Governance

If AI shaped the decision,
someone must answer for it.

For a leadership team, the question is not whether AI is capable. It is whether you can answer for what it does. Governance is how you answer: written rules, a named owner, and a record you can show.

Book a 20-minute call →Talk to me about governance →

Governance model: read and draft run on their own, change, send and delete need approval, and every action is recorded in an audit trail

Questions your board will ask

Six questions.
Straight answers.

Who is accountable?

A named person, for every AI-influenced decision that matters. When AI shapes a hire, a price, a claim or a client message, one human can explain it and stands behind it.

What may AI touch?

Only what has been decided in advance and written down: which tools, which data, which systems. If it was not granted, it is not allowed. The rule lives in policy and in the tools’ own settings, not in anyone’s good intentions.

What may it do without asking?

Read and draft. Looking things up and writing a first version are safe to leave running. Every step beyond that stops for a person’s yes.

How do we know what happened?

Keep a record: what AI was used for, who reviewed the result, and what was decided. Written in plain language, kept in order, and available when a board, a client or an insurer asks.

How do we stop it?

Someone has the authority to pause a use, turn off a tool or pull back an access, and everyone knows who. A rule you cannot enforce is only a wish.

What if it is wrong?

Plan for it. Review before use, sources on anything that matters, unverified facts flagged for checking, and a simple way for people to report a problem without fear.

The permission model

Four levels. Always the same.

Read

Looks things up and gathers information.

Can run on its own

Draft

Writes emails, reports and plans for a person to review.

Can run on its own

Change

Updates files, calendars, tasks and records.

Needs approval

Send or delete

Anything that leaves the building, or cannot be undone.

Always approval

Controls at a glance

What good governance puts in place.

An inventoryA list of the AI already in use, including the tools people adopted quietly.
A written policyOne plain-language page that people read, understand and follow. See a sample.
Named ownersEvery AI-influenced decision that matters has a person who can explain it.
Least privilegeEach use gets only the access and data its job needs, and nothing more.
Human approvalChanges, sends and deletes stop until a person says yes.
Approval tied to what was shownA yes approves exactly what the person saw, nothing else.
A review recordA simple log of how AI-assisted work was checked, and by whom.
Incident reportingA known way to raise a problem, and a person who answers it.
A regular reviewThe rules are revisited on a schedule, because the tools keep changing. Security detail.

Earned, not assumed

Trust is promoted
one use at a time.

Every new use of AI starts with approval on. When one has proven itself safe, a person can choose to let it run without asking. The decision is theirs, it applies to that one use only, and it can be taken back. Nothing is promoted on its own.

Day oneeverything needs approval
After it proves safea person promotes that one use
Sending and deletingalways stay with a person

Where to start

Start with what is already in use.

For leadership teams

AI governance review

Find the AI in use, name the owners and write the rules.

See what is included → Free to adapt

A sample AI use policy

One plain page you can adapt today.

Read the sample → Keep it safe

Security basics

What stays home, what goes out, and where the real risks are.

Read the security detail →

Straight answers

Questions owners ask me.

Is this a compliance certification?

No. I am not a lawyer, and this is not a certification or legal advice. It is a working set of controls and a plain-language policy, built so your leadership can answer for AI use. Have your counsel review it for your industry.

Can we show how AI-assisted work was checked?

Yes, if you keep a simple record: what AI was used for, who reviewed the result and what was decided. That record is what you show a board, a client or an insurer.

Does governance slow people down?

Not when it is written in plain language. Reading and drafting can keep running on their own. Only changes, sends and deletes wait for a person, which is where a mistake costs the most.

Does AI replace people?

No. Governance keeps people in the decisions. AI takes on reading, research, drafting and follow-up so your team can spend time on judgment and relationships.

Where do we start?

With the AI already in use and one page of rules. A 20-minute call tells you which to do first, even if we never work together.

Start here

One short call.
You leave knowing what to govern first.

Twenty minutes. Tell me what you want AI to do and we decide what it is allowed to do. No pitch theater, and you can stop at any point.

Book a 20-minute call → Or send an email →
Book a call