Trust Center

Key facts about data ownership, AI model use, approvals, retention and where data is processed.

Frequently asked questions

Who owns the data in my workspace?

You do. We process it only on your instructions.

Is my data used to train AI models?

No. Not by us and not by the model providers we use.

Can the AI send messages or change data on its own?

It depends on your settings. Execute-level skills, including requests from ChatGPT or Claude, always wait for a person to approve them, and admins can switch them off. Workflows, agent tools and auto-replies your team sets up run as configured; an admin can make every outbound step wait for approval with one setting.

Can I see what the AI did?

Yes. Every run is a task with input, output, model, cost and a full trace; the audit trail is exportable.

Where is data processed?

The application and primary database are hosted in the EU (Frankfurt). OpenAI, SendGrid, Replicate and Google also process data in the United States under Standard Contractual Clauses. EU hosting does not mean all processing stays in the EU. You can use your own EU-hosted model endpoint, and Sovereign offers private deployment.

Can I choose the AI model?

Yes. OpenAI by default, Google Gemini, or your own endpoint.

How long are AI records kept?

You decide: 30 days to 10 years, 365 days by default, deleted automatically.

Can ChatGPT or Claude use my skills?

Yes, through revocable keys. Their requests follow the same skill rules, and execute-level skills always wait for approval.

Certifications?

Not yet independently certified. We will publish reports here when available.