Security

How Cydrah AI keeps your work secure

Local-first architecture

Conversations, attached files and settings are written to disk on the machine Cydrah AI runs on. There is no Cydrah server that stores or sees your content.

Open-source model runtime

Models run through Ollama, a free, open-source, auditable runtime. Updating a model is a manual action you trigger — nothing downloads in the background.

No telemetry by default

Cydrah AI does not phone home. Optional web search is off by default, requires a per-message opt-in, and is clearly flagged in the conversation when used.

Audit logging, kept local

Teams can enable a local audit log for compliance review and export it to CSV on demand. Message content logging can be switched off, keeping only timestamps and timing.

Report a vulnerability

Found an issue? Email [email protected] — we credit responsible disclosures publicly on request.

Set it up once. Use it every day.

Join the waitlist