Trust & security,in plain answers

How we handle your data, where systems run, who has access, and how AI is tested and watched. If something here doesn't answer your question, ask us directly.

Where does our data go?

It stays in your systems or your cloud, and is sent to an AI provider only for the task at hand. We use it for your project and nothing else, and we never sell it.

Who can see it?

Only named people working on your project, each with their own account. You'll have the list before work starts.

How long is it kept?

Only as long as the project needs. At the end, we delete or return any copies we hold within 30 days, and confirm it in writing.

Is it used to train AI models?

No. Each AI provider we use is set up so your data isn't used to train its models, and we'll show you those settings in writing.

Where do the systems run?

In your own cloud by default, so your data stays in your environment. If you'd prefer, we can run them in ours, with the same controls.

How is access managed?

Every person and system gets the least access it needs, through named accounts. Access is reviewed during the project and removed when it ends.

Can our security team review how you work?

Yes. We'll answer your security questionnaire, walk your team through our practices and agree access with them before anything connects.

How are AI systems tested before launch?

Against your own examples, with each result marked Passed or Needs review, and against misuse such as prompt injection. See Evaluation & Monitoring and AI Security Testing.

How are they watched after launch?

Quality, cost and usage are monitored, every action is logged, and a named owner is alerted when results change.

Can a system be stopped?

Yes. Every system can be paused, and agents can have individual tools switched off without taking the whole system down.