Contents · 5 min read
01 / 05 · The choice
What is your leadership team deciding?
In short
the boundary between preparation and authority matters because a stronger case does not remove the need for a named leader to decide.
Before your next leadership meeting, AI may have compared plans and found gaps in the case. It may also bring past choices into view, so your leadership team can focus on the choice and the risks your company must accept. But one named leader still makes the final call.
Your meeting can then focus on the choice and the risks your company must accept.
But as AI does more, its draft may start to look like the decision itself. The decision owner may be unclear even when the case is well made.
The AI-augmented management team keeps a clear split. AI prepares and checks the work, while the people who face the result choose what the company will do.
A named leader must own the risk and the result, so that leader must make the final call.
02 / 05 · Authority
Why must AI stop before the final call?
In short
AI can improve the case for a choice, but it cannot carry the consequences, so a named leader must make the final call.
AI may find facts your team missed.
The Skills That Survived describes a board simulation in which AI scored higher than the human boards on decision quality and evidence use.
But the same work showed what AI could not do. It could not build trust or read the room. It could not sense when one person was being pushed aside.
Assign the work in four parts.
Facts. AI gathers the facts and finds what is missing. A named leader decides which facts should shape the call.
Possible paths. AI tests what may happen and makes the case against each path. A named leader chooses which risk the company will take.
Draft action. AI sets out what the company could do. A named leader approves what the company will do.
Monitoring after the decision. AI flags a change after the call. A named leader decides when to stop or change course.
The next question is which final calls must stay with a person.
03 / 05 · Final call
Which decisions must stay with a named leader?
In short
a named leader must make the final call whenever the decision sets your company’s values or requires someone to carry its consequences.
Use four tests before your team gives AI a role in a consequential decision.
Does the call set direction?
Strategy stays with your leaders because they choose the bet and what your company will leave aside.
Does it put people or funds at risk?
Choices about who gets a role stay with your leaders. So do choices about where funds go, because a person must own the result.
Must someone answer for the outcome?
A crisis call needs a person who can accept the risk and carry the result.
Does the call depend on trust?
A hard message needs someone who can read the room and respond with care.
THIS IS KILLING YOUR COMPANY explains why trying to appear caring can weaken trust when the care is not genuine.
Naming a leader is only the start because review can still become a quick approval. The next step is to give that leader a real point of control.
04 / 05 · Human review
When does human review protect the boundary?
In short
human review protects the boundary only when a named leader can stop or change the proposed action before it takes effect.
The word “review” does not say what the person can do. It also does not show when they will see the proposed action.
For an important use, put a named leader at the point before the action takes effect. That leader must see the case and be able to reject or change it. The decision record should record the reason and the owner. It should also keep any dissent and the trigger for reconsideration.
The EU AI Act checklist sets the same requirement for consequential uses: name the person and the exact point where they can step in.
A broad review step may only wave the output through. An exact stop point gives your leader the time and power to judge it.
Once that point is clear, your team can set the roles before each consequential call begins.
05 / 05 · The boundary
How should you set the boundary before each decision?
In short
before every consequential decision, state what AI prepares and name the leader who makes the call.
Use this sequence for a consequential decision in your next leadership meeting:
State the call
The owner writes the exact choice your company needs to make.
Assign the preparation
The team states which facts AI may gather and which options and risks it should test.
Set AI’s limit
The record says what AI may suggest and what it may never approve.
Mark the stop point
Human review happens before the action can affect people or bind the company.
Record the reason
After the call, the owner keeps the choice and the grounds behind it. The record also states what would cause a new review.
Your leaders should use AI in this work because they need firsthand experience to judge its limits. Your Leadership Team Is Behind on AI argues that leaders cannot hand off AI strategy while staying new to it themselves.
Make the decision question and owner part of the management team’s weekly record, then add the stop point and reason.
Make the boundary work
Your next task is to turn the boundary between preparation and authority into a working rule your management team can apply before each consequential decision. The paper on the AI-augmented management team shows how leaders can separate preparation from judgment while keeping the final choice with the person who must carry the result. Use it to set the first changes in how your team prepares decisions and records who owns the call.
The paper on what changes in the management team, with the first changes to make. Free, with a PDF to share.
Questions CEOs ask about this
Which leadership decisions should we never delegate to AI?
Never delegate the final call to AI when a decision sets your company’s values or requires a leader to own the result. You must decide where AI’s preparation ends before its draft starts to look like the decision itself and leaves ownership unclear. Name who decides and where that leader can stop or change the plan before it takes effect.
Who should own AI governance across the company?
Your board owns how much AI risk the company will carry and for what return. One accountable AI-policy owner maintains the framework and register. Each use-case owner runs the controls around their use, while Legal advises on difficult classifications.
Does this only apply when your company builds AI?
No. The EU AI Act checklist says a company that uses third-party AI systems is a deployer. Its duties are lighter than a builder’s, but they are still real.
Founder, GRAIL
Johan has spent fifteen years in management consulting and executive leadership, including as Chief Product Officer of a SaaS company. He founded GRAIL to help European leadership teams work out what AI changes at the core of their business. He is the author of The Execution Revolution and the host of the ThinkRoom podcast.
Related reading
Sources
- What does an AI-augmented management team look like?: how AI prepares management work while people keep the final choice. (accessed October 29, 2026)
- EU AI Act Checklist for Companies: why human review needs a named person and an exact point to step in. (accessed October 29, 2026)
- The Skills That Survived: what AI did well in tested board work and which human skills it lacked. (accessed October 29, 2026)
- THIS IS KILLING YOUR COMPANY: why false care can weaken trust. (accessed October 29, 2026)
- Your Leadership Team Is Behind on AI: why leaders need direct use of AI in high-stakes work. (accessed October 29, 2026)
Next scheduled review: February 14, 2027