Skip to main content

Intervention Rights™

Every AI system needs a brake pedal.

Most AI advice talks about governance, controls and policy. Very few ask the question that matters most when something goes wrong: who can stop the AI?

Check your readiness See the Intervention Rights service

The questions nobody has answered

Ask these about any AI tool your business uses. If the answer is "I'm not sure", you have an accelerator and no brake.

The five intervention rights

A brake pedal is really five different controls. Each one needs a named owner.

Stop

A named person can switch the AI off for a process, immediately, without needing permission first.

Pause

Outputs are held for human review while something is checked — work carries on, nothing goes out unseen.

Override

Staff can replace an AI recommendation with their own judgement, and the reason is recorded.

Reverse

A decision that has already reached a customer can be corrected, with a clear owner and a time limit.

Escalate

Agreed triggers move a problem up — from team lead to director to board — before it becomes a crisis.

What Intervention Rights™ puts in place

Who holds the brake

A rights matrix naming who can stop, pause, override, reverse and escalate each AI use — by role, with cover for holidays.

When to press it

Plain triggers: error rates, complaints, unusual outputs, data leaving your control. No one has to guess whether it's serious enough.

Proof it was pressed

Every intervention leaves a record — who, when, why and what happened next. That is your control evidence when someone asks.

Design your brake pedal

AI Intervention Rights & Escalation Design: we map your AI uses, name who holds each right, set the escalation triggers and build the intervention log. From £1,750, with optional ongoing review at £395/month.

View the service

Then prove it worked

An intervention is only useful if you can show it happened. See how authority, control and outcome evidence fit together.

Governance isn't a policy. It's proof