When to use Hive Server
What you'll learn
Estimated time: 9 minutes
Hive Server runs the same knowledge-graph model as the Aiqbee Platform, on infrastructure you operate yourself. The decision to self-host rarely comes down to price; it comes down to which controls your organisation actually needs and which operational load you are willing to take on.
- Compare Platform and Hive on features, not just hosting location.
- Name the Hive-exclusive controls that justify the extra operational work.
- List what Hive shifts onto your own team: Postgres, identity and upgrades.
Video lesson
The animation walks a fictional compliance team through a requirements checklist and lands on Hive because of groups, per-neuron-type access and an audit trail.
Two products, one graph, different jobs
Aiqbee ships two products from two different codebases. The Platform is a.NET SaaS application at app.aiqbee.com, with a chat assistant called Bea, a Marketplace and a wizard that builds a brain from uploaded files. Hive Server is a Python application you run yourself, in Docker or Kubernetes, against your own PostgreSQL database.
Some capability sits on only one side. Do not expect a Platform screenshot in someone else’s demo to match your Hive instance, or the reverse: the two products diverge on purpose, and this course only teaches what Hive itself does.
What Hive gives you that the Platform does not
Four capabilities exist only on the self-hosted side:
- User groups: grant read or read-write access to a whole team at once, without handing anyone ownership.
- Per-neuron-type access: restrict specific neurons to specific people, service accounts or groups, even inside a brain that is otherwise open.
- Audit history: a queryable record of who created, changed or deleted what, retained for a configurable number of days (28 by default).
- A management API and a richer MCP prompt set built for automation, covered in the rest of this course.
What you take on in return
Self-hosting means you now run the database. Hive stores everything it holds (brains, neurons, embeddings, users, the audit log) in the PostgreSQL instance you point it at; nothing is copied to Aiqbee. That is good for control and inconvenient in practice: backups, point-in-time recovery and disaster planning become entirely your responsibility, a point this course returns to directly.
The container is also intentionally minimal. TLS termination, a web application firewall, DDoS protection and rate limiting are not built in; they belong at your edge, such as a reverse proxy, an API gateway or a cloud front door. Licence keys set your own ceilings on users, brains and neurons; the published free-tier numbers differ between documents, so check the current licensing page rather than trust a specific figure here.
Before you continue
Platform-only and Hive-only features are listed correctly, not guessed.
At least one real organisational requirement points specifically to a Hive-exclusive control.
The operational load of self-hosting (Postgres, edge security, upgrades) has been named, not assumed away.
