Add Ezra to Slack.
The next screen is Slack's own consent page. Here's the whole story before you click.
What happens: you approve Slack's permission list. We start an isolated Ezra for your workspace — its own apps, its own database. It joins as @Ezra, DMs you once, and starts reading the channels you've invited it into.
Ezra can
- ✓read channels it's invited to
- ✓reply in threads and DMs
- ✓read documents you drop
- ✓draft work for your approval
Ezra can't
- ✕see channels it isn't in — Slack enforces that, not our promises
- ✕send anything outside your workspace — you send
- ✕act on any workspace but yours
One isolated instance per workspace. No pooling across companies.
Invitation is the boundary. Ezra reads only what it's invited to.
Nothing gets sent without your click. Ezra drafts; a person sends, and every action is logged.
Change your mind anytime — uninstall Ezra from Slack; that revokes its access at once, and we remove your deployment.
Ezra uses a large language model and can produce inaccurate answers or summaries; read what it drafts before you use it.
Invitation is the boundary. Ezra reads only what it's invited to.
Nothing gets sent without your click. Ezra drafts; a person sends, and every action is logged.
Change your mind anytime — uninstall Ezra from Slack; that revokes its access at once, and we remove your deployment.
Slack lists every permission Ezra requests — the list is short, and it never includes admin rights or acting as you. About 30 seconds.