Brain
Brain settings
Owners and admins decide what kinds of page the brain keeps, what detail it captures, whether new pages wait for review, and whether every source is kept as its own page.
How the brain behaves is configured once, by workspace owners and admins. Everyone else teaches, browses and asks; nobody else changes the rules.
Categories
The kinds of page your brain keeps. It ships with people, companies, concepts, notes and sources, and you can:
- Switch one off if it does not apply to how you work
- Add your own — Meeting, Client, Product, Incident — with a description and an icon
- Remove a category you added
The description is not decoration: it tells the AI what belongs in that category when it proposes pages. “Client — an organisation we have a signed contract with, not a prospect” produces noticeably better proposals than “Client”.
Switch every category off and teaching will read your material and save nothing. The settings page warns you when you have done that.
Details to capture
| Setting | What it adds |
|---|---|
| Subtitles | A one-line descriptor on new pages — a person’s role, a company’s stage |
| Open questions | Questions a source raises but does not answer, surfaced as gaps |
Open questions are the setting most teams underuse. They turn “this document was vague about approval limits” from a thing one reader noticed into a tracked gap anyone can close.
Review
Require review before saving puts new pages from teaching into the review queue until an owner or admin approves them. With it off, teaching writes pages straight into the brain.
Turn it on when more than a couple of people teach the brain, or as soon as answers from it are being used for anything that matters. See Reviewing what it proposes.
Provenance
Keep source pages saves each taught file, link or note as its own page, linking everything it produced. It costs a page per teach run and buys you the ability to answer “where did this come from?” months later.
Leave it on unless your brain is being taught constantly from throwaway material.
Deciding as a team
These four settings encode a judgement about how much your organisation trusts automation with its own facts. There is no universally right answer, but there is a pattern:
| Stage | Sensible settings |
|---|---|
| Getting started, small team | Review off, sources on, both details on |
| Several teachers, real use | Review on, sources on |
| Large workspace, brain used by staff and apps | Review on, curated category list, both details on |
Revisit them once the brain is actually in use. The setting that made sense when five pages existed is rarely the right one at five hundred.