What your AI knows: the Brain
The Brain is everything your team has read. It is the difference between an assistant that sounds plausible and one that knows your business, and it is the first place to look when an answer comes back wrong.
Three tabs#
| Tab | What it is for |
|---|---|
| Brain | The shape of what your team knows, and what it has learned recently |
| Sources | Every document, file and connected system feeding it |
| Answers | Questions already asked and answered, kept so the work is not repeated |
Asking a question#
Ask in plain language. “What did we agree with the supplier about lead times?”
Every answer cites the sources it came from. An answer without citations is one to distrust.
Citations open the underlying source at the passage used, so you can check the answer rather than take it.
Answers are kept in the Answers tab, so the next person asking gets the work already done.
Citations are the whole point. If you take one thing from this page: an answer you cannot trace to a source is a draft, not a fact.
Keeping it current#
Sources go stale, and a stale source produces a confidently wrong answer. The Sources tab shows what is connected and when it was last read. Adding a document, connecting a system, or removing something out of date all happen here.
Your team reads what is in the Brain and nothing else. That is a constraint worth understanding: it is why the answers stay on-topic, and why a missing document produces a gap rather than a guess.
If something goes wrong#
- An answer is wrong. Check the citations first. Wrong source means a Sources problem; right source misread means a feedback problem, and the answer has a feedback control.
- An answer is missing information you know exists. The document is probably not a source yet. Add it in the Sources tab.
- A source shows as failed. Some formats and some connected systems fail on specific files. The Sources tab names which file and why.
- The Brain looks empty. A new workspace starts empty. It fills during onboarding. See Adding sources.