How much of your agent risk is unmanaged?

One qualifying question, then eight. An indicative read on whether the controls around your operational agents are keeping pace with what you are handing them.

Built for the operator deploying agents inside the business, not the marketer preparing for agents outside it. Operational agent risk is exposure net of control: what could go wrong, minus how well you have contained it. This snapshot estimates where you sit.

Indicative band · 2 min · No data stored

No band returned

This instrument has nothing to score for you, and that is the useful answer.

The Operational Agent Readiness Index scores unmanaged risk arising from committing actions. An agent that drafts a purchase order is not an agent that places one. If nothing an agent does commits money, inventory, a supplier relationship, or a customer refund, there is no unmanaged operational exposure to score.

We could show you a band. It would be a false positive. A company running advisory agents will answer every control question poorly, because it has never needed an authorization model or an agent audit trail, and the instrument would report serious risk where none exists. Returning nothing is the honest result.

If committing agents are coming. The exposure appears the moment an agent completes an action without a human approving that specific action. It does not ramp gradually. And the controls take considerably longer to build than the agent does, which is why the useful time to look at this is the quarter before you deploy, not the quarter after.

If propose-only is the policy. Some organizations answer no because they have not started. Others answer no because they decided to, and enforce it. If that is you, you are not upstream of this instrument, you are running a control posture, and the risk you carry is a different one: a blanket propose-only rule is a single authorization decision made once, for everything. It holds until a business owner needs throughput on one workflow, and then it erodes one exception at a time, with no record of what changed or who agreed to it.

If agents outside your business concern you

The Agent Readiness Index

The demand-side instrument. Whether AI agents can find, evaluate, and select your brand when a buyer delegates the decision. Different exposure, same practice.

If committing agents are coming

Retake this when you deploy

Come back when an agent first acts without per-action human approval, or when a deployment is committed. That is the point this read becomes meaningful.

If propose-only is deliberate

Write the boundary down before it moves

Decide in advance which workflows may ever leave propose-only, on what evidence, and who signs. That work does not need a band, and it is far cheaper before the first exception than after the fifth.

Indicative band

Widest gap

The next step

The snapshot names the gap. The sprint closes it.

The Operational Agent Governance Sprint is a six-to-eight-week engagement that turns this indicative read into a scored Operational Agent Readiness Index and a governance blueprint your team implements: an authorization map, a data readiness assessment, and a bounded autonomy framework. The practice architects it. Your team builds it.

Start the conversation

This snapshot returns an indicative band, not a scored diagnosis. The full Operational Agent Readiness Index scores five weighted dimensions and a separate observability flag, and is delivered inside the sprint. Nothing you enter here leaves your browser.