Per-tenant isolation
Each organization gets its own scoped workspace. Queries, models, and uploads are bound to your org context, with tenant-isolation checks in the authorization path.
Security & Trust
Security & Trust
Geist is built in Munich for teams that take their data seriously. Here is our posture today, our commitments, and what is on the roadmap, stated plainly, with no overclaiming.
Data protection
Tenant-scoped data. Processing locations defined for your deployment.
Geist is EU/Munich-based. Every customer runs in a logically isolated tenant; data is partitioned per organization and protected with org-scoped access controls.
EU / Munich-based
We operate from Munich. Hosting regions, AI processing locations, and subprocessors need to be confirmed for each customer deployment.
Least-privilege by default
Access is scoped to the minimum needed for a role. Service-to-service calls carry explicit org and actor context on every request.
Authentication & access
Sign in with the identity you already trust.
No passwords to leak. No tokens sitting in the browser. Sessions are opaque and server-held.
OIDC: Google & Microsoft
Single sign-on via Google or Microsoft (OpenID Connect), plus invite codes for controlled onboarding. We can align with your Azure AD or Auth0 directory.
Opaque, httpOnly sessions
Sessions are opaque server-side records delivered as httpOnly cookies. No access tokens are exposed to client-side JavaScript.
Audit logging
The product includes organization-scoped audit records. Event coverage, retention, and monitoring need to be verified for each deployment.
AI & your data
Know which providers process your data.
Geist supports customer-connected AI accounts on supported routes. The provider account, processing location, and terms depend on the configured feature.
BYOK: bring your own keys
Supported integrations can use your provider credentials. Confirm account ownership, billing, retention, and processing terms for every enabled integration.
We do not train on your data
Geist does not use your process data, documents, or prompts to train models. Your operational data is yours.
Data shared with providers
AI features can send prompts, documents, images, and tool results to external providers. The enabled features and their data flows need to be reviewed before customer use.
Compliance roadmap
Where we are and where we are headed.
Security controls are implemented in the product. Deployment verification, legal review, and independent assurance remain separate work. We do not currently hold ISO 27001 certification or a SOC 2 report.
In place today
- Privacy request management workflow
- Per-tenant data isolation
- OIDC SSO + opaque server sessions
- Audit logging & least-privilege access
On the roadmap
- ISO 27001 readiness: priority for EU pilots
- SOC 2: preparatory controls, no observation period started
- Customer DPA, security measures, and subprocessor schedule: under preparation
- Third-party penetration testing: planned
Need something specific for your procurement review? Ask us; we would rather scope it honestly than overclaim.
Found a security issue?
Read the disclosure policy, then report the issue privately. Do not open a public issue or include credentials in your message.