AIEnterprise

Your company brain, read by every agent.

The company brain reads the documents, diagrams, plans, work items and database records you choose, and writes down what they say: the topics, how things connect, and the facts, each with the days it held and where it came from. Orbit, your agents and every AI tool you connect answer from it, each following the access of the person asking and the project's Sharing choice.

Enterprise plan only

The company brain is part of the Enterprise plan. It is off until an owner or admin switches it on, and it only reads the workspaces and projects they choose. Talk to us to add it.

Check it before you act on it

The Brain's facts and answers are AI estimates made from your content, and they can be wrong. Every tab of the Brain says so. A fact shows the words and the item it came from, and an answer lists the sources it used, so you can check them. What that means for your organisation is in the Brain terms of our Terms of Service, which an owner or admin accepts when they switch the Brain on.
The Brain overview: what it has found in a project, and how it all links.

What it does

Connects ideas across documents

It finds the people, systems, decisions and terms your work mentions, and how they link, so an answer can draw on several documents at once.

Writes down dated facts

A fact is something your content says, with the days it held and the words it came from. Ask what was true on a past day, or how something changed.

One source for every AI tool

Orbit, your agents and the AI tools you connect over MCP all read the same brain. It keeps reading as your work changes, and nobody has to tag or file anything first.

Follows who can open what

People, agents and AI tools see what their own access lets them open. For what the Brain knows from several documents, the Sharing setting decides.

Switch it on

  1. Turn on the Brain

    An owner or admin opens Organisation settings, General, and turns on Knowledge Graph in the Features card. The switch waits until they have ticked the box below it, which accepts the Brain terms for the organisation.
  2. Choose what it reads

    In Brain settings, owners and admins switch on the workspaces and projects the Brain may read, and untick any folder or document to leave it out. The storage each workspace and project takes shows beside its switch. The Brain reads only what is switched on, and switching something off, or leaving it out, asks first: the Brain deletes what it learnt from it, for good.
  3. Let it read

    Open the project's Brain. The pill at the top says how far it has read, from Not read yet to Up to date, and pressing it lists each step with its count. Owners and admins can rebuild it from Brain settings.
Organisation settings, Brain: how much it reads and stores, and the storage each workspace and project takes, beside its switch.

The Brain in a project

  • Overview: the numbers (topics, things, links, documents read and wiki pages), the topics it covers most, and a map of how things link.
  • Playground: ask Orbit about this one Brain. A map shows what Orbit looks at as it works, and an answer lists the sources it used.
  • Ontology: the kinds of things and links it found. Owners and admins can rename a kind or pin it.
  • Wiki: pages written from what the Brain read, one for each larger topic.
  • Facts: the facts it has recorded, with the days they held (see below).
  • Sharing: who sees what the Brain knows from several documents: strict (recommended), permissive, or open to everyone in the organisation.

Ask Orbit in the Playground

Ask in plain words, or press a question the Brain offers. Orbit searches the Brain, walks to the things it reads on the map, and shows each fact it finds beside its thing, with its days and its source. An answer lists the sources it used: press one to open it in place. In the Playground Orbit only answers, and changes nothing. A question here counts like any other question to Orbit.

In the organisation's left menu, Brain opens a Playground across every Brain you may open. Orbit answers only from what you can see. The same Playground is the second tab of Brain settings.

Ask in plain words. Orbit shows what it looked at, and where each fact came from.

Facts, with their days

A fact is something your content says is true, written down with the days it held and the words it came from: Sam Okoro is responsible for Acquirer certification, since 20 October 2026. The Facts tab lists them. Each one shows:

  • What it says, in plain words, and when it held: since a day, until a day, or planned from a day.
  • Where it came from: the documents, work items and plans that state it, with the words they used. Press one to open it.
  • How it ended, if it has: the text says it no longer holds, a newer fact replaced it, an edit took its words out, or it happened as planned.

