworks (GDPR, SOC 2, ISO 27001). Formal third-party certifications are currently in progress.
</Warning>

***

## What you can rely on today

<CardGroup>
  <Card title="Audit trails" icon="file-lines">
    Complete logging of user and agent actions with immutable audit trails.
  </Card>

  <Card title="Access controls" icon="user-lock">
    RBAC, authentication, and permission management across the platform.
  </Card>

  <Card title="Encryption & isolation" icon="shield-check">
    Encryption in transit and at rest with organization-level data isolation.
  </Card>

  <Card title="Retention controls" icon="clock">
    Configurable retention policies for conversations, logs, and user data.
  </Card>
</CardGroup>

<Info>
  These controls are operational today. Certifications will formally validate what is already in place.
</Info>

***

## Certification status

<Update label="In progress" description="Formal audits underway">
  * SOC 2 Type II: third-party audit in progress
  * ISO 27001: certification process initiated
  * GDPR: features supporting data subject rights implemented; conduct DPIA as appropriate
</Update>

<Warning>
  WonkaChat does not yet hold formal certifications. We are actively working toward them.
</Warning>

<Check>
  You can use WonkaChat confidently today; certifications will formally recognize the existing controls.
</Check>

***

## Audit trails and logging

* **Comprehensive logging system**: Winston-based structured logging
* **Log levels**: TRACE (10), DEBUG (20), INFO (30), WARN (40), ERROR (50), FATAL (60)
* **Security event logging**: Auth attempts, permission changes, rate limits, suspicious activity
* **Migration audit trail**: Schema changes, upgrades, config modifications, version tracking
* **User activity logging**: Sessions, agent and tool usage, data access/exports

<Info>
  Sensitive data (passwords, API keys, tokens) is automatically redacted from logs.
</Info>

***

## Data retention policies

### Conversation data

<AccordionGroup>
  <Accordion title="Temporary chat retention">
    Enable automatic deletion of conversation history after a configured period.

    * Temporary chat: enabled
    * Retention window: 24 hours
    * Features: automatic cleanup jobs, user-controlled deletion, admin override, soft delete with recovery
  </Accordion>

  <Accordion title="Permanent conversations">
    Retain conversations indefinitely until manually deleted. Best for knowledge bases and long-term projects.
  </Accordion>
</AccordionGroup>

### Log retention

Recommended retention:

* Error logs: 90 days (debugging and troubleshooting)
* Combined logs: 30 days (general activity)
* Security logs: 365 days (compliance requirements)
* Audit logs: 7 years (legal and regulatory requirements)
* Violation logs: 180 days (violation tracking)

Rotation settings:

* Daily rotation with date-based filenames
* Max file size: 50 MB
* Keep last 30 days of files
* Compress older files

### User data

Retention rules:

* Active users: retained while account is active
* Inactive users: 90 days after last activity
* Deleted users: 30 days (soft delete with recovery)
* Conversations: user-controlled retention
* Files: 90 days after last access
* Sessions: 7 days
* Tokens: expire on refresh (typically 15 minutes to 7 days)

***

## Need more information?

<Card title="Contact us" icon="handshake" href="/en/support-resources/getting-help">
  For questions about compliance, security documentation, or vendor assessments, reach out to our team for assistance.
</Card>


# Overview
Source: https://help.wonka.chat/en/security-governance/index

Understand how WonkaChat protects your data and learn how to enhance security in your organization.

Security is a shared responsibility. WonkaChat provides robust built-in security measures, while you control access and implement best practices to ensure safe AI collaboration in your organization.

<Info>
  This section covers both **what we do** to protect your data and **what you can do** to enhance security when using WonkaChat.
</Info>

***

## What We Do: Built-In Security

Understand the security measures and protections that WonkaChat provides out of the box.

<Card title="Built-In Security" icon="shield-check" href="/en/security-governance/security-overview">
  Explore WonkaChat's security architecture, data privacy measures, and how MCP connections respect your existing tool permissions
</Card>

<Tip>
  WonkaChat maintains strict security boundaries where AI agents cannot override or escalate permissions in your connected tools.
</Tip>

## What You Can Do: Enhance Your Security

Take control of your security posture with access management and proven best practices.

<CardGroup>
  <Card title="Access Control & Tool Restriction" icon="user-shield" href="/en/security-governance/access-control">
    Manage users, control agent permissions, and restrict which tools agents can use with Safe Mode and granular access controls
  </Card>

  <Card title="Best Practices" icon="clipboard-check" href="/en/security-governance/best-practices">
    Follow recommended guidelines for safe AI usage, including scope management and avoiding excessive permissions
  </Card>
</CardGroup>

<Note>
  Combining WonkaChat's built-in security with proper access controls and best practices creates a comprehensive security framework for your organization.
</Note>


# Security Overview
Source: https://help.wonka.chat/en/security-governance/security-overview

Understand WonkaChat's built-in security architecture, governance controls, and data privacy measures.

WonkaChat is an enterprise AI chat platform built on top of [LibreChat](https://www.librechat.ai/), a robust open-source project with a thriving community. We've enhanced it with enterprise-grade security, governance controls, and compliance features designed for organizations that need to use AI securely with their data.

<Info>
  This page covers the built-in security measures and governance controls that WonkaChat provides out of the box.
</Info>

***

## Governance

Control what AI models, tools, and agents are available in your organization. WonkaChat comes with extensive capabilities out of the box, but you can customize your deployment to include only what's relevant for your company. This helps reducing complexity, minimizing unnecessary access, and maintaining a focused, secure environment.

### Organization Access

<CardGroup>
  <Card title="MCP Selection" icon="plug">
    Wonka configures which MCP servers are available to your organization, ensuring only approved tools are accessible.
  </Card>

  <Card title="AI Model Selection" icon="brain">
    Administrators control which AI models and providers (OpenAI, Anthropic, Google, etc.) users can access within the organization.
  </Card>
</CardGroup>

<Tip>
  These governance controls ensure your organization uses only approved AI resources that align with your security policies and compliance requirements.
</Tip>

Beyond WonkaChat's built-in security measures, your organization maintains direct control over user access. For example, admins manage which agents are available to users and grant granular permissions per agent. See [Access Control](/security-governance/access-control) for details.

***

## Security Architecture

WonkaChat implements enterprise-grade security with multiple layers of protection.

### Authentication System

WonkaChat uses a secure, token-based authentication system to protect user access:

<AccordionGroup>
  <Accordion title="Access Tokens (15 minutes)" icon="key">
    Short-lived access tokens allow users to authenticate and interact with the platform. These tokens expire after 15 minutes, minimizing the risk of unauthorized access if compromised.
  </Accordion>

  <Accordion title="Refresh Tokens (7 days)" icon="rotate">
    Refresh tokens maintain user sessions for up to 7 days without requiring re-authentication. After 7 days, users need to log in again.
  </Accordion>

  <Accordion title="JWT Technology" icon="shield-check">
    WonkaChat uses JSON Web Tokens (JWT) for secure, stateless authentication. JWTs are cryptographically signed to prevent tampering and verify authenticity.
  </Accordion>
</AccordionGroup>

<Check>
  This dual-token approach balances security (short-lived access) with user convenience (session persistence).
</Check>

### Authentication Providers

Multiple authentication methods are available to integrate with your existing identity systems:

<Tabs>
  <Tab title="Available Now">
    WonkaChat currently supports these authentication providers:

    * **Local Authentication**: Username and password authentication
    * **OpenID Connect**: Integration with OpenID-compliant identity providers
    * **Azure AD / Entra ID**: Direct integration with Microsoft Azure Active Directory

    <Tip>
      Choose the authentication method that aligns with your organization's identity management strategy.
    </Tip>
  </Tab>

  <Tab title="Coming Soon">
    Additional social login providers are planned for future releases:

    * Google authentication
    * GitHub authentication
    * Other popular social providers

    <Info>
      These providers are not currently available but are on the roadmap.
    </Info>
  </Tab>
</Tabs>

### Rate Limiting & Abuse Prevention

WonkaChat implements rate limiting to protect against brute force attacks and system abuse. Rate limits are applied across critical operations including login attempts, registration, password resets, administrative actions, and general API usage.

<Check>
  Rate limits protect your organization from automated attacks while allowing normal user activity.
</Check>

***

## Data Privacy

Your data remains secure and private throughout all interactions with WonkaChat.

### Privacy Guarantees

<CardGroup>
  <Card title="Multi-Tenant Isolation" icon="building-lock">
    Complete data isolation between organizations through subdomain-based routing (https\://.wonka.chat). No accidental data exposure across tenants.
  </Card>

  <Card title="No Model Training" icon="ban">
    Your conversations and data are never used to train AI models. Model providers process requests temporarily without retention. Your data stays in WonkaChat where you control it.
  </Card>

  <Card title="Encrypted in Transit" icon="lock">
    All data transmitted to and from WonkaChat is encrypted using HTTPS/TLS protocols for secure API communications.
  </Card>

  <Card title="Encrypted at Rest" icon="shield-halved">
    Data stored in WonkaChat's MongoDB databases is protected with encryption-at-rest support to secure your information.
  </Card>
</CardGroup>

### Data Encryption

<AccordionGroup>
  <Accordion title="In Transit Protection">
    **HTTPS/TLS for all communications:**

    * All API calls use secure HTTPS/TLS protocols
    * Data is encrypted during transmission between your browser and WonkaChat servers
    * Secure connections to AI model providers (Azure, AWS Bedrock)
  </Accordion>

  <Accordion title="At Rest Protection">
    **MongoDB encryption and secure storage:**

    * Passwords: Hashed with bcrypt using salt rounds (never stored in plain text)
    * Tokens: JWT signing with strong cryptographic secrets
    * Database: MongoDB encryption-at-rest support for all stored data
    * API Keys: Never logged in plain text
  </Accordion>

  <Accordion title="Automatic Redaction">
    **Sensitive data protection in logs:**

    * All logs automatically redact sensitive patterns (passwords, tokens, keys)
    * API keys and authentication tokens are never logged in plain text
    * Token rotation prevents long-term exposure
    * Passwords are hashed with bcrypt before any storage

    <Check>
      Even system administrators cannot access your passwords or sensitive credentials in logs or databases.
    </Check>
  </Accordion>
</AccordionGroup>

### Multi-Tenant Data Isolation

WonkaChat ensures complete separation between organizations:

<Tabs>
  <Tab title="Organization-Scoped Data">
    Every piece of data in WonkaChat is scoped to your organization:

    * Conversations and chat history
    * AI agents and configurations
    * User accounts and permissions
    * MCP connections and tool settings
    * Audit logs and activity records

    <Check>
      Data from one organization is completely isolated and inaccessible to other organizations.
    </Check>
  </Tab>

  <Tab title="Subdomain Routing">
    Each organization has its own unique subdomain:

    * **Format**: `https://{your-org}.wonka.chat`
    * **Benefit**: Built-in namespace isolation at the network level
    * **Security**: Prevents cross-organization data leakage

    <Info>
      Subdomain-based routing adds an extra layer of separation beyond database-level isolation.
    </Info>
  </Tab>
</Tabs>

### Model Provider Security

<AccordionGroup>
  <Accordion title="Azure & AWS Bedrock APIs">
    WonkaChat connects to AI models through secure Azure and AWS Bedrock APIs. Your data is processed in real-time and never stored or retained by model providers.
  </Accordion>

  <Accordion title="Ephemeral Model Processing">
    When you interact with an AI model, your prompts and responses are sent to the model provider, processed in memory, and immediately discarded by the provider after generating the response.

    **Your conversations are securely stored in WonkaChat** for you to access your history, but model providers (Azure, AWS) never retain, store, or use your data for training or any other purpose.
  </Accordion>
</AccordionGroup>

<Check>
  Your conversation history is securely stored in WonkaChat where you control it. Model providers only process requests temporarily without any data retention or reuse.
</Check>

***

## MCP Security Controls

WonkaChat supports MCP (Model Context Protocol) servers with granular security configuration and multi-user isolation.

