Skip to main content

Overview

The Brain is your organization’s shared memory — the facts, decisions, interpretations, and policies your agents draw on to do their work. Every piece of knowledge is backed by evidence you can trace, and nothing becomes active until a person confirms it. Open it from Agents → Brain.

Layout

The Brain is split into two panes:
  • Left — pages list. A searchable list of knowledge pages. Each page represents a concept your organization knows about (a person, team, process, decision, policy, topic, and so on) and shows its name and type.
  • Right — page detail. The knowledge, evidence, relationships, and history for the page you select.

Searching

Type in the Search Brain box at the top of the list to find pages by name or type. Results update as you type. Select any result to open its detail on the right.

Reading a page

A Brain page brings together everything the organization knows about one concept.

Knowledge

The Knowledge section lists the page’s claims. Each claim shows:
  • What it asserts — the statement itself.
  • A state badge — for example Active (confirmed knowledge) or Disputed.
  • A type badge — the kind of claim (such as a policy, preference, inference, or recommendation).
  • Evidence — one or more citations back to the source that supports the claim, including the source title, where in it the claim came from, and a quoted excerpt where available.
The evidence shown on a page is recomputed from only the sources you can access. You’ll never see a claim supported by evidence tied to a project you don’t have access to.

Relationships

The Relationships section lists connections between this page and other concepts — for example, a person who belongs to a team, or a decision that relates to a policy. Each relationship shows its direction (inbound or outbound) and its type. Click any relationship to jump to that page.

Revision history

The Revision history section lists how the page has changed over time, showing each revision’s number, when it was recorded, how many source revisions supported it, and its status.

Giving feedback

Anyone on your team can help keep the Brain accurate. Click Feedback on a page (or on an individual claim) to open the feedback dialog, choose a feedback type, add context, and submit.
Feedback typeUse it to
ConfirmVouch that a claim is correct
CorrectPoint out an error and describe the right answer
DisputeFlag a claim you believe is wrong
Mark importantSignal that a claim matters more than its current weighting
Your feedback enters the organization’s pending review queue — it doesn’t change the Brain directly.

Reviewing pending knowledge

Organization admins see a few extra controls for approving what enters the Brain.

The Pending filter

Above the pages list, a All / Pending toggle appears for reviewers. Switch to Pending to see only pages that have knowledge or feedback waiting for a decision. A count badge shows how many items are outstanding.

Confirming or rejecting claims

Newly proposed claims — for example, candidate policies, preferences, inferences, and recommendations extracted by the Brain Maintainer — appear on the page marked Pending. As a reviewer you can:
  • Confirm — accept the claim so it becomes active knowledge.
  • Reject — discard the proposed claim.

Resolving feedback

Pending feedback submitted by your team appears on the relevant page (or claim). Reviewers can Approve or Reject each feedback item to act on it.
Candidate policies, preferences, inferences, and recommendations stay pending until an organization admin confirms them. Agents can propose knowledge, but only a person can make it active.

Sources

The documents and Royaltyport data the Brain learns from.

Playbooks

Procedural guidance that steers extraction agents.

Agent Registry

See which agents read from the Brain.

AI Chat

The assistant that draws on confirmed organizational knowledge.