Assure

Independent controls around agentic AI.

Before an AI agent goes into production in a regulated firm, someone has to answer for what it can actually do — not what the policy says it does. Assurance is how you get that answer, independently and in evidence, before go-live or before a supervisor asks.

Where to start

Two steps, in order.

Step one · Free

Readiness Check

Twenty questions, five minutes, no registration. It records what your firm believes is true about the controls around your agents, scored across the four boundaries an action passes through on its way to the world. Your result is on screen; nothing is stored unless you ask for the PDF.

Take the Readiness Check

Step two · Engagement

Health Check

An independent, evidence-graded review of a live platform against thirty controls in six areas, tested against the running system rather than the document. Four weeks from engagement letter to a board decision, ending in a board paper and a one-page Statement written for insurers and supervisors.

It exists because human in the loop does not scale for agents. The controls at the boundaries around the agent are what hold, and they have to be tested, not asserted.

How the Health Check works

What comes after

Once the controls are known.

  • Governance framework. A control framework your agents are held to, mapped to the standards and regulation you answer to.
  • Pilot Rescue. A stalled agentic pilot, diagnosed and put on a path to production, or stopped for the right reasons.
  • Continuous Assurance. A retainer that keeps the controls tested as the platform and the models change.

These do not need their own brochures yet. If one is what you need, get in touch.

We also work with brokers and MGAs who want an independent view of the AI platforms they cover.

Start with what your firm believes.

Five minutes. No registration. The Health Check tells you which of those beliefs hold.

Take the Readiness Check

or book a scoping call