### Multi-User Isolation

<AccordionGroup>
  <Accordion title="User-Specific Token Storage">
    Each user's MCP authentication tokens are stored separately and securely:

    * Tokens are isolated per user account
    * No cross-user token access or sharing
    * Automatic token cleanup on user deactivation

    <Check>
      Your MCP credentials are never accessible to other users, even within the same organization.
    </Check>
  </Accordion>

  <Accordion title="Session-Based Authentication">
    MCP connections use session-based authentication for security:

    * Automatic user identification via request headers
    * Session validation on every MCP operation
    * Timeout enforcement to prevent stale sessions
  </Accordion>

  <Accordion title="Authentication Strategies">
    WonkaChat supports multiple MCP authentication methods:

    * **in-tool**: User authenticates directly within the connected tool
    * **oauth**: OAuth 2.0 flow for secure, delegated access
    * **api-key**: API key-based authentication for service integrations

    <Info>
      The authentication strategy is determined by the MCP server provider and cannot be changed by your organization.
    </Info>
  </Accordion>
</AccordionGroup>

### No Permission Escalation

<CardGroup>
  <Card title="Inherit Your Permissions" icon="user-lock">
    MCP connections inherit your exact permissions when accessing connected tools. They cannot access anything you couldn't access manually.
  </Card>

  <Card title="No Admin Bypass" icon="shield-halved">
    MCP connections cannot be used to override, bypass, or escalate permissions in connected systems. Your security policies remain fully enforced.
  </Card>
</CardGroup>

<Warning>
  If you cannot access a resource or perform an action manually in a connected tool, the AI models cannot do it either (even if instructed to try).
</Warning>

***

## What You Can Control

While WonkaChat provides robust built-in security, you control additional layers:

<CardGroup>
  <Card title="Access Control" icon="user-shield" href="/en/security-governance/access-control">
    Manage user roles and control who can access specific agents
  </Card>

  <Card title="Best Practices" icon="clipboard-check" href="/en/security-governance/best-practices">
    Follow recommended guidelines for safe AI usage in your organization
  </Card>
</CardGroup>

<Check>
  Combining WonkaChat's built-in security with your organization's access controls and practices creates a comprehensive security framework.
</Check>


# FAQ
Source: https://help.wonka.chat/en/support-resources/faq

Frequently Asked Questions about WonkaChat.

Find quick answers about WonkaChat across setup, models, external connections (MCP), agents, security, troubleshooting, and best practices.

<Tip>
  New to WonkaChat? Start with our [Introduction](/welcome/introduction), then explore [Tools & Connection](/tools-connection/introduction) and [AI Agents](/ai-agents/introduction).
</Tip>

## Getting started

<AccordionGroup>
  <Accordion title="What is WonkaChat?" icon="circle-question">
    WonkaChat is an enterprise AI assistant that connects to your business tools and data to execute multi-step workflows through natural conversation, while keeping your data within your security perimeter. See [What is WonkaChat?](/welcome/introduction).

    <Check>
      Think of it as one interface that orchestrates actions across your apps (email, calendar, CRM, docs, ERP, and more).
    </Check>
  </Accordion>

  <Accordion title="How do I get started in 5 minutes?" icon="rocket">
    1. Create your account at [wonka.chat/register](https://wonka.chat/register) or sign in via your organization's SSO

    2. Send your first message to verify everything works

    3. Connect your first connection (email is a great start) via the Connectors settings

    4. Try a simple workflow like "Send an email to my team about tomorrow's meeting"

    See [Quick Start Guide](/welcome/quick-start).

    <Tip>
      Start with your email provider (Outlook Mail or Gmail). It unlocks high-value daily workflows.
    </Tip>
  </Accordion>

  <Accordion title="Do I need to code to use WonkaChat?" icon="seedling">
    No. You interact through conversation. Advanced users can configure agents and connections without writing code.
  </Accordion>

  <Accordion title="What can WonkaChat help me accomplish?" icon="wand-magic-sparkles">
    Common examples include onboarding workflows, intelligent search across softwares, automated reporting, meeting scheduling, and customer support triage. Explore examples in [Welcome → Introduction](/welcome/introduction).
  </Accordion>

  <Accordion title="Can I switch AI models?" icon="sliders">
    Yes. Choose pro models for complex reasoning or fast models for quick tasks. You can switch per task or conversation. See [Why Choose WonkaChat?](/welcome/why-wonka-chat).
  </Accordion>
</AccordionGroup>

## Models and usage

<AccordionGroup>
  <Accordion title="Which AI models are supported?" icon="microchip">
    WonkaChat is model-agnostic. You can use providers like OpenAI (GPT-5.2, GPT-5, GPT-4o), Anthropic Claude (Opus 4.5, Sonnet 4.5, Haiku 4.5), Google Gemini (2.5 Pro, 3 Flash), and Mistral. Choose per task based on complexity, speed, and cost. See [Why Choose WonkaChat?](/welcome/why-wonka-chat).
  </Accordion>

  <Accordion title="When should I use a pro model vs a fast model?" icon="brain">
    * **Pro models**: Deep analysis, multi-step planning, complex workflows, external connections
    * **Fast models**: Quick answers, formatting, short lookups, high-volume operations

    Details and examples in [Why Choose WonkaChat?](/welcome/why-wonka-chat) and [Creating Your First Agent](/ai-agents/creating-your-first-agent).
  </Accordion>

  <Accordion title="How do context windows affect conversations?" icon="memory">
    Models only “remember” what fits in the current context window. Very long threads may push earlier details out of scope. If responses degrade, start a new conversation with a short summary. See [Conversation Management](/improved-usage/conversation-management).
  </Accordion>

  <Accordion title="How can I ask better questions?" icon="question">
    Be specific, set scope and audience, provide context, and request structured outputs. See patterns and templates in [Asking Great Questions](/improved-usage/asking-great-questions).
  </Accordion>
</AccordionGroup>

## External connections (MCP)

<AccordionGroup>
  <Accordion title="What is MCP and why does it matter?" icon="plug">
    MCP (Model Context Protocol) securely connects AI models to your softwares (email, calendar, CRM, storage, ERP, etc.) so the AI can read data and take actions with your permission. See [Tools & Connection → Understanding MCP](/tools-connection/introduction).
  </Accordion>

  <Accordion title="How do I connect my first external software?" icon="link">
    Open the Connectors settings (bottom left) → select a connection → authenticate (OAuth/credentials) → enable the connection for your AI model → test with a simple query. Full walkthrough in [Quick Start Guide](/welcome/quick-start).

    <Check>
      Start with email (Outlook Mail or Gmail) to unlock daily productivity.
    </Check>
  </Accordion>

  <Accordion title="What can WonkaChat access and do with connected softwares?" icon="hand">
    WonkaChat can only access data you already have permission to view and only when you explicitly ask. Actions depend on each connection's (MCP) capabilities (read, write, search, execute workflows). You control all actions through your natural language requests. See [Tools & Connection](/tools-connection/introduction).
  </Accordion>

  <Accordion title="How are my credentials stored?" icon="lock">
    WonkaChat stores authentication tokens securely (encrypted at rest), never your plain-text passwords. Tokens are per-user, auto-refreshed when needed, and deleted when you disconnect a tool. See [Security Overview](/security-governance/security-overview).
  </Accordion>

  <Accordion title="Why can't the AI access a file or record I asked for?" icon="triangle-exclamation">
    Most common reasons: you don't have access in the underlying tool, you denied permissions during authentication, or the tool isn't connected. Verify you can access it directly in the software, then reconnect and grant the requested permissions. See [Troubleshooting Connections](/tools-connection/troubleshooting-connections).
  </Accordion>

  <Accordion title="Which external connections are supported?" icon="puzzle-piece">
    WonkaChat supports 40+ connections across communication, development, operations, analytics, and more including Outlook, Gmail, Slack, GitHub, Notion, Google Drive, Salesforce, HubSpot, Asana, and many others. See the complete list in [Tools & Connection](/tools-connection/introduction).
  </Accordion>
</AccordionGroup>

## AI agents

<AccordionGroup>
  <Accordion title="What are AI Agents and how are they different from regular chat?" icon="robot">
    AI Agents have persistent instructions, dedicated tools, and consistent roles across conversations - like specialized team members. Regular chat starts fresh each time. Agents are ideal for recurring tasks with specific requirements. See [AI Agents Introduction](/ai-agents/introduction).
  </Accordion>

  <Accordion title="When should I create or use an AI Agent?" icon="bullseye">
    Use AI Agents for recurring workflows, multi-tool processes, or when you need standardized outputs. Create personal agents for your specific workflows, or use organization-wide agents shared by administrators. See [Creating Your First Agent](/ai-agents/creating-your-first-agent).
  </Accordion>

  <Accordion title="How do I find and access agents?" icon="magnifying-glass">
    Access agents through the Agent Marketplace or "My Agents" section in your WonkaChat interface. You can search, filter, and select agents based on your needs. See [Finding and Accessing Agents](/ai-agents/finding-and-accessing-agents).
  </Accordion>

  <Accordion title="What if an agent doesn't have the information or tools it needs?" icon="wrench">
    * **As a user**: Provide missing details in your prompt, or choose a different agent with the appropriate tools configured
    * **As an administrator**: Update the agent's instructions, add relevant documents, or connect the required external tools

    See [Using Existing Agents](/ai-agents/using-existing-agents).
  </Accordion>

  <Accordion title="What is Business Teaching and why should I use it?" icon="graduation-cap">
    Business Teaching allows you to add your company's terminology, policies, and processes to AI Agents through custom instructions and uploaded documents. This ensures responses match your organization's specific context rather than generic assumptions. See [Business Teaching](/ai-agents/business-teaching).
  </Accordion>
</AccordionGroup>

## Security and governance

<AccordionGroup>
  <Accordion title="Does WonkaChat train AI models on my data?" icon="ban">
    No. Your conversations are stored securely in WonkaChat for your history and reference, but model providers process requests ephemerally without data retention or training. Your data never trains AI models. See [Security Overview](/security-governance/security-overview).
  </Accordion>

  <Accordion title="How is my data isolated and protected?" icon="building-lock">
    WonkaChat uses organization-scoped isolation, encryption in transit (TLS) and at rest, secure token storage, and automatic redaction of sensitive data in logs. Your data stays within your security perimeter. See [Security Overview](/security-governance/security-overview).
  </Accordion>

  <Accordion title="What authentication methods are supported?" icon="key">
    WonkaChat supports Google Workspace, Microsoft 365, and email/password authentication. Enterprise plans can configure custom SSO providers. Access tokens and refresh tokens balance security with usability. See [Security Overview](/security-governance/security-overview).
  </Accordion>

  <Accordion title="How does access control and sharing work?" icon="user-shield">
    WonkaChat has two roles: **User** and **Administrator**. Administrators manage users, share AI Agents organization-wide, and control tool access. Users can create personal agents but cannot share them across the organization. See [Access Control](/security-governance/access-control).
  </Accordion>

  <Accordion title="Can AI Agents escalate privileges or bypass permissions?" icon="shield-halved">
    No. AI Agents inherit your exact permissions from connected softwares. If you can't perform an action manually, agents can't either. Tool restriction features let you remove specific capabilities from agents for additional safety. See [Access Control](/security-governance/access-control).
  </Accordion>

  <Accordion title="What is Safe Mode and when should I use it?" icon="shield-check">
    Safe Mode requires your approval before AI Agents execute actions. Use it for sensitive operations like sending communications, modifying production data, deployments, and financial transactions. See [Access Control](/security-governance/access-control).
  </Accordion>
</AccordionGroup>

