SECURITY
Built for business. Designed for control.
The controls that are actually implemented, described plainly — and an honest statement of what we do not claim.
- Sign-in is Google OAuth 2.0 with PKCE. There is no password to store, because the platform does not use one.
- Session cookies are set HttpOnly, so page scripts cannot read them.
- Role-based access control across four roles, checked on the server for every authenticated route.
- Multi-tenant data isolation: requests are scoped to the caller's own organization in the application layer.
- Audit logging of privileged and tenant-affecting actions.
- Server-side credential protection: provider and API keys stay on the server, and provider error messages are scrubbed so key material cannot be returned to a browser.
- Human oversight and deployment controls: a call handler must pass readiness checks and be approved before it can take a live call.
- Rate limiting on public endpoints.
- We don't claim uncertified compliance. Where we have not been audited against a standard, we say so rather than implying otherwise.
- Security practices will keep changing as the platform develops. This page describes what is in place today, not an aspiration.
Questions about anything on this page, or a request about your own data? Get in touch.