> ## Documentation Index
> Fetch the complete documentation index at: https://docs.xpander.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Company Brain

> One place agents look up what your company knows: files stored in xpander and live sources such as Google Drive, SharePoint, OneDrive, Confluence and Jira, read as the person asking and answered with citations.

The Company Brain is what your agents know about your company. It answers from two kinds of content:

* **Collections**: files, pasted text and web pages stored in xpander, parsed into sections and searched by meaning and by exact words.
* **Live sources**: Google Drive, SharePoint, OneDrive, Confluence and Jira, searched at question time with the account of the person asking. Nothing from a live source is copied into xpander unless an admin keeps a copy of a folder.

Every agent asks the Company Brain by default. Admins can also build **specific brains**, such as Legal or Support, that hold only chosen collections and parts of sources, and attach them to the agents that need them.

## Answers carry citations

Each passage an agent receives from the brain names where it came from. In a conversation, an answer that used the brain ends with **Sources**: numbered chips, one per document or live item. Selecting **Sources** lists each one with its title and where it came from, and opens it:

* A **stored document** opens over the answer at its passages, with the cited one highlighted.
* A **live item** opens in its source (for example Confluence), with your own account.

See [Citations in the conversation](https://docs.xpander.ai/use/company-brain/agents.md#citations-in-the-conversation).

## Each person gets only what they can already open

The brain never widens access. When it reads as the person asking, which the Company Brain always does:

* **Stored documents** are filtered by who may read them before ranking, so a document someone may not open never reaches the answer or the citation.
* **Live sources** are read with the person's own account. A file shared with five people is found only for those five, and a permission removed in the source applies on the next question.
* **People who have not connected a source** get nothing from it. There is no shared or service account to fall back on. The agent says which sources it skipped, and the answer offers **Connect** for each one.

A specific brain only narrows where an agent looks. It cannot give anyone a document they could not open themselves. The one exception is a brain an admin sets to **Shared agent access**, for public agents whose visitors have no company login; see [Whose access it uses](https://docs.xpander.ai/use/company-brain/admins.md#whose-access-it-uses).

## What it reads

| Source | How it is read | An admin can limit it to |
| - | - | - |
| Collections (stored in xpander) | Searched in xpander | The people, groups and agents in its audience, and narrower per document |
| Google Drive | Live, as the person | Shared drives, folders or single files |
| SharePoint and OneDrive | Live, as the person | Sites, libraries, folders or single files |
| Confluence | Live, as the person | Spaces |
| Jira | Live, as the person | Projects |

On every live source an admin can also exclude by path, label, file type and age.

Uploaded files can be PDF, Word, PowerPoint, Excel, Markdown, HTML or images, up to 25 MB each. Scanned pages are read with OCR.

## Where you manage it

| Who | Where | What |
| - | - | - |
| Admins | **Platform > Agents > Knowledge** | Brains, sources and their scope, collections, organization settings, learned facts, access log |
| Everyone | **Settings > Knowledge** in Xpander Chat | Your own source sign-ins, what the brain knows for you, your own collections, facts waiting for your approval |
| Agent owners | **Agent settings > `KNOWLEDGE`** | The Company Brain switch, and the brains and collections attached to the agent |

> **[Set it up for your organization](https://docs.xpander.ai/use/company-brain/admins.md)**
>
> Brains, collections, sources, settings, learned facts and the access log.

> **[Your knowledge](https://docs.xpander.ai/use/company-brain/members.md)**
>
> Connect your sources, see what the brain knows for you, keep your own collections.

> **[Agents and MCP clients](https://docs.xpander.ai/use/company-brain/agents.md)**
>
> Attach brains to an agent, read citations, and ask the brain from Claude, Cursor and other MCP clients.

> **[Self-hosted and hybrid](https://docs.xpander.ai/self-hosted/company-brain.md)**
>
> Where the data stays, air-gapped installs, and the chart values to set.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.