Skip to content

Security overview

Sail is built for companies that can’t send their work through a consumer AI app. This section answers the questions a security team asks, plainly, including the limits.

These pages describe Sail itself, wherever it runs. If Msty hosts Sail for you, ask Msty for the hosting details as well: regions, sub-processors and the data processing agreement.

QuestionShort answer
Is our data mixed with other customers’?No. Each Sail install serves one company, with its own database and file storage.
Where does our data live?In your Sail’s Postgres database and file volume. Self-hosted, that’s on your servers.
Does Sail send data to Msty?No. A self-hosted Sail never calls Msty and sends no telemetry.
Who sees our prompts?Your Msty Nexus and the model providers it’s set up to use. Sail holds no model keys.
Can admins read people’s chats?No. Nobody in Sail can read another person’s chats or memories, owners included.
How do people sign in?Through your own identity provider, over OpenID Connect. Sail stores no passwords. MFA stays in your provider.
What’s logged?Who changed what, in an audit log. Never message content.
Is data encrypted?In transit, with HTTPS. Stored secrets are encrypted by Sail. Chats and files rely on your disk and backup encryption.
Can people delete their data?Yes. They can export everything, and ask for erasure. An admin approves.
Does the AI run code?No. Sail never runs code a model writes. Files are made and read by Sail itself.
Is it certified?No certifications are claimed yet.
  • Data and privacy: what’s stored, who can see it, what leaves Sail, and how it’s deleted.
  • How Sail is protected: sign-in, sessions, encryption, AI-specific safeguards, the audit log, and how Sail is built and shipped.

Found a vulnerability in Sail? Report it to Msty privately before sharing it anywhere else. Self-hosted installs publish their own contact at /.well-known/security.txt. Set yours with SAIL_SECURITY_CONTACT (see Configuration).