## Troubleshooting

<AccordionGroup>
  <Accordion title="I'm seeing authentication errors when connecting a tool" icon="circle-xmark">
    Verify your credentials by logging into the tool directly first. Then disconnect and reconnect through WonkaChat's Connectors settings, ensuring you grant all requested permissions during authentication. See [Troubleshooting Connections](/tools-connection/troubleshooting-connections).
  </Accordion>

  <Accordion title="I get 'permission denied' when accessing data" icon="ban">
    Common causes: your organization blocks third-party apps, you didn't grant all required permissions during setup, or your account lacks access to that data. Contact your IT team to approve WonkaChat, then reconnect with full permissions. See [Troubleshooting Connections](/tools-connection/troubleshooting-connections).
  </Accordion>

  <Accordion title="Responses are slow or timing out" icon="clock">
    Try these steps: refresh WonkaChat, check your internet connection, reduce query scope (use date ranges or filters), switch to a faster model, or verify the connected software's status. See [Troubleshooting Connections](/tools-connection/troubleshooting-connections).
  </Accordion>

  <Accordion title="AI Agent responses don't match my expectations" icon="pen-to-square">
    Be more specific about desired format and scope in your prompt. Review the agent's instructions to ensure alignment. Administrators can refine agent instructions, add business context, or connect additional tools. See [Using Existing Agents](/ai-agents/using-existing-agents) and [Asking Great Questions](/improved-usage/asking-great-questions).
  </Accordion>

  <Accordion title="Where can I get help?" icon="life-ring">
    Visit [Getting Help](/support-resources/getting-help) for support options, contact our team at [meetwonka.com/contact](https://www.meetwonka.com/fr/contact), or explore [Troubleshooting](/support-resources/troubleshooting) for common issues and solutions.
  </Accordion>
</AccordionGroup>

## Prompting, conversations, and the prompt library

<AccordionGroup>
  <Accordion title="How do I write better prompts?" icon="list">
    Structure prompts with clear context, specific goals, desired output format, and relevant constraints. Provide examples when helpful and specify your audience. The more specific you are, the better results you'll get. See [Asking Great Questions](/improved-usage/asking-great-questions).
  </Accordion>

  <Accordion title="When should I start a new conversation?" icon="arrows-turn-to-dots">
    Start a new conversation when changing topics, after very long threads (context window limits), or when response quality declines. Begin the new conversation with a brief summary of relevant context if needed. See [Conversation Management](/improved-usage/conversation-management).
  </Accordion>

  <Accordion title="How do I save and reuse prompts?" icon="bookmark">
    Use the Prompt Library to save effective prompts with variables like `{{current_date}}` or `{{current_user}}`. Create reusable templates for common tasks and invoke them quickly with slash commands. See [Prompt Library](/improved-usage/prompt-library).
  </Accordion>
</AccordionGroup>

## Policies and risk management

<AccordionGroup>
  <Accordion title="What practices reduce risk when using AI?" icon="clipboard-check">
    * Enable Safe Mode for high-impact actions
    * Test workflows in non-production environments first
    * Use least-privilege access (only grant necessary permissions)
    * Break complex workflows into reviewable steps
    * Remove destructive capabilities from AI Agents when not needed

    See [Best Practices](/security-governance/best-practices).
  </Accordion>

  <Accordion title="Can I undo actions performed by AI?" icon="rotate-left">
    Many tool actions are irreversible (sending emails, deleting data, deployments, financial transactions). Always use Safe Mode for sensitive operations, restrict destructive capabilities, and review outputs before final execution. See [Best Practices](/security-governance/best-practices).
  </Accordion>
</AccordionGroup>

***

<Card title="Still have questions?" icon="life-ring" href="/en/support-resources/getting-help">
  Visit Getting Help to reach our team or explore more support resources.
</Card>


# Getting Help
Source: https://help.wonka.chat/en/support-resources/getting-help

Information on the available support channels and how to get help when you need it.

**How to Get Help:**

<CardGroup>
  <Card title="Email Support" icon="envelope" href="mailto:support@meetwonka.com">
    Need help or want to provide feedback?\
    Email our support team anytime:

    * Report issues or errors
    * Ask questions about the platform
    * Share suggestions for improvements

    **Response Time:** Within 1 business day\
    [support@meetwonka.com](mailto:support@meetwonka.com)
  </Card>

  <Card title="Book a Support Call" icon="headset" href="https://outlook.office.com/bookwithme/user/b57675e919ea45ee863231c0e4fa8e18@meetwonka.com/meetingtype/aUxDhvQJAUSwOd-foKjQZg2?anonymous&ismsaljsauthenabled&ep=mcard">
    Have questions about our services or prefer a live conversation?

    * Schedule a 1:1 support call
    * Get personalized assistance
    * Discuss your needs with a product specialist

    **Book a call:** Choose a time that works for you.
  </Card>
</CardGroup>


# Examples of Connection Usage
Source: https://help.wonka.chat/en/tools-connection/examples-of-tool-usage

See external connections (MCP) in action with real examples of Outlook, Odoo, and other business systems.

Learn how to use these connections effectively by watching real examples of WonkaChat working with your business systems.

<Info>
  These examples demonstrate the power of external connections, showing how AI can interact with your business systems through natural conversation.
</Info>

***

## Real-World Examples

Watch these examples to see how the external connections work with different types of business systems.

<Tabs>
  <Tab title="Email Management">
    ### Managing Emails with AI

    See how WonkaChat connects to Gmail to help you manage emails, create drafts, and stay organized only by using natural language conversation.

    <iframe title="Gmail connection in action" />

    ### Common Gmail Use Cases

    <CardGroup>
      <Card title="Daily Briefing" icon="newspaper">
        "Give me my daily briefing: today's calendar, unread emails from my team, and any urgent items"
      </Card>

      <Card title="Meeting Preparation" icon="clipboard-list">
        "Find all emails about the Q4 planning meeting and summarize the key discussion points"
      </Card>

      <Card title="Email Cleanup" icon="trash">
        "Show me emails older than 30 days from newsletters and promotional senders"
      </Card>

      <Card title="Follow-Up Tracking" icon="clock-rotate-left">
        "Which emails am I waiting for responses on? Show me emails I sent that haven't been replied to"
      </Card>
    </CardGroup>

    <Info>
      The same capabilities shown with **Gmail** work identically with **Outlook**, or any other email and calendar provider that has an MCP server available on WonkaChat. Just connect the corresponding provider instead!
    </Info>
  </Tab>

  <Tab title="Odoo ERP System">
    ### Managing Business Operations with Odoo

    Watch how WonkaChat integrates with Odoo to help you manage customers, inventory, sales, and operations through AI-powered conversation.

    <iframe title="Odoo connection demonstration" />

    ### Go further

    <AccordionGroup>
      <Accordion title="Customer Management" icon="users">
        **Example Queries:**

        * "Show me our top 10 customers by revenue this quarter"
        * "Find the contact information for Acme Corporation"
        * "Create a new customer record for ABC Company with email [contact@abc.com](mailto:contact@abc.com)"

        **What Happens:**

        1. AI connects to Odoo CRM module
        2. Queries customer database
        3. Retrieves and formats customer information
        4. Can create or update customer records

        **Benefits:**

        * Quick access to customer data
        * No need to navigate Odoo interface
        * Natural language queries instead of filters
        * Update records through conversation
      </Accordion>

      <Accordion title="Sales Pipeline" icon="chart-line">
        **Example Queries:**

        * "What's the status of our sales pipeline this month?"
        * "Show me all opportunities in negotiation stage"
        * "Update the TechCorp deal to 'Proposal Sent' stage"

        **What Happens:**

        1. AI accesses Odoo sales module
        2. Retrieves pipeline data and opportunities
        3. Presents current status and metrics
        4. Can update opportunity stages and details

        **Benefits:**

        * Real-time pipeline visibility
        * Quick status updates
        * Identify bottlenecks faster
        * Track deals without logging into Odoo
      </Accordion>

      <Accordion title="Inventory Management" icon="warehouse">
        **Example Queries:**

        * "What's our current stock level for Product SKU-12345?"
        * "Show me products below reorder level"
        * "Create a purchase order for 100 units of SKU-67890"

        **What Happens:**

        1. AI queries Odoo inventory module
        2. Retrieves real-time stock levels
        3. Identifies low stock items
        4. Can initiate purchase orders or transfers

        **Benefits:**

        * Instant inventory visibility
        * Proactive low-stock alerts
        * Faster reordering process
        * Reduce stockouts and overstocking
      </Accordion>

      <Accordion title="Order Processing" icon="receipt">
        **Example Queries:**

        * "Show me orders that need to be shipped"
        * "What's the status of order SO-2024-001?"
        * "Create a sales order for customer ABC Company: 50 units of Product X"

        **What Happens:**

        1. AI accesses Odoo sales and warehouse modules
        2. Retrieves order information
        3. Shows fulfillment status
        4. Can create new orders

        **Benefits:**

        * Track orders without manual lookup
        * Faster order creation
        * Identify fulfillment bottlenecks
        * Improve customer response time
      </Accordion>

      <Accordion title="Financial Insights" icon="chart-pie">
        **Example Queries:**

        * "What's our revenue this quarter compared to last quarter?"
        * "Show me outstanding invoices"
        * "Which customers have overdue payments?"

        **What Happens:**

        1. AI connects to Odoo accounting module
        2. Retrieves financial data and reports
        3. Performs calculations and comparisons
        4. Identifies receivables and payables

        **Benefits:**

        * Quick financial insights
        * No need to generate formal reports
        * Track receivables proactively
        * Better cash flow management
      </Accordion>
    </AccordionGroup>

    ### Common Odoo Use Cases

    <CardGroup>
      <Card title="Operations Dashboard" icon="gauge">
        "Give me an operations overview: pending orders, low stock items, and overdue invoices"
      </Card>

      <Card title="Customer Service" icon="headset">
        "Find the order history and open support tickets for customer TechCorp"
      </Card>

      <Card title="Sales Reports" icon="file-chart-column">
        "Compare our sales performance: this month vs. last month, broken down by product category"
      </Card>

      <Card title="Procurement" icon="cart-shopping">
        "Which products do I need to reorder? Show me everything below minimum stock level and create purchase orders"
      </Card>
    </CardGroup>

    <Info>
      Odoo has many modules (CRM, Sales, Inventory, Accounting, HR, Manufacturing, etc.). The external connection (MCP) works with all installed modules in your Odoo instance.
    </Info>
  </Tab>

  <Tab title="Multi-Tool Example">
    ### Multi-Tool: Connecting Gmail and Odoo for a Unified Overview

    See a real-world example of how WonkaChat connects to both Gmail and Odoo to provide a unified view of your Odoo contacts, triggered from your email context.

    <iframe title="Multi-tool: Gmail and Odoo contact overview" />

    Connect Gmail and Odoo with WonkaChat to instantly see who you need to contact.
    WonkaChat pulls your contacts from your CRM (e.g. Odoo) and creates a ready-to-send draft in your mailing system (e.g. Gmail) with all the details.

