Suggestions from channel messages
Noots reads new channel messages and offers to update the board. What that means, and how to stop it.
Somebody posts *“done — created the legal@ mailbox”* in a project channel, and a small card appears under it: Tick “Create the legal inbox” on Launch checklist? One click and the checklist item is ticked. That is this feature.
What it can offer
- Tick a checklist item that the message says is finished.
- Move a task to another column — into review, or into done.
- Add a task, in a column it names.
- File a note on the project, when nothing more specific fits.
Every suggestion is grounded in that project’s real open work — it can only point at a task or a checklist item that actually exists, and the target is re-checked against the database before anything happens. At most two suggestions per message, and “nothing to suggest” is the usual answer.
A suggestion is a suggestion. Noots never moves a task, ticks an item or creates anything on its own — somebody has to press Add. Accepting one goes through the same permission checks as doing it by hand, so it can never do something you could not.
What leaves your workspace, and when
To make the suggestion, two things are sent to our AI provider: the text of the message, and enough of the project to ground it — the names of the candidate projects, their column labels, the titles of their open tasks and the text of unticked checklist items. The second half is why a suggestion can name a real task instead of inventing one, and it is the part worth understanding before you decide whether to leave this on.
- Which messages
- Channel messages only.
- Direct messages
- Never scanned.
- Group chats
- Never scanned.
- Slash commands
- Never scanned — anything starting with / is skipped.
- Attachments & files
- Never sent.
- Message bodies of OTHER messages
- Never sent — only the one just posted.
- On or off
- On by default, for the whole workspace.
- Who can change it
- Owners and admins, in Org settings → AI.
The switch is checked on the server before anything is read or sent — it is not a hidden button. Switch it off and message text stops leaving the workspace immediately. The rest of chat is unaffected: mentions, search, previews, calls and canvases all work exactly as before.
The card itself
- Add does the thing it describes and says what it did.
- Dismiss removes it for everyone.
- Carry on typing and it tucks itself away into a small icon beside the message that prompted it — click that to bring it back. It is not lost, and it stops sitting in the way of a conversation.
See also Which AI does what for how this differs from Autopilot, Nooties and Hoot.