> ## Documentation Index
> Fetch the complete documentation index at: https://docs.get-hive.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Business Brain: what Hive knows about your business
> Explore the Business Brain, the shared knowledge graph Hive builds from your tools: entities, facts, sources, trust levels, relationships and how it learns.
The **Business Brain** is the shared memory Hive uses to understand your business: your customers, suppliers,
deals, systems, teams and the facts that connect them. It is what lets Ask Hive answer questions in your context,
lets Signals judge how much a problem matters, and lets proposed actions reflect how your business actually works.
Open **Business Brain** in the sidebar (`/brain`). The page has two tabs:
* **What Hive knows** — the knowledge graph: "How Hive understands your business — one shared memory across every
department."
* **Learned behaviours** — the rules, workflows and exceptions Hive has learned from your team. See
[Learned behaviours](/brain/learned-behaviours).
Admins can switch the Business Brain product off for a workspace. If you do not see it in the sidebar, ask an
Owner or Admin. See [Workspace settings](/admin/workspace-settings).
## Explore the knowledge graph
The **What Hive knows** tab shows a map of entities and the relationships between them.
* **Search entities & facts…** finds a customer, system or fact by name.
* Filter by **department** to see one part of the business.
* The counter **Hive has learned N things** opens a list of the facts, baselines and rules Hive has added.
* Double-click a node, or press **F** while it has keyboard focus, to see just its neighbourhood. Choose **Exit
focus** to return to the whole map.
Select any entity to open its inspector:
| Section | Shows |
| - | - |
| **Facts** | Each fact with its source, the date Hive learned it, and its confidence. |
| **"Normal" baselines** | The learned normal range for a metric, and the date Hive learned it. |
| **Relationships** | For an entity with no recorded facts yet: how it connects to others, such as "owned by" or "customer of". |
| **Discovered through** | For an entity with no recorded facts yet: the connected tools it was found through. |
If Hive has not learned anything yet, you will see **Nothing learned yet** (or **Your brain is forming** once entities
exist without facts): connect a tool and the Brain fills in as Hive collects from it, then on every hourly run.
## Trace how two things connect
In an entity's inspector, choose **How is this connected to…**, then search for a second entity. Hive shows the
shortest chains of learned relationships between the two — each link written as a sentence with its role and, where
recorded, the date it was learned. A record that appears only as the target of a relationship, and is not an
entity of its own, is marked **External reference**.
Hive looks for chains of up to **four** links and shows at most **three** equally short ones. Links through
your own business entity, or through Hive's own monitoring hub, do not count, because almost everything connects through them.
## Trust levels
Not every fact is equally reliable, and Hive does not treat them as if they were. Every fact carries a trust level,
and Hive weights it accordingly when it answers questions or ranks signals:
| Trust | Typical source | Weight in retrieval |
| - | - | - |
| **Verified** | Confirmed by a person, or a signed system of record | 1.0 |
| **Reported** | A connected system of record, such as your payments or accounting tool | 0.85 |
| **Inferred** | Computed by Hive itself, such as a learned baseline | 0.6 |
| **Untrusted** | Ingested content such as inbound email or chat text and web pages | 0.25 |
A fact derived from several sources takes the trust of its **least** trusted source, so an inference can never be
more reliable than its weakest input. During onboarding, facts Hive gathers from your public website start as
low-trust context.
## How the Brain learns
Hive builds the Brain in the background:
1. **Collect.** Every hour, Hive reads fresh data from your connected tools and synced file stores (Google Drive,
OneDrive, SharePoint, Dropbox). Synced files keep the sharing permissions they have in the source.
2. **Reflect.** Hive extracts facts, relationships and patterns from what it collected. All content is treated as
untrusted data, never as instructions. Candidates with low confidence or only stale evidence are discarded.
3. **Keep history.** Writes never overwrite silently. When a value changes, the old fact is closed and the new one
recorded, so you can always answer "what did Hive believe on that date?".
Facts cannot be edited directly in the Brain today. When a fact is wrong, correct it in the source system — the
connected tool is the source of truth, and the next run picks up the change. If a *behaviour* is wrong rather than
a fact, pause the relevant rule in [Learned behaviours](/brain/learned-behaviours).
## When an answer or signal looks wrong
Find it with **Search entities & facts…**.
Is it stale? Is it from a low-trust source?
Correct the record in the connected tool if that is where the error lives.
If the fact is right but Hive acted on it wrongly, review the matching rule.
Better source data improves Ask answers, signals and proposed actions all at once.
## Who can see what
The Brain is one shared memory, but access still follows your workspace roles and department scopes. Files synced
into the knowledge base keep their original permissions, and mail is stored as private to the mailbox owner.
## Related
Rules, workflows and exceptions Hive follows.
What Hive remembers about you in chat.
Ask questions grounded in the Brain.
How the Brain shapes signal priority.