Build your first agent

This walkthrough takes about fifteen minutes and assumes you have a workspace and at least one connected source.

1. Connect something to read

An agent with no context can only guess. Connect a source from Data, then wait for the first sync to finish. A few hundred records is enough to see useful behaviour.

2. Describe the outcome

Write what the agent is for in plain language: who asks it things, what a good answer looks like, and what it must never do. This description is the agent's brief, so be specific.

3. Give it tools

Tools are the actions the agent may take. Start with read access only. You can add tools that write or send later, once you trust the behaviour.

4. Run and read the trace

Every run leaves a trace showing what the agent read, which tools it called, and why it stopped. Read the trace before you change the brief. Most early problems are missing context rather than wrong instructions.