Answers before you ask.

Straight answers about scope, evidence and how engagements start.

What exactly do you check in Secure?

We test the real attack surface of your AI systems against recognised frameworks — OWASP LLM Top 10, MITRE ATLAS, CSA MAESTRO — not a method we invented ourselves.

How long does Analyze take?

First result within 48 hours, full presentation within 5 days. Your own effort is around 5 hours.

Do we need to commit to a full programme to start?

No. Every route starts with the smallest useful engagement — you approve scope before any work begins.

What happens if the initial analysis doesn't convince us?

You pay nothing for it. The Autorea Confidence Promise means we present evidence first — payment follows only once you're convinced.

Which standards and frameworks do you test against?

Depending on the service: OWASP AI Testing Guide, OWASP LLM Top 10, MITRE ATLAS, CSA MAESTRO, Google SAIF, NIST AI RMF, ISO/IEC 42001 and more — see the full matrix on How We Work.

Does this work with the tools we already use?

Yes. We connect your existing stack — Shopify, HubSpot, Zapier, n8n and others — into one system that can be observed, instead of replacing it.

Do you need admin access to our systems?

Only the minimum required for the agreed scope, defined in writing before we begin. Nothing broader.

How is Autorea different from a general AI or security consultancy?

We combine hands-on AI engineering with recognised security methodology — the same pattern that secured every prior technology wave, applied before the damage instead of after.