    <Check>
      Effortlessly bridge information between systems and take action in a single step.
    </Check>
  </Tab>

  <Tab title="More Examples">
    ### Additional Business System Examples

    While we have detailed video examples for Outlook and Odoo, here are quick demonstrations of other popular external connections:

    <AccordionGroup>
      <Accordion title="Slack - Team Communication" icon="slack">
        **What You Can Do:**

        * Read messages from specific channels
        * Send messages or announcements
        * Search conversation history
        * Monitor mentions and threads
        * Post automated updates

        **Example Queries:**

        ```
        "Show me messages in #engineering channel from today"
        "Post in #general: 'Team meeting moved to 3 PM'"
        "Search Slack for conversations about the new feature"
        "What are my unread mentions?"
        ```

        **Use Cases:**

        * Stay updated without switching to Slack
        * Automate team notifications
        * Search across all channels at once
        * Monitor important channels
      </Accordion>

      <Accordion title="GitHub - Code Repository" icon="github">
        **What You Can Do:**

        * View repository information
        * List and filter issues
        * Check pull request status
        * Read code files
        * Review commit history

        **Example Queries:**

        ```
        "Show me open issues in the main repository"
        "What's the status of PR #234?"
        "List recent commits to the backend folder"
        "Show me high-priority bugs assigned to me"
        ```

        **Use Cases:**

        * Quick code reviews
        * Issue tracking without leaving WonkaChat
        * Monitor team activity
        * Code search across repositories
      </Accordion>

      <Accordion title="Google Drive - File Storage" icon="google-drive">
        **What You Can Do:**

        * Search for files and folders
        * Read document contents
        * View file metadata
        * List shared files
        * Access recent files

        **Example Queries:**

        ```
        "Find the Q3 planning document"
        "Show me files shared with me this week"
        "What's in the 'Project Alpha' folder?"
        "List my recent spreadsheets"
        ```

        **Use Cases:**

        * Fast file search
        * Document retrieval
        * Content review
        * Shared file tracking
      </Accordion>

      <Accordion title="Salesforce - CRM" icon="salesforce">
        **What You Can Do:**

        * Query leads and opportunities
        * View account information
        * Check pipeline status
        * Update records
        * Run reports

        **Example Queries:**

        ```
        "Show me leads from this month"
        "What's the value of our pipeline in 'Proposal' stage?"
        "Update the TechCorp opportunity to closed-won"
        "Which accounts haven't been contacted in 30 days?"
        ```

        **Use Cases:**

        * Sales pipeline monitoring
        * Lead management
        * Account research
        * Quick record updates
      </Accordion>

      <Accordion title="Jira - Project Management" icon="jira">
        **What You Can Do:**

        * View and filter issues
        * Check sprint progress
        * Update issue status
        * Assign tasks
        * Track project boards

        **Example Queries:**

        ```
        "Show me issues assigned to me"
        "What's the status of the current sprint?"
        "Update PROJ-123 to 'In Review'"
        "List all bugs with high priority"
        ```

        **Use Cases:**

        * Task management
        * Sprint tracking
        * Bug tracking
        * Team workload visibility
      </Accordion>
    </AccordionGroup>

    ### Multi-Business System Workflow Examples

    The real power of external connections (MCP) comes from combining multiple business systems:

    <CardGroup>
      <Card title="Research to Documentation" icon="arrows-turn-right">
        **Business Systems:** Google Drive + Odoo

        "Search my Google Drive for market research files, summarize the findings, and create a summary page in Odoo"
      </Card>

      <Card title="Issue to Communication" icon="bug-slash">
        **Business Systems:** GitHub + Slack

        "Find critical issues in the main repo and post a summary to #engineering channel on Slack"
      </Card>

      <Card title="Sales Pipeline to Email" icon="envelope-circle-check">
        **Business Systems:** Salesforce + Outlook

        "Show me opportunities closing this week and draft follow-up emails for each contact"
      </Card>

      <Card title="Task Sync" icon="arrows-rotate">
        **Business Systems:** Jira + ClickUp

        "Find completed tasks in Jira this week and summarize them for my weekly update in ClickUp"
      </Card>

      <Card title="Meeting Preparation" icon="presentation-screen">
        **Business Systems:** Outlook + Google Drive + Odoo

        "Get tomorrow's meeting agenda from Outlook, find related documents in Drive, and create a prep page in Odoo"
      </Card>

      <Card title="Customer Intelligence" icon="magnifying-glass-chart">
        **Business Systems:** Salesforce + Outlook + Slack

        "Gather all information about Acme Corp: Salesforce account, recent emails, and Slack mentions"
      </Card>
    </CardGroup>
  </Tab>
</Tabs>

***

## Tips for Effective Business System Usage

<AccordionGroup>
  <Accordion title="Start with Simple Queries" icon="seedling">
    When you first connect a business system:

    1. **Test basic access**: "Show me my recent items"
    2. **Verify permissions**: Try read operations before write operations
    3. **Explore capabilities**: Ask "What can you do with \[business system]?"
    4. **Gradually increase complexity**: Start simple, then combine operations

    <Tip>
      Don't worry about getting the query perfect. Just describe what you want in natural language and the AI will figure out the right business system operations.
    </Tip>
  </Accordion>

  <Accordion title="Be Specific When Needed" icon="crosshairs">
    While AI handles natural language well, specific details improve results:

    **Less Specific:**

    ```
    "Show me emails"
    ```

    Result: May return too many results or default to today

    **More Specific:**

    ```
    "Show me unread emails from my manager this week"
    ```

    Result: Precisely filtered results

    **Even More Specific:**

    ```
    "Find emails from john@company.com with 'budget' in subject, received in the last 30 days"
    ```

    Result: Exactly what you need
  </Accordion>

  <Accordion title="Use Time Constraints" icon="clock">
    Adding time boundaries improves performance and relevance:

    * "today" / "this week" / "this month"
    * "in the last 7 days"
    * "between January 1 and January 31"
    * "after March 15th"
    * "before the end of Q2"

    **Example:**

    ```
    "Show me Slack messages in #engineering from yesterday"
    ```

    Much faster than querying all messages ever sent.
  </Accordion>

  <Accordion title="Leverage Multi-Step Requests" icon="list-ol">
    You can ask AI to perform multiple operations in sequence:

    ```
    "Check my calendar for availability tomorrow afternoon, 
    then email those times to sarah@company.com"
    ```

    AI will:

    1. Connect to the calendar
    2. Find available slots
    3. Connect to the email
    4. Draft and send email

    **No need to break it into separate queries!**
  </Accordion>

  <Accordion title="Review Before Action" icon="eye">
    For operations that modify data (send, create, update, delete):

    * Ask the AI to show you a preview before any changes are made
    * Review the proposed action thoroughly before approving it
    * Request modifications as needed: "Make the email more formal"
    * Always verify recipient addresses and all critical information
    * Direct the AI when you want to pause: "Don't send this yet, just show me the draft"

    <Info>
      To enhance safety, we provide a <strong>Safe Mode</strong> that keeps you in control whenever agents or AI models perform actions. When Safe Mode is enabled, you must explicitly approve or reject all modifications before they take effect.

      <br />

      Learn more in the <a href="#safe-mode-human-in-the-loop">Safe Mode: Human-in-the-Loop</a> section.
    </Info>

    <Warning>
      Some actions (such as sending emails or creating orders) may be irreversible. Always double-check your intent before confirming the AI to proceed.
    </Warning>
  </Accordion>
</AccordionGroup>

<Check>
  You now have a complete understanding of how MCP connections work in practice. Start connecting business systems and experimenting with your own use cases!
</Check>


# External Connections
Source: https://help.wonka.chat/en/tools-connection/introduction

Connect WonkaChat to your softwares, business tools and data sources to access real information and take actions on your behalf.

WonkaChat can connect to your business tools, databases, and software, enabling you to access real information and take actions through natural conversation. These are called **Connectors** or **MCP connections** for more technical people.

<Info>
  Want to understand the technical details of how these connections work?
  [Jump to Technical Details](#technical-details) to learn about the Model Context Protocol (MCP) that powers these integrations.
</Info>

<Tabs>
  <Tab title="Getting Started">
    ## What External connection Enables

    Connecting your external softwares and data sources transforms WonkaChat from a standalone AI assistant into a connected system that can work with your actual data and workflows. Instead of just answering general questions, WonkaChat can access your real emails, manage your calendar, update your CRM, create documents in your systems, and coordinate actions across multiple platforms, all through natural conversation.

    ### Real-World Capabilities

    <CardGroup>
      <Card title="Access Your Data" icon="database">
        Read emails, calendar events, documents, and databases through natural conversation without switching between apps.
      </Card>

      <Card title="Take Actions" icon="bolt">
        Send emails, create tasks, update records, and trigger workflows, all by asking WonkaChat in plain language.
      </Card>

      <Card title="Multi-Platform Workflows" icon="layer-group">
        Combine multiple connections in a single request. For example: "Check my calendar, find open slots, and email them to John."
      </Card>

      <Card title="Always Up-to-Date" icon="arrows-rotate">
        Access live data from your systems rather than relying on static knowledge or manual updates.
      </Card>
    </CardGroup>

    ### How to Get Started

    <Steps>
      <Step title="Start a connection">
        Go to the settings and connect to the tool (like Outlook, Gmail, or Slack) by providing your credentials. Then add it to your chat.

        <Frame>
          <img alt="Outlook Mail button" />
        </Frame>

        <Info>
          These external connections are called "MCP" (Model Context Protocol).

          While **you don't need to understand the technical details to use them**, you can learn more in the [Technical Details](#technical-details) section if you're curious.
        </Info>
      </Step>

      <Step title="WonkaChat gets secure access">
        A secure connection is established between WonkaChat and your business tools. WonkaChat works on your behalf using natural language, operating within your existing permissions.

        <Warning>
          **Important:** WonkaChat inherits your exact permissions. It can only access data and perform actions that you personally have permission to do. If you can't access a file, send an email from a specific account, or modify a record in your regular workflow, WonkaChat can't do it either.
        </Warning>
      </Step>

      <Step title="Chat naturally">
        Ask questions or make requests in plain language. WonkaChat automatically uses the right tools and actions to help you.

        ```
        "What emails did I receive from Sarah this week?"
        ```

        WonkaChat connects to your email, retrieves the information, and presents it clearly.
      </Step>
    </Steps>

    ***

    ## External Connections (MCP) Example

    Connecting your business tools dramatically changes what WonkaChat can accomplish. Here's a real example showing the difference:

    <Tabs>
      <Tab title="Without Connections">
        **Your request:**

        ```
        "Schedule a meeting with the marketing team for next Tuesday at 2 PM to discuss the Q2 campaign results."
        ```

        **WonkaChat's response (limited):**

        ```
        I can help you draft a meeting invitation, but I cannot access your calendar 
        or email system to actually schedule the meeting. 

        Here's what I suggest:
        1. Check your calendar for availability on Tuesday at 2 PM
        2. Send a meeting invite to your marketing team
        3. Include "Q2 Campaign Results Discussion" as the subject

        Would you like me to draft the meeting invitation text for you?
        ```

        <Warning>
          Without connections, WonkaChat can only provide general guidance and cannot access your actual data or take real actions.
        </Warning>
      </Tab>

      <Tab title="With Connections">
        **Your request:**

        ```
        "Schedule a meeting with the marketing team for next Tuesday at 2 PM to discuss the Q2 campaign results."
        ```

        **WonkaChat's response (powerful):**

        ```
        I'll schedule that meeting for you.

        Checking your calendar... Tuesday at 2 PM is available.

        Looking up "marketing team" members... Found 5 team members:
        - Sarah Johnson (Marketing Director)
        - Mike Chen (Content Lead)  
        - Emma Rodriguez (Social Media Manager)
        - David Kim (Analytics Specialist)
        - Lisa Wang (Designer)

        Creating meeting invite with:
        - Title: "Q2 Campaign Results Discussion"
        - Time: Tuesday, October 21, 2025 at 2:00 PM - 3:00 PM
        - Location: Conference Room B (reserved)
        - Attendees: All 5 marketing team members

        ✓ Meeting scheduled and invitations sent.
        ✓ Conference room reserved.
        ✓ Calendar updated.

        Is there anything else you'd like me to add to the meeting?
        ```

        <Check>
          With connections, WonkaChat accesses your real calendar, identifies team members, checks room availability, and completes the entire task automatically.
        </Check>
      </Tab>
    </Tabs>

    This example shows how connecting your business tools enables WonkaChat to move from giving advice to taking real actions with your actual business data.

