Skip to main content
Create a specialized AI agent tailored to your workflow. This guide walks through the guided creation flow, from your first idea to a shared, working agent.
Building an agent means defining its identity, writing instructions, choosing a model size, and optionally connecting tools.

​
Starting an Agent

1

Open New Agent

In the sidebar, open Agents → New agent. You land in a guided creation flow, in the main conversation area (there’s no separate side panel to look for).
2

Start from Scratch or a Template

Choose to start blank, or click Browse templates to start from a ready-made template and customize it. Templates are a fast way to see a working example before writing your own instructions from scratch.

​
The Guided Creation Wizard

The wizard walks you through five steps:
1

Identity

Choose a clear, descriptive name and a short description.Good: “Customer Support Assistant,” “Q4 Sales Data Analyzer” Avoid: “Helper,” “Agent 1”
If you plan to share the agent with your organization, pick a name that makes sense to everyone, not just to you.
2

Model

Pick a model size for your agent. WonkaChat groups models as fast, standard, or advanced rather than surfacing every provider by name, and each model shows a region badge indicating where it runs.
  • Fast models suit simple, high-volume tasks (data lookup, formatting, quick replies)
  • Standard models are a good default for most agents
  • Advanced models suit complex, multi-step reasoning, coding, or planning
For a full comparison of model sizes and when to use each, see Choosing a Model.
3

Instructions

Write your agent’s instructions: its role, tone, operational rules, and any business context it needs.
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.”
Define tone and format.Example: “Communicate in a friendly, professional tone. Be concise but thorough. Use bullet points for clarity.”
Provide specific rules for how the agent should execute tasks and make decisions.Example: “When analyzing sales data, always compare against the previous quarter. If data is incomplete, note the gaps explicitly rather than making assumptions.”
Include company-specific information.Example: “Support hours: 9 AM-6 PM. We offer three tiers: Basic, Pro, Enterprise.”
Clear, detailed instructions are the foundation of agent effectiveness. See Business Teaching for more on this.
4

Capabilities

Turn on any capabilities your agent needs:
  • Code Execution — if enabled for your organization
  • Web Search — if enabled for your organization, toggle Allow searching the web for this agent
  • Connectors — add the connectors (MCP) the agent should use
Agents only access connectors explicitly assigned to them, and inherit your personal permissions on those systems. See Access Control for details.
Tool selection principles:
  • Only add connectors the agent actually needs
  • Too many connectors can cause selection confusion
  • Make sure you personally have permission for any connector you assign
5

Review

Review your configuration, test it with a few real requests, and create the agent.

​
After Creation: Audit and Share

Once the agent is created, the flow moves into two more stages.
1

Audit

Review the agent’s activity and configuration before putting it in front of others. This is a good time to test edge cases and confirm the agent behaves as expected.
See Audit Your Agents for details.
2

Share

Decide who else should be able to use the agent.
  • Personal use: keep it private, it’s ready to use immediately
  • Share with specific people or groups: as Viewer, Editor, or Owner
  • Share with everyone in your organization: makes the agent available to every member
Whether you can share agents — and with everyone in your organization — depends on your role’s permissions and your organization’s settings. If sharing options are missing, ask your administrator.See Sharing Agents for the full walkthrough.

​
Best Practices

Be Explicit

Define specific goals, actions, and expected results to reduce ambiguity.

Include Edge Cases

Anticipate unusual situations and provide handling guidance.

Provide Context

Give agents background information about your business processes and terminology.

Iterate

Start simple, test thoroughly, refine based on real-world use.

​
Common Mistakes

Problem: “You are a helpful assistant” lacks guidance. Solution: Be specific about role, expertise, tone, and guidelines.
Problem: One agent handling unrelated tasks like customer support and data analysis. Solution: Create focused agents for specific workflows.
Problem: Assigning every available connector. Solution: Only add the connectors the agent needs, to avoid selection confusion.
Problem: Sharing an agent without thorough testing. Solution: Test extensively with realistic scenarios during the Audit stage first.

​
Security Best Practices

​
Limit Connector Access

Only grant the connectors that are strictly necessary for the agent’s task. If your agent drafts emails, consider leaving out a “send email” capability so you keep final control over sending.

​
Isolate Integrations

Avoid connecting systems that shouldn’t interact with each other in 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. Recommended approach:
  • Start with read-only access when possible
  • Test with non-production data first
  • Gradually add connectors as you verify behavior
See Security & Governance for comprehensive security practices.