Creating your first agent

AI agents are configurable AI assistants with custom behaviour, tools, and knowledge bases. Create agents to automate tasks, answer questions, and interact with your data.

Navigate to AI Agents

In the dashboard sidebar, click AI Agents, then click Create Agent to open the creation wizard.

Creation wizard

The agent creation wizard walks you through five steps.

Step 1: Basic info

Enter a Name and Description for your agent. Optionally assign it to a Group to keep your agents organised. Built-in groups include development, compliance, content, and analytics, or you can enter a custom group name.

Step 2: Model and behaviour

Select an AI model for the agent. Only models that support tool use are shown in the list.

Set the Temperature to control how creative or deterministic the agent’s responses are:

Value	Behaviour
0	Precise, deterministic
0.5	Balanced
1	Creative, varied

Optionally set a Max tokens limit (1—200,000) to cap the length of the agent’s responses. Leave this empty to use the model’s default.

Step 3: System prompt

Write a system prompt that defines the agent’s personality, expertise, and boundaries. The system prompt is the most important factor in shaping agent behaviour.

Choose from a built-in template to get started quickly:

• Code Reviewer — analyses code for quality, security, and best practices
• Content Generator — creates and edits written content
• Data Analyst — interprets data and produces insights
• Compliance Checker — reviews content against regulatory requirements

Or write your own system prompt from scratch.

Tip

Start with a template system prompt and customise it for your use case. Templates provide a solid foundation for common agent roles.

Step 4: Tools and knowledge

Assign capabilities to your agent:

• Built-in tools — tools provided by Quant AI (e.g. web search, content retrieval)
• Custom tools — your own edge functions registered as callable tools
• Skills — reusable instruction sets grouped by namespace
• Vector database collections — knowledge bases for retrieval-augmented generation (RAG)

You can add or change these after creating the agent.

Step 5: Review

Review all settings and click Save to create the agent. You can edit any setting later from the agent detail page.

Next steps

• Agent configuration — detailed reference for all agent settings
• Testing agents — try out your agent using the built-in chat interface