    ***

    ## Available External Connections (MCP)

    WonkaChat supports MCP connections to **40+ business tools** across all major categories. Here are all the currently available connections:

    <AccordionGroup>
      <Accordion title="Microsoft 365 & Productivity" icon="microsoft">
        Connect your Microsoft tools for comprehensive office productivity.

        **Available Connections:**

        * Outlook Mail (email)
        * Outlook Calendar (scheduling)
        * Microsoft Teams (communication)
        * OneDrive (storage)
        * Excel (spreadsheets)
        * Word (documents)
      </Accordion>

      <Accordion title="Google Workspace" icon="google">
        Integrate with Google's suite of productivity and collaboration tools.

        **Available Connections:**

        * Gmail (email)
        * Google Calendar (scheduling)
        * Google Drive (storage)
        * Google Docs (documents)
        * Google Sheets (spreadsheets)
      </Accordion>

      <Accordion title="Project Management & Collaboration" icon="list-check">
        Connect your project management and team collaboration platforms.

        **Available Connections:**

        * Asana (project management)
        * ClickUp (productivity)
        * Jira (issue tracking)
        * Linear (project management)
        * Monday (work management)
        * Notion (knowledge base)
        * Confluence (documentation)
      </Accordion>

      <Accordion title="CRM & Sales" icon="handshake">
        Integrate with customer relationship management and sales platforms.

        **Available Connections:**

        * Salesforce (CRM)
        * HubSpot (CRM & marketing)
        * Attio (CRM)
        * Close (sales CRM)
        * Klaviyo (marketing automation)
      </Accordion>

      <Accordion title="Development & Engineering" icon="code">
        Connect your development tools, repositories, and deployment platforms.

        **Available Connections:**

        * GitHub (version control)
        * GitLab (version control)
        * Vercel (deployment)
        * Supabase (database)
      </Accordion>

      <Accordion title="Communication & Scheduling" icon="comments">
        Integrate messaging and scheduling tools for seamless communication.

        **Available Connections:**

        * Slack (team messaging)
        * Cal.com (scheduling)
        * LinkedIn (professional networking)
      </Accordion>

      <Accordion title="Storage & Databases" icon="database">
        Connect file storage and data management platforms.

        **Available Connections:**

        * Airtable (database)
        * Box (storage)
        * Fathom (analytics)
      </Accordion>

      <Accordion title="Design & Specialized Tools" icon="palette">
        Integrate design and specialized business tools.

        **Available Connections:**

        * Figma (design)
        * Odoo (ERP)
      </Accordion>
    </AccordionGroup>

    <Info>
      Need a connection that's not listed? [Contact our support team](/support-resources/getting-help) to request new MCP integrations. We're constantly expanding our connection library based on user needs.
    </Info>
  </Tab>

  <Tab title="Technical Details">
    ## Understanding the Model Context Protocol (MCP)

    WonkaChat uses the **Model Context Protocol (MCP)** to connect AI models with your business tools, databases, and software, enabling AI to access real information and take actions on your behalf.

    <Info>
      MCP is an open protocol that allows AI systems to communicate with external data sources and softwares in a uniform way.

      Think of it as a universal connector (like USB for AI systems) that allows different AI models and software tools to work together seamlessly without custom integrations for each combination.
    </Info>

    ### How MCP Works

    Instead of building custom connectors for each AI and data source pair, MCP allows all AI clients and data sources to be connected through a single integration.

    <Frame>
      <img alt="MCP simplifies connections between AI models and business tools" />

      <img alt="MCP simplifies connections between AI models and business tools" />
    </Frame>

    **In simple terms:** The AI Client (WonkaChat) sends structured requests through MCP. The MCP Server communicates with your underlying systems (Outlook, GitHub, Odoo, etc.) and sends back a structured response.

    ***

    ## Risk Management & Security

    <Frame>
      <img alt="MCP provides centralized security, compliance, and governance for AI systems" />

      <img alt="MCP provides centralized security, compliance, and governance for AI systems" />
    </Frame>

    <Warning>
      One of the biggest concerns around AI is how it accesses and uses sensitive data.
    </Warning>

    **MCP addresses this by acting as a security checkpoint for all your AI systems**, tracking every data interaction and enforcing company policies. Think of it as having a detailed log of who accessed what, when, and why, which is essential for regulatory compliance.

    ### How MCP Protects Your Data

    <AccordionGroup>
      <Accordion title="Central Security Layer" icon="shield-halved">
        MCP introduces a central layer through which all AI interactions are channeled. This provides **consistency in security policies**, enabling companies to enforce authentication, authorization, and data masking at the protocol level.

        All your AI tools connect through one secure gateway, rather than having separate security implementations for each connection.
      </Accordion>

      <Accordion title="Controlled AI Behavior" icon="gavel">
        MCP aligns AI behavior with internal business policies. Because all capabilities are exposed and managed through the MCP server, companies can define exactly **what an AI system is allowed to do** and nothing more.

        AI systems **can't access sensitive data** or trigger actions they haven't been explicitly granted. Every permission is controlled and auditable.
      </Accordion>

      <Accordion title="Complete Audit Trail" icon="clipboard-list">
        Every interaction through MCP is logged:

        * **Who** accessed the data (user identity)
        * **What** data was accessed (resources, actions)
        * **When** it happened (timestamps)
        * **Why** it was needed (context, request details)

        This detailed logging is essential for regulatory compliance and security audits.
      </Accordion>
    </AccordionGroup>

    ***

    ## MCP Architecture Overview

    The Model Context Protocol is an open standard developed by [Anthropic](https://www.anthropic.com/news/model-context-protocol) that enables secure, two-way connections between AI applications and external data sources.

    <Info>
      WonkaChat implements MCP as a client, connecting to MCP servers that expose your business tools and data to the AI models. Learn more at [modelcontextprotocol.io](https://modelcontextprotocol.io).
    </Info>

    ### Technical Implementation

    MCP uses a **client-server architecture** where WonkaChat acts as the MCP Client, connecting to MCP Servers that integrate with your business tools:

    <Frame>
      <img alt="WonkaChat MCP architecture diagram" />

      <img alt="WonkaChat MCP architecture diagram" />
    </Frame>

    **Architecture Flow:**

    * **WonkaChat (MCP Client)** sends structured requests and manages authentication
    * **MCP Server** acts as the integration layer, connecting to external systems
    * **External Systems** (Outlook, Odoo, Salesforce, etc.) provide data via their APIs
    * All responses flow back through the same secure channel

    **Key Technical Benefits:**

    * JSON-RPC based communication protocol
    * Capability negotiation during initialization
    * Clear security boundaries via client-host-server architecture
    * User consent mechanisms for AI interactions

    ### Security Implementation

    WonkaChat implements MCP with enterprise-grade security features that build upon the protocol's foundation:

    <AccordionGroup>
      <Accordion title="MCP Protocol Security" icon="shield">
        **Built into the Protocol:**

        * **User Consent**: All AI actions require explicit user authorization ([MCP Best Practices](https://modelcontextprotocol.io/specification/draft/basic/security_best_practices))
        * **Capability Negotiation**: Clients and servers agree on available features during initialization
        * **Token Validation**: Servers only accept tokens explicitly issued for them (OAuth 2.1 framework)
        * **Session Management**: Secure session handling to prevent hijacking and impersonation
        * **Security Boundaries**: Client-host-server architecture with clear isolation

        **Recommended by MCP:**

        * Input validation and sanitization for all parameters
        * Rate limiting and abuse monitoring
        * Secure transport (TLS encryption) for all connections
      </Accordion>

      <Accordion title="WonkaChat's Implementation" icon="lock">
        **Authentication & Access Control:**

        * OAuth 2.0 for third-party services (Google, Microsoft, Salesforce)
        * API key authentication for developer tools
        * Token encryption at rest (AES-256) with automatic refresh
        * Per-user token storage with complete isolation
        * Users inherit only their existing platform permissions

        **Enterprise Security:**

        * Role-Based Access Control (RBAC) at organizational level
        * Centralized security policies across all MCP connections
        * Connection timeouts and resource protection
        * Multi-region deployment with secure failover
      </Accordion>

      <Accordion title="Audit & Compliance" icon="clipboard-list">
        **WonkaChat's Audit Trail:**

        <Info>
          Detailed audit and compliance features documentation coming soon.

          These are WonkaChat implementation features, not built into the MCP protocol itself.
        </Info>
      </Accordion>
    </AccordionGroup>

    <Tip>
      If you have specific requirements or unique needs for MCP integrations, reach out to our support team. We're here to help expand or tailor connections to fit your workflows.
    </Tip>
  </Tab>
</Tabs>


# Your First External Connection
Source: https://help.wonka.chat/en/tools-connection/setting-up-your-first-connection

Connect WonkaChat to your business systems and start using AI-powered workflows.

Connect your first business system to WonkaChat and unlock AI-powered access to your business data. This guide walks you through the complete connection process, from selecting a connection to testing your first AI-powered request.

## Understanding Connections vs Tools

<AccordionGroup>
  <Accordion title="Connections" icon="usb">
    **Connections** are what you add to AI models - they link WonkaChat to your business systems like Outlook, Slack, or Google Drive.
  </Accordion>

  <Accordion title="Tools" icon="wrench">
    **Tools** are the individual capabilities within each connection. For example, an Outlook connection might include tools for reading emails, sending messages, and managing calendar events.
  </Accordion>
</AccordionGroup>

## Before You Start

<CardGroup>
  <Card title="Check Your Access" icon="user-check">
    Ensure you have login credentials for the business system you want to connect (e.g., your Outlook email and password).
  </Card>

  <Card title="Know Your Permissions" icon="shield">
    You'll only be able to access data in connected business systems that you already have permission to view or edit.
  </Card>
</CardGroup>

## Video Walkthrough

Watch this step-by-step guide to see the entire connection process in action:

<iframe title="Complete guide to setting up your first connection" />

<Tip>
  Follow along with the video while setting up your first connection. Pause at each step to complete it in your own WonkaChat instance.
</Tip>

## Step-by-Step Connection Process

<Steps>
  <Step title="Locate the Connectors Panel">
    In your WonkaChat interface, find the **Connectors** inside the settings at the bootom left of the screen. This panel displays all available business system connections.

    <Frame>
      <img alt="Connectors panel and Outlook Mail connection example" />
    </Frame>

    <Tip>
      If you don't see the right side panel, click the arrow icon in the middle-right of your screen to expand the sidebar.
    </Tip>
  </Step>

  <Step title="Select Your Connection">
    Browse or search for the connection you want to add. Click the connection's button to begin the connection process.

    **Popular first connections:**

    * **Outlook Mail** or **Gmail** for email access
    * **Slack** for team communication
    * **Google Drive** or **OneDrive** for document access

    <Check>
      Not sure which software to connect first? Start with your email provider, it's the most versatile connection for daily tasks.
    </Check>
  </Step>

  <Step title="Authenticate Your Account">
    A credential window will appear, requesting authentication for the selected business system. The process varies slightly by connection type:

    <Tabs>
      <Tab title="OAuth">
        <Info>
          **OAuth** authentication is used for most of the softwares like <strong>Outlook, Gmail, Google Drive</strong>.
        </Info>

        1. Click **"Connect"**.
        2. You'll be redirected to the software's official login page.
        3. Enter your credentials on the software's secure site.
        4. You'll be automatically redirected back to WonkaChat.
      </Tab>

      <Tab title="API Key">
        <Info>
          Some softwares require an API key. If you need assistance obtaining or using your API key, see the <a href="/en/support-resources/getting-help">Getting Help</a> page for personalized support options.
        </Info>

        1. Click the **Gear Icon**.
        2. Obtain your API key directly from the provider.
        3. Paste the API key into the WonkaChat credential field.
        4. Click **"Connect"**.

