AI teammates, each with a job.
Your workspace starts with an AI teammate, and an admin can add more: a researcher, a reviewer, someone for release notes. Each has its own name, brief, model and tools, and works in the same chat, tasks and docs as everyone else.
The next morning
StandupAI
Step three shipped. Launch plan updated. Retention watch ran at 9: no drop detected.
Nadia
@WeMachines every morning, go through our tasks, keep them up to date and challenge any blockers.
WeMachines AI
Done. Task sweep runs every morning at 9: it updates each task from its commits and comments, and asks the owner about anything that looks blocked.
Take a look around.
Click around the app. Illustrative demo data.How it works
-
Add a teammate
In Settings, give it a name, a brief and a model.
-
Set how it works
Its length, thinking, tone and pushback, and whether it cites sources, searches the web or stays read-only.
-
Mention it
It answers in chat, tasks and pages, and asks you back with a few options rather than guess.
A team of specialists.Sharing one workspace.
A primary agent
One teammate runs ⌘O, call summaries and conflict checks, and any routine that names no one else.
Its own tools
Narrow an agent to a set of tools, give it skills, or add an MCP server just for it.
Read-only when you want
An agent in read-only mode can look anywhere and change nothing.
It sees images
Screenshots and photos shared in chat, tasks and pages are part of what it reads.
Joins in, or doesn’t
Set how readily it speaks up without being mentioned.
Reminders
Ask it to remind you on Friday at 3. The reminder is yours alone.
Set it the way the job needs.
A researcher can be thorough, think first, write prose and cite its sources, and still be barred from changing anything. Each setting is one choice on a scale, so the ends explain the middle.
Researcher Settings → Agents
Good to know
- Agents work inside your workspace. They don’t join calls or shared channels, and they have no calendar or mailbox of their own.