Skip to content
FAQ

Questions, answered.

How we test, what we need from you and what you get back.

What is continuous red teaming?

We attack your apps, APIs, cloud and AI agents the way a real attacker would, on a schedule and on every release. Every finding comes with proof and a fix, and we retest after you ship it.

How is this different from a penetration test?

A pentest checks one version of your product once a year. We keep testing as you ship, so new code and new agent tools get the same scrutiny as the old ones.

Do you test in production?

Only if you authorize it. Most clients give us a staging environment and test accounts. In production we use read-only, low-impact methods and agreed time windows.

Do you need our source code?

No, but it helps. With source access we point to the exact line that failed. Without it, you still get the request, the response and the fix.

Which AI models do you use?

Open models with refusals removed, running in our own attack harness. They cover ground that hosted models refuse to test. Every result is reviewed by an engineer.

What happens to our data?

Your data never trains an AI model. We use only providers with short retention, named in your contract, and keep only what the report needs.

Can we use your report for SOC 2 or ISO 27001?

Yes. The report shows what was tested, how, what we found and what was fixed, in a format you can hand to your auditor.

How fast do we hear about a critical finding?

As soon as an engineer confirms it. You do not wait for the end of the cycle.

How do we start?

Tell us which app, API or agent matters most. We agree the scope, test accounts and stop conditions in writing, then start testing.

Still have a question?

Email us and an engineer will answer.