Skip to main content
Business plan and above. Off by default — enable under Settings → Modules.
The assistant works over your real workspace data — projects, tasks, time — and can act on it when you ask.

​
What it does

Command agent

Ask in plain language: find work, create tasks, log time, update status.

Morning brief

What changed, what’s at risk, what needs you today.

Standup drafts

A draft update from what actually moved.

In-thread answers

Mention the assistant in a task comment for an answer in place.

​
It can only do what you can

The assistant runs as you. It sees the projects you can see and can only make changes you’re permitted to make — a viewer asking it to create a task gets refused, exactly as they would in the UI.

​
Confirmation before changes

In the app, actions that change data show you what will happen before they run.
Over a connected AI host (MCP Server) there is no separate confirmation step — a write tool executes when called. Grant read-only access to any connection you don’t want making changes.

​
Governance

Admins can set AI policy per workspace, and per project where needed, under Settings → AI — including which features are available and how credits are used.