Transcription runs on your laptop by default. Data is walled off between tenants. Every credential is encrypted. Every admin action is logged. Security was the first architectural decision, not a late retrofit.
By default, audio stays on the device — the laptop runs on-device transcription and syncs the resulting text when you reconnect. When a segment's on-device confidence is low, just that audio segment may be securely uploaded and transcribed in the cloud to keep your transcript accurate. Details in our Privacy Policy.
The desktop app transcribes on-device, and text transcripts sync when you reconnect. Low-confidence audio segments may be securely uploaded and transcribed in the cloud to improve accuracy — encrypted in transit, used only to generate your transcript, and automatically deleted within 30 days.
Every table has a tenant-scoped filter. Data separation is enforced in SQL, not just in policy. Your leads, transcripts, and analytics are only ever visible to your team.
TLS on every request. Integration credentials encrypted at rest. Passwords hashed with bcrypt. API keys stored as SHA-256 hashes — the plaintext only exists on your machine.
Role-based permissions. Scoped booth tokens. Every admin action logged with actor, resource, and old/new values.
GDPR-ready data handling, right to deletion, retention controls, and a complete audit history your legal team can pull.
Hosted on SOC 2 certified cloud providers. Continuous monitoring, automated deployments, enterprise uptime.
Happy to answer deep questions about our architecture, our data handling, and anything else your legal or security review needs. Drop a note — we reply by email.