        <Tip>
          Need help finding your API key? Most softwares have it under <strong>Settings → API</strong> or <strong>Settings → Integrations</strong>. Check the software's documentation if you're not sure.
        </Tip>
      </Tab>

      <Tab title="Custom">
        <Info>
          Some business systems, such as <strong>Odoo</strong>, require custom configuration beyond standard authentication. If you need assistance to connect, see the <a href="/en/support-resources/getting-help">Getting Help</a> page for personalized support options.
        </Info>

        Some business systems require additional connection parameters:

        * Database name or instance URL
        * Server endpoint
        * Username and password combination
        * Custom authentication tokens or credentials

        **Connection Process:**

        1. Click the **Gear Icon**.
        2. Review the connection requirements displayed in the WonkaChat credential window.
        3. Enter all required fields.
        4. Verify everything is accurate before proceeding.
        5. Click **"Connect"** to establish the connection.

        <Info>
          Each software displays its specific connection requirements in the credential window.

          If you need assistance with connection details or encounter any issues, <a href="https://www.meetwonka.com/fr/contact">contact our support team</a> for personalized help.
        </Info>

        <Warning>
          Always verify you're on the official WonkaChat page of your organization before entering credentials or connecting to any software.
        </Warning>
      </Tab>
    </Tabs>
  </Step>

  <Step title="Enable The Connection">
    After successful authentication, you need to activate the connection for the AI model you're using.

    1. Click the **"Connectors"** button under the chat input field
    2. Find your business system in the list
    3. Click on it to enable the connection for the current chat

    <Frame>
      <img alt="Outlook Mail connection example" />
    </Frame>

    <Check>
      Once enabled, the AI model can access this business system whenever you ask it to. You can enable or disable connections at any time.
    </Check>
  </Step>

  <Step title="Test Your Connection">
    Verify the connection is working by asking the AI a simple question that requires the business system.

    ### Example For Email (Outlook/Gmail):

    ```
    What emails did I receive today?
    ```

    If the AI successfully retrieves and displays information, your connection is working correctly!

    <Check>
      Connection successful! You can now use natural language to access and interact with this business system through WonkaChat.
    </Check>
  </Step>

  <Step title="Disable A Connection">
    To remove a connection:

    1. Click the **"Connectors"** button under the chat input field
    2. Find your business system in the list
    3. Click on it to disable the connection for the current chat by removing it from the selected connections

    <Warning>
      Disconnecting a business system immediately revokes WonkaChat's access to that service. However, your authentication credentials remain stored securely for future reconnections.
    </Warning>
  </Step>
</Steps>

***

## Understanding Permissions and Actions

<AccordionGroup>
  <Accordion title="What Can WonkaChat Access?" icon="eye">
    WonkaChat can only access:

    * Data you personally have permission to view in the connected business system
    * Information you explicitly ask it to retrieve
    * Resources within your organizational boundaries (if applicable)

    **WonkaChat cannot:**

    * Access data from other users unless they've shared it with you
    * Bypass your business system's permission system
    * Read or modify data without your explicit request
  </Accordion>

  <Accordion title="What Actions Can WonkaChat Perform?" icon="hand">
    WonkaChat's capabilities depend entirely on what tools are implemented for each business system. If a provider exposes an action through their connection (MCP), the AI can perform it. If an action isn't implemented in the MCP, the AI cannot access it, regardless of whether the underlying business system supports it.

    **Common capabilities include:**

    * **Read**: View emails, documents, tasks, calendar events, database records, etc.
    * **Write**: Send emails, create tasks, update records, schedule meetings, add comments
    * **Delete**: Remove items, archive records, clear data (only when you explicitly ask)
    * **Search**: Query across documents, find specific records, filter data
    * **Execute**: Run workflows, trigger automations, perform calculations

    <Check>
      Each connection (MCP) defines its own set of available actions. Check the specific business system's documentation to see what operations are supported through its MCP implementation.
    </Check>

    **Important:** The AI model is restricted to only the actions that the MCP exposes. Even if a business system's native interface supports a feature, WonkaChat can only use it if the corresponding MCP server has implemented that capability.

    All actions require your explicit instruction through conversation. WonkaChat only performs actions when you ask it to.
  </Accordion>

  <Accordion title="How Are Credentials Stored?" icon="lock">
    Your authentication tokens are:

    * Encrypted at rest in our secure database
    * Unique to your user account (never shared with others)
    * Automatically refreshed to maintain connection
    * Deleted immediately when you disconnect a business system

    WonkaChat never stores your actual passwords. It only stores the secure access tokens provided by the OAuth flow or API key you input.
  </Accordion>
</AccordionGroup>

***

## Multiple Platform Workflows

Once you've connected several business systems, you can combine them in powerful ways:

<CardGroup>
  <Card title="Cross-Platform Queries" icon="arrows-left-right">
    Ask questions that span multiple business systems in one request.

    **Example:** *"Check my Google Calendar for availability and send those times to Sarah via Outlook."*
  </Card>

  <Card title="Data Transfer" icon="arrow-right-arrow-left">
    Move information between business systems without manual copy-paste.

    **Example:** *"Take the action items from this Slack thread and create tasks in ClickUp."*
  </Card>

  <Card title="Automated Workflows" icon="gears">
    Build agents that monitor one business system and take actions in another.

    **Example:** Agent that checks Salesforce for new leads and creates follow-up tasks in Asana.
  </Card>

  <Card title="Unified Reporting" icon="chart-line">
    Combine data from multiple sources for comprehensive insights.

    **Example:** *"Summarize my week: emails from Outlook, meetings from Calendar, and completed tasks from Notion."*
  </Card>
</CardGroup>

***

## Best Practices

<AccordionGroup>
  <Accordion title="Start with Essential Connections" icon="star">
    Connect your most-used business systems first:

    1. Email (Outlook or Gmail)
    2. Calendar (Google Calendar or Cal.com)
    3. Primary project management tool
    4. Main document storage

    You can always add more connections later as you discover needs.
  </Accordion>

  <Accordion title="Test Before Complex Workflows" icon="vial">
    After connecting a business system:

    * Ask simple queries first to verify it works
    * Gradually increase complexity
    * Understand what data the business system exposes before building workflows
  </Accordion>

  <Accordion title="Review Permissions Periodically" icon="calendar-check">
    Every few months:

    * Review which business systems are connected
    * Disconnect business systems you no longer use
    * Verify connections are still active and working
  </Accordion>

  <Accordion title="Create AI Agents with Focused Tool Sets" icon="robot">
    **Why This Matters:**

    AI agents with purpose-built business system connections deliver significantly better results than general-purpose setups. When an agent has access only to relevant business systems for its specific task, it:

    * **Reduces hallucinations** by limiting context to what's actually needed
    * **Improves accuracy** by focusing on domain-specific data sources
    * **Delivers faster responses** with fewer irrelevant tool checks
    * **Maintains better context** within its area of expertise

    **Best Practice Examples:**

    <Accordion title="Sales Agent" icon="handshake">
      **Connect only:** Salesforce, HubSpot, Gmail, Calendar

      * Focused on customer data and communication
      * Won't accidentally query unrelated developer tools
      * Understands sales pipeline context deeply
    </Accordion>

    <Accordion title="Development Agent" icon="code">
      **Connect only:** GitHub, GitLab, Linear, Slack

      * Specialized in code and project management
      * Faster because it's not checking email or CRM systems
      * Better at understanding technical context
    </Accordion>

    <Accordion title="Executive Assistant Agent" icon="briefcase">
      **Connect only:** Outlook, Google Calendar, Google Drive, Notion

      * Dedicated to scheduling and document management
      * Won't accidentally access technical or sales systems
      * More reliable for personal productivity tasks
    </Accordion>

    <Accordion title="Operations Agent" icon="gears">
      **Connect only:** Odoo, QuickBooks, Airtable, Slack

      * Focused on business operations and finance
      * Specialized knowledge of operational workflows
      * Better at cross-referencing operational data
    </Accordion>

    <Tip>
      Start by creating 2-3 specialized agents for your most common workflows. You'll quickly see the quality difference compared to one agent with all tools and softwares connected.
    </Tip>
  </Accordion>
</AccordionGroup>

***

## Next Steps

<CardGroup>
  <Card title="Troubleshooting" icon="wrench" href="/en/tools-connection/troubleshooting-connections">
    Having issues with connecting your business systems? Find solutions to common connection problems.
  </Card>

  <Card title="See Examples" icon="video" href="/en/tools-connection/examples-of-tool-usage">
    Watch real examples of MCP connections in action with Outlook, Odoo, and more.
  </Card>
</CardGroup>


# Troubleshooting Connections
Source: https://help.wonka.chat/en/tools-connection/troubleshooting-connections

Quick solutions to common external connection problems and how to get help when you need it.

Having trouble connecting a business system or experiencing issues with an existing connection? This guide provides quick solutions to the most common problems.

<Info>
  Most connection issues can be resolved in a few minutes by following the steps below. If you're still stuck, see the [Getting Help](/support-resources/getting-help) page for all the ways to reach our help desk.
</Info>

***

## Quick Fixes to Try First

Before contacting support, try these simple solutions:

<Steps>
  <Step title="Disconnect and Reconnect">
    This resolves most connection issues:

    1. Go to the Connectors panel
    2. Click on the problematic business system
    3. Select **"Disconnect"**
    4. Wait 10 seconds
    5. Click **"Connect"** again and re-enter your credentials
    6. Enable the connection for your AI model
  </Step>

  <Step title="Refresh Your Page">
    Sometimes the connection just needs a refresh:

    1. Refresh your browser page (F5 or Ctrl+R).
    2. Wait for WonkaChat to reload.
    3. Try your query again.

    This can resolve temporary connection hiccups.
  </Step>

  <Step title="Clear Cache and Site Data">
    If refreshing the page doesn't work, try clearing your browser cache and site data:

    1. Open your browser's Developer Tools (press `F12` or right-click anywhere and select **Inspect**).
    2. Click the **Application** tab in the Developer Tools panel.
    3. In the left sidebar, find the **Storage** section and click **Clear site data**.
    4. Once cleared, reload the page and log in again.
    5. Reconnect your business system on WonkaChat.
    6. Test your connection to see if the issue is resolved.

    <Tip>
      After clearing site data, you may be logged out of other browser tabs for WonkaChat and will need to sign in again.
    </Tip>
  </Step>

  <Step title="Check Your Internet Connection">
    Ensure you have a stable connection:

    * Test your internet speed
    * Try loading other websites
    * Switch networks if possible (WiFi vs. mobile hotspot)
    * Restart your router if needed

    <Check>
      These troubleshooting steps often fix most connection problems by refreshing the authentication token and clearing any cached credentials that may have become stale or corrupted.
    </Check>
  </Step>
</Steps>

## Common Connection Problems

<AccordionGroup>
  <Accordion title="Authentication Failed" icon="circle-xmark">
    **Symptoms:**

    * "Authentication failed" error message
    * Can't complete the login process
    * Redirected to login page repeatedly

    **Solutions:**

    **1. Check Your Credentials**

    * Make sure you're using the correct username and password
    * Try logging into the business system directly (outside WonkaChat) to verify your credentials work
    * If you recently changed your password, update it in WonkaChat

    **2. Reconnect the Business System**

    * Disconnect the business system from the Connectors panel
    * Wait 10 seconds
    * Reconnect with your current credentials

    <Tip>
      Test your credentials on the business system's website first before trying to connect through WonkaChat.
    </Tip>
  </Accordion>

  <Accordion title="Permission Denied" icon="ban">
    **Symptoms:**

    * "Access denied" or "Insufficient permissions" error
    * AI says it cannot access certain data
    * "403 Forbidden" message

    **Why This Happens:**

    **1. Your Organization Blocks Third-Party Apps**

    Some companies restrict employees from connecting external applications for security reasons.