Editing a document does not end a fact, unless the edit takes out the words that stated it. Choose Current, Ended or All, search by words or names, or pick a day to see what held then. You see a fact only when you can open a document, work item or plan that states it.

Facts, newest first, with their days and where each came from.

Correct a fact

Check a fact against its source, and change what the Brain got wrong. The people who manage a Brain can: the organisation's owners and admins, the admins of its workspace and the admins of its project. Press the dots beside a fact.

  • Correct: change what it is about, or its value, and the days it held.
  • End: say it is no longer true from a day.
  • Confirm: mark it right as it is.
  • Remove: say it was never true. It no longer feeds answers and stays in All, where you can put it back.

A corrected fact shows what it said before. A decision stands: reading the content again does not bring back a removed fact or undo a correction. Orbit and your AI tools say who decided, and when. A fact that content in another project also states can be changed only by someone who manages that project's Brain too.

Fact summary emails

The Brain can email the people who manage it the facts it recorded or ended, each morning or each Monday. No AI writes the email: it counts the facts, lists up to 25 of the newest, and links to the rest in the Facts tab. It names a person's decisions the way the tab does.

  • A project's managers choose Off, Daily or Weekly under Fact summary emails in its Brain settings.
  • Each manager also turns the emails on for themselves, in their notification settings (Brain fact summaries, off until they do).

Daily emails go out at 8:00 am in each person's own time zone with the facts from the day before, and weekly ones at 8:00 am on Mondays with the facts from the week before. Each person is sent only facts they can open, and nothing is sent for a day or week with none. Organisations on a dedicated database are not sent these emails.

Brain settings: databases, sharing, fact summary emails, rebuild and storage.

Database records, as they were

Rows in your databases can join the Brain, so a question can draw on a customer record as well as the documents about it. A database stays out until an owner or admin switches it on, and then chooses what the Brain uses from each table: its names, exact identifiers such as an email or a code, facts such as dates, choices and amounts, text to read like a document, and links to other tables. Records keep their own rules: people see only the records they can open.

Ask what a record said on a past day, such as “What did the Brisemont Shipping record say on 15 October?” The Brain works the answer out from the record's own history: how it stood at the end of that day in your time zone, and who made each change. The lookup tells apart a record that did not exist yet, one that had been deleted, and one whose history does not reach back that far. It dates only what the Brain uses: links, calculated fields and who last edited a record are never dated. How far back it can look is as far back as your plan keeps each record's history (see History in Databases).

A record as it stood on a past day, with the record shown as a source.

Who sees what

The Brain uses the same organisation, workspace, project, folder and document permissions as everywhere else, in the app or through an AI tool. Anything it knows from a single document is shown only to people who can open that document. A connected AI tool sees no more than the key or sign-in it uses allows.

The Sharing tab decides what the Brain shows when something it knows comes from several documents.

  • Strict (recommended): shown only if the person can open every source.
  • Permissive: shown if they can open one of the sources.
  • Open to everyone: shown to everyone in the organisation, even without access to the documents it came from.

For your AI tools

Connected AI tools search the Brain, read a topic and what it connects to, list the documents about it, follow the links back, and read the wiki pages. They can also ask what held on a past day, or how a fact changed, for facts and for database records. They see no more than their key or sign-in allows, and what they get back is the same kind of estimate to check. See the knowledge graph tools. Finding a document by its title or id works on every plan, without the Brain.

Cost and storage

Building and using the Brain uses no credits on Enterprise: the Brain is paid for by the storage it uses. The Brain shows when it is approaching its storage allowance; at the allowance, it reads nothing new until there is room, and nothing already built is removed. Search and answers keep working. More storage is available by agreement.

Brain settings shows how much each workspace and project takes. Switching one off deletes what the Brain learnt from it and frees its room, and reading starts again by itself once there is room.

Search works the same with or without the Brain. The Brain adds what search alone cannot: answers that connect ideas spread across many documents, and what held on a past day. Turn the Brain off and everything else carries on as before.