How We Handle Data

What your brain stores

Your brain is a structured knowledge base: pages, decisions, references, and the relationships between them, distilled from the sources you connect or upload. Source files you upload (documents, PDFs, transcripts) are kept in your brain's own isolated storage, linked to the pages derived from them, so every answer can cite its evidence. Each company's brain lives in its own dedicated storage; membership is re-verified on every request.

Who sees what: roles

Every person in your brain has one of three roles. What each role can see is enforced by a single permission layer that filters every read (search, page reads, graph walks, listings) before anything is returned. A page you cannot see never appears, not even as a search snippet.

Owner

The account holder. Full access to content, settings, members, and billing.

Admin

Can read all content levels and manage members and settings. Every admin read of confidential or private content is written to the audit log, so elevated access is always accountable.

Member

Sees standard content, plus confidential content only if they are in its named audience. Companies can additionally set members to AI-only mode, where members work with the brain through their AI tools without browsing the content dashboard.

Who sees what: visibility levels

Every page in a brain carries one of three visibility levels:

Standard

The default. Visible to every authenticated member of your brain.

Confidential

Scoped to a named audience, for example Finance or Legal. Only people in that audience (and accountable admins) can see it.

Private

Personal pages visible to their owner. Admin access to private content exists but is always audit-logged.

Sensitive content never enters silently

When ingestion detects likely sensitive material, such as financial data, compensation, cap tables, legal matters, personal data, or credentials, the content is held instead of written. A human decides its visibility explicitly before it enters the brain. Nothing sensitive lands in shared memory by accident.

What our AI does with your content

We use Anthropic's Claude models to process and organize content during ingestion and upgrades. Those calls run under configurations that exclude your content from model training. Your data is never used to train AI models, ours or anyone else's, and we do not hold your AI provider tokens: you bring your own AI clients and keys.

How your data is protected

Encryption

TLS in transit, encryption at rest.

Audit trail

Every action against your brain, by a person or an AI, is logged: who, when, what was touched, including denials. Audit events are retained for up to 400 days.

History

Your brain's history is append-only, and any change can be reverted exactly, byte for byte.

Backups

The whole brain is snapshotted daily; snapshots rotate out on a schedule of roughly five weeks.

Adversarial review

Every access-control change is attacked from multiple angles (malicious insider, forged credentials, stale caches, spec drift) against the real merged code before it reaches production.

Your rights and controls

Access

Request a copy of your personal data at any time via info@lanbrain.ai.

Deletion

Request deletion of specific content or your whole account. Deletion removes your brain's storage, its index, and uploaded source files; rotating backups clear within roughly five weeks.

Visibility control

Set any page to standard, confidential, or private, and define named audiences for confidential content.

Connection control

Every third-party connection is authorized by you and revocable by you at any time.