    **Solution:**

    * Contact your IT or Security team
    * Ask them to approve WonkaChat as a trusted third-party application
    * They may need to whitelist WonkaChat in your organization's security settings

    **Examples:**

    * Google Workspace admin settings
    * Microsoft 365 admin center
    * Company firewall or VPN restrictions

    **2. You Didn't Grant All Permissions**

    During the OAuth connection process, you may have denied some permissions that WonkaChat needs.

    **Solution:**

    * Disconnect the business system
    * Reconnect and **grant all requested permissions** when prompted
    * Some business systems let you selectively approve permissions. You need to check if WonkaChat need broader access or not.
  </Accordion>

  <Accordion title="Connection Timeout or Slow Response" icon="clock">
    **Symptoms:**

    * "Connection timed out" message
    * Requests take a very long time
    * No response from the business system

    **Quick Fixes:**

    **1. Refresh Your Page**

    * Press F5 or Ctrl+R to reload WonkaChat
    * Try your request again

    **2. Check Your Internet Connection**

    * Test your network speed
    * Try switching networks
    * Close unnecessary browser tabs

    **3. Simplify Your Request**

    * If you're asking for a lot of data, try a smaller request first
    * Instead of "all emails this year," try "emails from this week"
    * Break complex requests into smaller steps

    **4. Check the Business System's Status**

    * The business system itself might be experiencing issues
    * Visit the business system's status page (e.g., status.google.com, status.microsoft.com)
    * Wait for the service to come back online

    <Info>
      If the business system's servers are down, there's nothing WonkaChat can do until the provider fixes it.
    </Info>
  </Accordion>

  <Accordion title="Tool Not Responding" icon="triangle-exclamation">
    **Symptoms:**

    * AI says "Unable to connect to \[business system]"
    * Connection shows as active, but queries fail
    * No data returned from the business system

    **1. You Don't Have Access in the Business System Itself**

    Sometimes you're trying to do something through AI that you can't do directly in the business system either.

    <Warning>
      **Important:** WonkaChat can only do what YOU can do. If you don't have permission to access a file, read an email, or edit a record in the business system itself, WonkaChat can't do it for you through its external connection (MCP).
    </Warning>

    **Example:** If you can't access a specific Google Drive folder because you're not a member, WonkaChat can't access it either.

    **Solution:** Get the proper permissions in the business system first, then try again in WonkaChat.

    **2. Check the Business System Directly**

    * Log into the business system outside of WonkaChat
    * Verify it's working normally
    * Confirm you can access the data you're requesting

    **3. Business System API Limits**

    * Some business systems limit how many requests you can make
    * If you've made many requests in a short time, wait 15-30 minutes
    * Then try again
  </Accordion>
</AccordionGroup>

***

## Understanding Error Messages

<Info>
  WonkaChat provides detailed error messages that help us diagnose connection issues. We're currently working on user-friendly documentation for these error messages.

  **Coming Soon:** A comprehensive error code reference guide that will help you understand and resolve specific errors faster.
</Info>

For now, if you see an error message:

1. Take a screenshot of the full error
2. Note what you were trying to do when it happened
3. Contact support with this information

## When to Contact Support

If you've tried the quick fixes above and you're still having issues, our support team is here to help.

**Contact Support When:**

* Connection problems persist after trying all quick fixes
* You see error messages you don't understand
* You need help with organizational permissions or IT approval
* Multiple business systems are having issues at the same time
* You suspect a security or data issue

## Additional Resources

<CardGroup>
  <Card title="Setup Guide" icon="plug" href="/en/tools-connection/setting-up-your-first-connection">
    Learn how to properly connect business systems from the beginning to avoid common issues.
  </Card>

  <Card title="Connection Examples" icon="video" href="/en/tools-connection/examples-of-tool-usage">
    Watch examples of successful external connections (MCP) with Outlook, Odoo, and more.
  </Card>

  <Card title="FAQ" icon="question" href="/en/support-resources/faq">
    Find answers to frequently asked questions about WonkaChat connections.
  </Card>

  <Card title="Getting Help" icon="life-ring" href="/en/support-resources/getting-help">
    All the ways to reach our support team and get personalized assistance.
  </Card>
</CardGroup>


# What is WonkaChat?
Source: https://help.wonka.chat/en/welcome/introduction

Your enterprise AI assistant that connects to your business tools and runs securely in your environment.

WonkaChat is an enterprise AI assistant that connects directly to your business tools, software, and data to help you accomplish complex tasks through simple conversation while maintaining your privacy and data integrity.

## Core Concept

Wonka Chat consolidates your work across multiple software applications into a single chat interface. Instead of switching between different tools and softwares, you interact with one interface that coordinates actions across your connected systems.

Think of it as your intelligent business assistant that:

* Understands your company's processes and data
* Connects to all your existing softwares and tools
* Executes multi-step workflows across different platforms
* Keeps your data secure within your environment

<Frame>
  <img alt="Onboarding Agent orchestrating tasks across Notion, Outlook, Odoo, and other business tools" />
</Frame>

In this example, a request to "Run New Employee Onboarding" automatically:

1. ✅ Gets information from **Notion** (onboarding checklist)
2. ✅ Creates email account in **Microsoft 365**
3. ✅ Sends access codes via **Outlook**
4. ✅ Schedules onboarding meeting in **Outlook Calendar**
5. ✅ Creates **Odoo** account and sends invitation
6. ✅ Sends onboarding presentation via email

<Check>
  This is the power of WonkaChat: complex multi-step workflows across different softwares, managed through natural conversation.
</Check>

In this example, a simple request triggers actions across Notion, Outlook, Odoo, and other tools, all coordinated through one platform and one conversation.

***

## Key Capabilities

<CardGroup>
  <Card title="AI Agents" icon="robot">
    Create specialized agents for specific tasks like onboarding, reporting, or customer support that run autonomously on your behalf.
  </Card>

  <Card title="External Connections" icon="plug">
    Connect your existing softwares and business tools (from PM to CRM to communication platforms) so WonkaChat can work with your data.
  </Card>

  <Card title="Model Flexibility" icon="sliders">
    Choose between thinking and fast/non-thinking models based on task complexity, and choose any AI model from any provider.
  </Card>

  <Card title="Business Teaching" icon="graduation-cap">
    Use your company's knowledge by uploading documents and providing context. The AI will adapt to your processes, terminology, and workflows.
  </Card>
</CardGroup>

***

## What Can You Accomplish?

Real examples of what teams are doing with WonkaChat:

<AccordionGroup>
  <Accordion title="Employee Onboarding in Minutes" icon="user-plus">
    Automatically create accounts, schedule meetings, send welcome materials, and set up access across all your systems with a single request.

    **Example:** "Onboard John Smith starting Monday" → Creates email, schedules HR meeting, adds to Slack, creates CRM account, sends laptop request.
  </Accordion>

  <Accordion title="Intelligent Document Search" icon="magnifying-glass">
    Find information across all your connected systems (emails, documents, databases, wikis) without remembering where things are stored.

    **Example:** "Find the Q3 marketing budget proposal" → Searches Google Drive, SharePoint, Notion, and Confluence simultaneously.
  </Accordion>

  <Accordion title="Automated Reporting" icon="chart-line">
    Generate reports by pulling data from multiple sources, analyzing trends, and formatting results with no manual spreadsheet work.

    **Example:** "Create a sales performance report for last month" → Pulls CRM data, calculates metrics, generates formatted report.
  </Accordion>

  <Accordion title="Cross-Platform Workflows" icon="arrows-spin">
    Execute complex multi-step processes that span different tools, eliminating manual coordination and context switching.

    **Example:** "Create project in PM tool, notify team in Slack, send welcome email, schedule kickoff" → All happens automatically.
  </Accordion>

  <Accordion title="Meeting Intelligence" icon="calendar">
    Schedule meetings considering everyone's availability, send agendas from templates, capture notes, and distribute action items, all in one flow.

    **Example:** "Schedule quarterly review with leadership team next week" → Finds time slots, sends invites with agenda, creates notes doc.
  </Accordion>

  <Accordion title="Customer Support Automation" icon="headset">
    Answer support questions using your knowledge base, create tickets, escalate when needed, and update customers to reduce response time.

    **Example:** "Customer asking about refund policy" → Retrieves policy, drafts response, creates ticket if needed, logs interaction.
  </Accordion>
</AccordionGroup>

<Tip>
  These workflows run through simple conversation with no coding or complex configuration required.
</Tip>

***

## Ready to Get Started?

<CardGroup>
  <Card title="Why Choose Wonka Chat?" icon="star" href="/en/en/welcome/why-wonka-chat">
    Discover the unique advantages and security benefits of using WonkaChat for your team.
  </Card>

  <Card title="Quick Start Guide" icon="rocket" href="/en/en/welcome/quick-start">
    Get up and running with your first conversation in just 5 minutes.
  </Card>
</CardGroup>


# Quick Start Guide
Source: https://help.wonka.chat/en/welcome/quick-start

Get started with WonkaChat in 5 minutes, from first login to your first AI-powered workflow.

Get up and running with WonkaChat quickly and start experiencing AI-powered automation across your business tools.

## Your 5-Minute Success Path

<Steps>
  <Step title="First Login and Account Access">
    Access WonkaChat through one of four methods:

    <Tabs>
      <Tab title="Organization Invitation">
        If you received an invitation to join your organization's WonkaChat:

        1. Check your email for the WonkaChat invitation
        2. Click the invitation link
        3. Create your account using the provided setup flow
        4. Complete your profile
        5. Start using WonkaChat with your team's configuration
      </Tab>

      <Tab title="Team SSO Login">
        If your organization already has WonkaChat set up, use your work credentials:

        **Sign in with:**

        * **Google Workspace** (your company Google account)
        * **Microsoft 365** (your company Microsoft account)
        * **Other SSO providers** your organization configured

        <Check>
          No new password needed, just use your existing work account to access WonkaChat instantly.
        </Check>
      </Tab>

      <Tab title="Individual Subscription">
        Create your personal WonkaChat account and subscribe directly:

        1. Visit [wonka.chat/register](https://wonka.chat/register)
        2. Enter your email address and create a password
        3. Accept the Terms and Conditions
        4. Click **Continue** to create your account
        5. Complete payment to activate your account

        <Check>
          Your subscription is managed through Stripe with secure payment processing.
        </Check>
      </Tab>

      <Tab title="Enterprise Setup">
        If you're setting up WonkaChat for your entire organization:

        1. Contact our team through the [Getting Help](/support-resources/getting-help) page
        2. Discuss your organization's requirements and user count
        3. Receive your custom enterprise invitation
        4. Configure SSO, governance policies, and team settings
        5. Invite your team members with pre-configured access

        <Tip>
          Enterprise plans include dedicated support, custom integrations, advanced security features, and volume pricing.
        </Tip>
      </Tab>
    </Tabs>

    **What you'll see after login:**

    * Clean chat interface in the center
    * Connected tools and AI Agents in the sidebar
    * Model selection options in the settings
    * Your prompt library and saved conversations
  </Step>

  <Step title="Quick Navigation Tour">
    <Info>
      **Watch the WonkaChat Dashboard Walkthrough**

      <iframe title="WonkaChat Dashboard Walkthrough" />

      In this video, you'll learn how to navigate the WonkaChat interface, access essential features, and customize your workspace for maximum productivity.
    </Info>
  </Step>

  <Step title="Test Your First Message">
    Let's make sure everything works with a simple test:

    ### Send a Basic Message

    Type something simple in the chat:

    ```
    Can you say "Hello World" please?
    ```

    WonkaChat will respond, confirming the platform is working correctly.

    <Check>
      If you receive a response, you're all set! The AI is connected and ready to help.
    </Check>
  </Step>

  <Step title="Connect Your First Tool (Outlook Mail)">
    Now let's connect WonkaChat to your email account so it can access your emails and calendar.

