An agent inventory with owners
The agents you run, including ones nobody approved, each with a named owner.
Our main entry engagement
A fixed-scope, read-only engagement designed to give you a clear starting point: which agents you run, who owns them, what they can reach, and whether your controls hold. Metadata only, never secret values.
How it’s designed
We’re designing it as a small, consent-based engagement, scoped with you in writing. We’re shaping it now with practitioners.
What you get
The agents you run, including ones nobody approved, each with a named owner.
What each agent can reach across your tools, systems and data (its Agent(ic) Perimeter), and where its access goes beyond what it needs.
Where your AI policy and the controls that should enforce it don’t yet match, in a short list you can act on.
Each finding with an owner and a next step, plus a plain summary for your board and evidence your auditors can follow.
We never call an agent “safe”. We show what we found and the evidence behind it.
What it isn’t
It’s designed as an independent check on what your existing platforms report, and a starting point for governing agents through their lifecycle.
We’re talking with CISOs, CROs, CIOs, compliance and internal audit leaders about what evidence their boards and auditors need.