Security & Trust
A plain-language overview of our infrastructure, data handling practices, and the commitments we make to every client before a project starts.
Core commitments
We never train on your data
Your prompts, documents, and workflow data are never used to train AI models — ours or any third party's.
We never sell your data
Personal data and client information are never sold, rented, or shared with third parties for commercial purposes.
Encryption in transit and at rest
All data is encrypted over TLS in transit. Stored data uses provider-level encryption at rest across our infrastructure.
GDPR-aligned by design
We are based in Berlin, Germany. Our data handling follows GDPR principles including purpose limitation, data minimization, and consent management.
Human review stays in place
All client deployments include human-in-the-loop checkpoints for sensitive decisions. Automation does not replace oversight.
90-day data retention
Analytics and session data are retained for a maximum of 90 days and deleted on request. We keep only what is needed to operate.
Hard limits
Sell, rent, or share your personal data with third parties for commercial gain
Train AI models on your prompts, documents, or workflow data
Store sensitive data beyond the minimum necessary for service operation
Deploy automation into your workflows without agreed review controls and rollback procedures
Use client project data for any purpose outside the agreed scope of work
Infrastructure
Edge & hosting
Cloudflare Workers
All API endpoints run on Cloudflare's global edge network. No origin server exposed.
Analytics storage
Supabase (EU region)
Consented analytics stored in Supabase with row-level security and access controls.
AI processing
OpenAI API
Prompts are processed under OpenAI's API terms. Your data is not used for OpenAI model training.
Email delivery
Resend
Transactional email for demo booking confirmations. No marketing without explicit opt-in.
Access control
Secret management + env vars
API keys and secrets never exposed in client-side code. All sensitive variables are server-side only.
Consent management
First-party, on-site
Consent is captured and stored first-party. Non-essential analytics are disabled until you opt in.
Client projects
Every client engagement includes a scoping session where we agree on data access scope, review controls, human checkpoints, and rollback procedures before a single line of automation is written.
Minimum-access data design
We request only the data access required to deliver the agreed workflow. No broad permissions, no access beyond scope.
Review checkpoints in every pilot
Automation outputs go through human review before affecting downstream systems. Control gates are agreed in advance.
Confidentiality by default
Client workflow data, process details, and business logic are treated as confidential and never shared outside the project team.
Rollback and exit procedures
Every deployment includes a defined rollback path. You can stop or revert any automation without depending on us to do it.
Questions
We're happy to discuss data handling, access controls, or compliance alignment before you commit to a pilot. Email us directly or book a call.