    <Warning>
      This example uses **Outlook Mail**, but the process is identical for **Gmail**, or any other email system. Simply click the corresponding the right connection button for your email provider and follow the same steps.
    </Warning>

    ### Step-by-Step Connection

    <Steps>
      <Step title="Open the Connectors Settings">
        On the bottom left of your WonkaChat interface, locate the settings and open the **Connectors** button.

        <Frame>
          <img alt="Outlook Mail button in the right panel" />
        </Frame>

        <Info>
          Learn about external connections in [Tools & Connection](/tools-connection/introduction).
        </Info>
      </Step>

      <Step title="Add Your Credentials">
        A connection window will appear asking for your credentials. Enter the required information:

        * Your Outlook email address
        * Authentication credentials (this will use your organization's authentication method)

        <Info>
          WonkaChat uses secure authentication and never stores your password. All credentials are encrypted.
        </Info>
      </Step>

      <Step title="Enable Connection for Your Model">
        After connecting, you need to authorize the external connection for the AI model you're using.

        <Frame>
          <img alt="Enable the connection for the AI model" />
        </Frame>

        Click **"Connectors"** and select which connection you want the AI model to use.

        <Check>
          Once the Outlook Mail connection is enabled, the AI can read your emails, access your calendar, send messages on your behalf, and handle any Outlook-related tasks through natural conversation.
        </Check>
      </Step>
    </Steps>
  </Step>

  <Step title="Your First Connected Conversation">
    Now that Outlook Mail is connected, let's use it in a conversation:

    ### Send a message to the connected AI model

    Type something like:

    ```
    Send an email to Sarah letting her know I'll be 5 minutes late to the meeting.
    ```

    <Warning>
      The AI will only interact with Outlook or send messages when you explicitly request it. It always follows your permissions and never takes action without your direction.
    </Warning>

    WonkaChat will:

    * Access your Outlook account through the configured connection
    * Draft an email to Sarah with your update
    * Send directly the mail or present it for review before sending (depends on your prompt and on the model used)

    <Check>
      You've successfully connected a tool and used it through natural conversation!
    </Check>
  </Step>

  <Step title="Understanding Model Selection">
    Choose the right model for your task:

    <CardGroup>
      <Card title="Pro Models" icon="brain">
        **Use for complex tasks:**

        * Strategic analysis
        * Multi-step problem solving
        * Data analysis and insights
        * Complex workflow planning
        * External tool connections

        **Example:** "Analyze our Q3 sales data and recommend improvements"
      </Card>

      <Card title="Fast Models" icon="bolt">
        **Use for quick tasks:**

        * Simple questions
        * Fast information retrieval
        * Basic formatting
        * Quick summaries
        * High-volume operations

        **Example:** "What's my next meeting?"
      </Card>
    </CardGroup>

    <Tip>
      When in doubt, start with pro models for important tasks. You can always switch to fast models for speed and cost once you understand the query complexity.
    </Tip>
  </Step>

  <Step title="Quick Tips for Success">
    <CardGroup>
      <Card title="Be Specific" icon="crosshairs">
        The more context you provide, the better WonkaChat can help. Include relevant details, timeframes, and desired outcomes.
      </Card>

      <Card title="Start Simple" icon="seedling">
        Begin with straightforward queries and gradually work up to complex multi-step workflows as you learn the platform.
      </Card>
    </CardGroup>
  </Step>
</Steps>

***

## Next Steps: Go Deeper

Now that you're familiar with the basics, explore more advanced capabilities:

<CardGroup>
  <Card title="Create AI Agents" icon="robot" href="/en/ai-agents/introduction">
    Build specialized agents focused on specific tasks for more accurate, reliable results with reduced hallucinations.
  </Card>

  <Card title="Understand External Connections" icon="plug" href="/en/tools-connection/introduction">
    Learn how external connections (MCP) work and scale from connecting a single tool to integrating your entire business ecosystem.
  </Card>

  <Card title="Improve Your AI Usage" icon="wand-magic-sparkles" href="/en/improved-usage/asking-great-questions">
    Discover techniques for asking better questions, managing conversations effectively, and getting more value from every AI interaction.
  </Card>

  <Card title="Security & Governance" icon="shield-check" href="/en/security-governance/index">
    Learn how to use AI securely, maintain data privacy, and implement governance policies that protect your organization while maximizing AI capabilities.
  </Card>
</CardGroup>


# Why Choose WonkaChat?
Source: https://help.wonka.chat/en/welcome/why-wonka-chat

Superior security, true flexibility, and seamless integration with your softwares and business tools.

WonkaChat goes beyond traditional AI assistants by offering enterprise-grade security, genuine tool and external software integration, and the flexibility to adapt to your specific business needs.

<Tabs>
  <Tab title="Getting Started">
    ## Unique Value Propositions

    <CardGroup>
      <Card title="External Softwares Integration" icon="puzzle-piece">
        Unlike copy-paste workflows with other AI tools, WonkaChat directly connects and acts within your ecosystem.
      </Card>

      <Card title="Choose Your AI Models" icon="toggle-on">
        You are not locked into a single provider and can choose any AI model that fits your task.
      </Card>
    </CardGroup>

    ## Security and Governance Advantages

    **Public AI platforms** like ChatGPT store your data on their servers, may use it for training, and give you limited control over security and access.

    **WonkaChat operates on a different foundation:**

    <CardGroup>
      <Card title="Built-In Security" icon="shield-check">
        You receive a completely isolated environment with encrypted communications and a robust, enterprise-ready security architecture designed to protect your data at every layer.
      </Card>

      <Card title="Team AI Training" icon="certificate">
        We offer comprehensive training to help your team use AI securely, follow best practices, and maintain compliance.

        To learn more or request a customized training session, [contact us](/en/support-resources/getting-help).
      </Card>

      <Card title="No Data Collection" icon="ban">
        Your conversations are never collected, stored by AI providers, or used to train models. Everything stays private.
      </Card>

      <Card title="Access & Control" icon="users-gear">
        Comprehensive access controls and configurable permissions to control user actions, feature availability, and agent usage.
      </Card>
    </CardGroup>

    For more details, see the [Security & Governance](/en/security-governance/security-overview) section.

    ***

    ## Flexibility and Customization Benefits

    ### Model Selection for Every Task

    WonkaChat lets you choose the right AI model for each situation:

    <Tabs>
      <Tab title="Pro/Thinking Models">
        **Best for:** Complex analysis, multi-step reasoning, strategic planning, and external tool connections

        **Available Models:**

        * **Claude Opus 4.5** - Highest capability for complex reasoning and coding
        * **Claude Sonnet 4.5** - Balanced performance for most advanced tasks
        * **GPT-5.2** - Latest OpenAI model with advanced capabilities
        * **GPT-5** - Strong reasoning and agentic workflows
        * **Gemini 2.5 Pro** - Massive 1M context window for long documents

        **Use when you need:**

        * Deep analysis of business data
        * Complex problem-solving with external software integration
        * Strategic recommendations
        * Multi-step workflow planning
        * Long document analysis

        **Example tasks:**

        * "Analyze our Q3 sales data and recommend territory adjustments"
        * "Create a comprehensive onboarding plan for new employees"
        * "Review this contract and identify potential risks"
      </Tab>

      <Tab title="Fast Models">
        **Best for:** Quick answers, simple tasks, fast responses, and high-volume operations

        **Available Models:**

        * **GPT-4o mini** - Most cost-effective for high-volume tasks
        * **Gemini 2.5 Flash** - Fast with large 1M context window
        * **Claude Haiku 4.5** - Fastest Claude model with near-frontier intelligence
        * **Gemini 3 Flash** - Next-generation speed and large 1M context window

        **Use when you need:**

        * Rapid information retrieval
        * Simple formatting or editing
        * Straightforward questions
        * Quick summaries
        * High-volume operations

        **Example tasks:**

        * "What's on my calendar today?"
        * "Send a meeting confirmation to the team"
        * "Summarize this email thread"

        <Warning>
          While fast models are excellent for quick and straightforward tasks, they may not handle complex queries requiring deep reasoning or extensive external software connections.

          Consider the task requirements carefully when choosing your model.
        </Warning>
      </Tab>
    </Tabs>

    <Info>
      Choose pro models for difficult tasks requiring deep analysis or external connections, and fast models for quick, straightforward queries or high-volume operations.
    </Info>

    ***

    ## Who Benefits Most from WonkaChat?

    <CardGroup>
      <Card title="Multi-Platform Organizations" icon="sitemap">
        Companies using diverse tech stacks who want to unify AI operations across all their systems.
      </Card>

      <Card title="Security-Conscious Teams" icon="user-shield">
        Organizations that cannot or will not send sensitive data to public AI platforms.
      </Card>

      <Card title="Custom Workflow Needs" icon="diagram-project">
        Businesses with unique processes that don't fit into standard SaaS AI tools.
      </Card>

      <Card title="Regulated Industries" icon="scale-balanced">
        Healthcare, finance, legal, and government organizations that need AI capabilities with strict data governance.
      </Card>
    </CardGroup>

    <Tip>
      WonkaChat is ideal when you are integrating AI deeply with your existing tools, concerned about data security, and maintaining flexibility in your AI strategy.
    </Tip>
  </Tab>

  <Tab title="Technical Details">
    ## External Tool/Software/Data-Source Connection via MCP

    ### Model Context Protocol (MCP)

    WonkaChat uses **MCP (Model Context Protocol)** to connect AI models to your business tools and software systems.

    <Info>
      **What is MCP?** MCP is an open protocol that enables secure, standardized connections between AI models and external tools. It allows WonkaChat to interact with your emails, calendars, databases, CRMs, and other business applications while maintaining security and permission boundaries.
    </Info>

    **Key MCP Capabilities:**

    * Direct integration with business systems (Outlook, Gmail, Salesforce, Notion, etc.)
    * Secure credential management and authentication
    * Real-time data access and write operations
    * Permission inheritance from your existing user access
    * Audit logging for all tool interactions

    <Tip>
      For comprehensive information about MCP architecture, supported integrations, connection setup, and advanced configuration options, visit the [Tools & Connection](/en/tools-connection/introduction) section.
    </Tip>

    ***

    ## Model Provider Flexibility

    ### Multi-Provider Architecture

    WonkaChat's architecture is designed to be **model-agnostic**, allowing you to:

    <Tabs>
      <Tab title="Provider Options">
        **Supported AI Providers:**

        * **OpenAI**: GPT-5.2, GPT-5, GPT-4o, GPT-4o mini
        * **Anthropic (Claude)**: Opus 4.5, Sonnet 4.5, Sonnet 4, Haiku 4.5, Sonnet 3.7
        * **Google Gemini**: 2.5 Pro, 2.5 Flash, 3 Flash Preview, 3 Pro Preview
        * **Mistral**: Large Latest, Medium Latest
        * Multiple providers available across all major AI companies
      </Tab>

      <Tab title="Selection Criteria">
        **Factors to consider when choosing models:**

        1. **Task Complexity**
           * Pro models: Complex reasoning, analysis, tool usage, multi-step workflows
           * Fast models: Quick queries, simple tasks, high-volume operations

        2. **Cost Optimization**
           * Token pricing per provider
           * Response latency requirements
           * Quality vs. cost trade-offs

        3. **Performance Characteristics**
           * Context window size (128K to 1M+ tokens)
           * Token limits and output capacity
           * Response generation speed
      </Tab>

      <Tab title="Switching Strategy">
        **Benefits of multi-provider support:**

        * **Avoid Vendor Lock-In**: Switch providers without changing workflows
        * **Cost Management**: Use cheaper models for simple tasks, pro models for complex ones
        * **Performance Optimization**: Route tasks to the fastest or most accurate model
        * **Redundancy**: Failover to alternative providers if one is unavailable
        * **A/B Testing**: Compare model performance for your specific use cases
      </Tab>
    </Tabs>

    <Tip>
      Model selection can be configured per organization, per team, or per individual user based on your governance policies.
    </Tip>
  </Tab>
</Tabs>