Last updated: 14 June 2026

Cairnmind is a team "second brain": it ingests the documents and data a workspace connects, builds a private per-workspace knowledge graph, and answers questions with citations grounded in that workspace's own data. This policy explains what we process and why. It applies to the hosted service at cairnmind.com; self-hosted deployments are operated by whoever runs them, who is the data controller for their own instance.

What we process

• Account data: name, email, hashed password or single-sign-on identifier, workspace membership and role, and authentication factors (e.g. 2FA).
• Workspace content: the documents, messages, files, and other data you upload or connect (for example via a connector or the ingest API), plus the entities, relations, and embeddings derived from them. This content is isolated per workspace (tenant) at the database level.
• Connector credentials: tokens you provide for sources you connect, stored encrypted at rest (AES-256-GCM). For local-agent connectors (e.g. Google Drive/Gmail), your credentials and the raw source data stay on your own machine and never reach our servers; only the extracted text you choose to send is ingested.
• Operational data: usage counters, audit logs of security-relevant actions, and standard request metadata used to run, secure, and rate-limit the service.

How we use it

• To provide the service: store your knowledge, answer your questions, and run the connectors you enable.
• To secure and operate the service: authentication, abuse and cost limiting, audit logging, and support.
• We do not sell your data, and we do not use your workspace content to train our own or third-party foundation models.

AI processing and sub-processors

To produce cited answers and to extract entities/embeddings, relevant excerpts of your content may be sent to the configured model provider (by default OpenAI) for inference. These providers act as sub-processors and, under their API terms, do not train on data submitted through their API. A self-hosted deployment can point Cairnmind at any OpenAI-compatible endpoint (including a self-hosted model), in which case no data leaves that operator's chosen infrastructure.

Agent and MCP access

You can connect external AI agents to your workspace over the Model Context Protocol (MCP) or the REST API using API keys you issue and can revoke. Each key carries a role (read-only, read-write, or maintainer); a key only ever sees the tools its role permits, and every tool call is recorded in your workspace audit log.

Retention and deletion

• Workspace content is retained while your workspace is active. You can disconnect a source and purge all data it contributed at any time.
• You can export your full workspace and you can delete your workspace, which removes its content from our systems (subject to short-lived, routine backups that age out).

Security

Hard multi-tenant isolation via PostgreSQL row-level security, encryption of secrets at rest, encryption in transit (HTTPS), CSRF/origin protection, rate limiting, and outbound-request (SSRF) guarding. No system is perfectly secure, but isolation and least-privilege are designed in from the first migration.

Your rights

Depending on your jurisdiction you may have rights to access, correct, export, or delete your personal data. Workspace owners can action most of these in-product (export and delete); for anything else, contact us.

Contact

Questions or requests: privacy@cairnmind.com.