The Brain is what Nisa knows about your company.
It is not a chat history and it is not a model that was trained on you. It is a set of files you can open, read, correct and delete — which is the whole point. When Nisa tells you something, you can ask where she got it, and there is an answer.
Packs
A pack is one folder in the Brain, with one subject.
| pack | holds |
|---|---|
company-brain | what your company knows — people, projects, decisions, numbers |
worker-log | what your Skills reported, so you can ask about this morning's brief this afternoon |
sales | your leads and deals, when you keep them here rather than in a CRM |
hiring | open roles and where each one stands |
operator-preferences | how you have told Nisa to work — tone, length, defaults |
operator-voice | how you write, so drafts under your name sound like you |
You will see the word "pack" in a few places in the product. It means folder, and the reason it has its own name is that a Skill declares which packs it may read — so a Skill that summarises your inbox never sees your sales pipeline.
Why files and not a database
Three reasons, and all three are things customers have asked for:
You can correct it. A wrong fact is a line in a file. Open it, fix it, done.
You can delete it. When a person asks to be forgotten, there is a specific thing to remove — not "somewhere in the weights".
You can take it with you. The files are markdown. If you leave, you leave with your knowledge.
What goes in, and what does not
Durable things go in: a decision and why it was made, a metric and where it came from, who owns what.
Passing things do not: this week's mood, a draft you already sent, anything you would not want a colleague to read. Nisa asks before writing something to the Brain that came from a source she is not sure about — a transcribed voice message, for instance, is quoted back to you first, because transcription can mishear.