Skip to content

AI Agent documentation

Set up the workspace where your agents plan, run, and report on growth work.

AI Agent brings agents, workflows, connected tools, knowledge, and supervised work into one workspace. Start by completing onboarding and testing your primary agent. Add a workflow when the same process should run again, and use a project when an initiative needs its own board and progress view.

Start here

Choose the path that matches what you want to accomplish first.

For a complete introduction, finish onboarding, take the workspace tour, and follow the weekly growth review tutorial.

How the pieces fit

  1. Your workspace holds shared agents, connections, knowledge, and operating surfaces.
  2. An agent brings instructions, skills, and permitted connection tools to a task.
  3. A workflow defines a repeatable sequence you can test and run.
  4. An autopilot keeps proven work operating from a configured trigger.
  5. The Board, Inbox, approvals, and runs show what needs attention and what happened.
  6. An optional project groups tickets for a focused initiative.

Explore the concepts in more detail:

Keep control

Begin with read-only connection tools where possible. Add actions only when the agent needs them, and keep approval steps for changes such as publishing, posting, or updating records. Learn how tool access and approvals work together before expanding an automation’s permissions.

Navigation