Is your AI agent deployment actually governed?
Most AI safeguards live inside the model — instructions telling it what not to do. Those are advisory, not enforcing. This 10-question assessment scores your exposure against your controls and returns the specific gaps worth closing first.
Two tracks
An executive track asking in plain language about stakes, ownership and accountability; and a technical track on enforcement points, identity propagation, read controls, action gating, injection isolation, observability and adversarial review.
What you get
An exposure score and a control-maturity score, your position on the effort–impact quadrant, observations on each weak dimension, priority actions in impact order, quick wins separated from foundations, and a short list of what to avoid.
What happens next
The output maps directly to a Phase 0 assessment of two to four weeks: a data-estate and IAM-maturity map, entitlement-gap analysis, threat model and a prioritized control roadmap. Book an introductory call or email pw@prima-partners.com.
Frequently asked
- How long does the assessment take?
- About three minutes. Ten questions, in either an executive or a technical track, and the result appears immediately.
- Do I have to give my email to see the result?
- No. The full result — scores, observations, priority actions and what to avoid — is shown on screen. You only share an email if you want a written follow-up.
- What does the result actually measure?
- Two axes: exposure (how much damage your agents could do today) and control maturity (how much of that is enforced outside the model rather than by prompt instructions).
- What happens after the assessment?
- The gaps map to a Phase 0 scoping engagement of two to four weeks that converts them into an owned, sequenced control roadmap.