Quick answer: An AI playbook for CEOs is not a list of tools or use cases. It is a decision lens you run every week. Mine is called MOMENTA, and it asks three questions in order: where does leverage actually compound here, what must NOT be multiplied yet, and what decision will this let us revisit faster? AI is a force multiplier. One unit in, ten out. So your job is placement judgment, not technology. MIT found that 95% of AI initiatives fail to turn a profit while 5% win with the very same tools. The gap is judgment. This playbook is how you build the judgment.
By Andreas Pettersson, founder of Leaders ADAPT and a former Canon AI executive who built and sold an AI company before ChatGPT existed.
Most CEOs I meet have plenty of AI activity. Pilots here. A tool there. Someone on the team running experiments nobody reviews. What they do not have is a way to decide. They are busy. They are not compounding.
Here's the thing. AI does not fail because the technology is weak. It fails because there is no operating system behind it. No simple rule a leader runs again and again until good placement becomes a habit.
I learned this the hard way. I built an AI company called Arcules and grew it from 3 people to 150 across 5 countries, real machine learning and computer vision at scale, before ChatGPT was a dinner-table word. We exited at nine figures. Then I watched the market commoditize the very thing we had built. I have stood on both sides of the AI commoditization curve. The lesson from both sides is the same. The tools level out fast. The judgment does not.
So this post gives you the judgment, packaged as something you can actually run. Let me be direct. This is the prescription from my book, not a summary of it.
Why a CEO needs an AI operating system, not a checklist
A checklist tells you what to do once. An operating system tells you how to decide every time. That difference is everything.
Here's the problem with checklists. AI moves too fast for a static list. The tool you picked this quarter is cheaper and weaker than next quarter's. A checklist freezes you at one moment. An operating system travels with you, because it is about how you think, not what you bought.
And the data says thinking is the bottleneck. MIT studied corporate AI and found 95% of initiatives fail to turn a profit. The other 5% win. Same tools. Same models. Same market. The only variable that moves is the judgment of the person placing the lever. That is not a technology gap. That is a leadership gap.
So stop asking which AI to buy. Start asking where to place it and what to leave alone. That is what an operating system does for you.
What is the AI MOMENTA framework?
MOMENTA is a weekly decision lens I built over more than nine years of running AI inside a large company and as a founder. It is not a project plan. It is the set of questions I run before any AI move, and the set I teach the CEOs I work with to run for themselves.
Picture AI as a force multiplier. One unit of effort in, ten units out. That sounds great until you remember a multiplier works on whatever you point it at. Point it at the wrong thing and you multiply the wrong thing. Tenfold. Fast.
So the whole game is placement. Where you put the lever, and just as important, what you refuse to multiply yet. MOMENTA is three questions, answered in this exact order.
1. Where does leverage actually compound here?
Not where can we use AI. Almost anything can use AI. The real question is narrower. Where does a small input create an output that keeps paying off?
Compounding is the word that matters. A one-time time-saver is fine. A place where the gain feeds the next gain is gold. Think of a process that touches many decisions downstream, or a bottleneck that slows everything behind it. Fix that with a multiplier and the whole system speeds up, not just one task.
Most leaders point AI at what is visible. Point it at what compounds instead. That is the first discipline.
2. What must NOT be multiplied yet?
This is the question almost nobody asks. And it is the one that separates the 5% from the 95%.
A force multiplier does not care if your process is good or broken. It scales both. So if you point it at a workflow with bad data, a fuzzy decision, or a step you do not actually trust, you do not get ten times better. You get ten times more of the mess, shipped faster, with more confidence than it deserves.
Restraint is the skill. Before you multiply anything, you ask what is not ready. Bad data. A judgment call that still needs a human. A process you have not cleaned up. Those stay off the multiplier until they are fixed. Saying not yet to the wrong thing protects you more than saying yes to the right thing.
Most leaders get this wrong. They treat AI as gas on every fire. The winners treat it as a lever they place with care, and they keep their hand off the things that should not scale yet.
3. What decision will this let us revisit faster?
The third question reframes every AI move as reversible and informational. You are not betting the company. You are buying a faster look at a decision you will make again.
Ask it plainly. If we do this, which decision do we get to revisit sooner, with better information? AI that shortens your learning loop is worth more than AI that automates a task, because speed of learning compounds across everything else you do.
This is also how you kill the fear. When every move is treated as reversible and informational, you stop waiting for certainty. You run the move, you learn, you adjust. Fast and 90% right beats slow and 95% right. Especially when the move was reversible the whole time.
Run those three questions in order, every week, and placement stops being luck. It becomes a practice. That is the operating system. For the strategy that sits above it, see AI Strategy for CEOs, and for the wider map, the hub on AI for CEOs.
How CEOs use gen AI for strategic planning with this lens
This is where the question I hear most often gets answered. How are CEOs actually using gen AI for strategic planning? The honest answer is that the tool is the easy part. The lens is what makes it strategic instead of busy.
Run a planning question through MOMENTA. Where does leverage compound? Maybe it is scenario work, where one prompt lets you pressure-test five versions of a plan in an afternoon instead of a week. What must not be multiplied yet? The final call. AI can draft the options and stress the assumptions. It should not own the decision, because the data behind your market is yours and the accountability is yours. What does it let you revisit faster? Your assumptions. You can re-run the analysis the moment a number changes, so your plan stops being a once-a-year document and starts being a living thing you revisit on demand.
That is the difference. Gen AI in planning is not about a slicker deck. It is about shrinking the loop between a changing reality and your next decision. The lens keeps you pointed at that, instead of at whatever the tool makes easy.
The starting loop: run this in one quarter
Frameworks are nice. Reasons aren't results. So here is the loop to run this quarter, before you read another word about AI. It is the same motion I teach every CEO I work with, and you can start it on Monday.
- Pick one process that is slow, expensive, or inconsistent. Just one. Resist the urge to transform everything. You want a single real problem you can feel.
- Name the outcome with a number. Not "make it better." Cut this from 10 days to 3. Raise this from 60% to 90%. A number turns opinion into a result you can check.
- Give one person ownership. One person, not a committee. Committees produce meetings. An owner produces an outcome. They run the MOMENTA questions on the problem and place the lever.
- Run it a few weeks. Short loop, not a roadmap. Long enough to see signal, short enough that killing it costs you little.
- Review honestly. Did the number move? Say it straight. No credit for activity. The whole point is to learn, and you only learn if the review is honest.
- Expand or kill. If it worked, point the multiplier at the next compounding spot. If it did not, kill it without drama and try a better placement. Either way you got smarter, because the move was reversible and informational from the start.
Do that three times and you will know more about AI in your business than a year of vendor demos could teach you. Not because you learned the tools. Because you built the judgment.
That loop is the operating system in motion. One real problem, one owner, one short loop, one honest review, run on repeat. It is how scattered AI activity turns into compounding advantage.
What the winning 5% actually do differently
Let me bring it back to the number, because it is the whole argument. MIT found 95% of AI initiatives fail to turn a profit. The 5% who win use the same tools as everyone else. So the difference cannot be the technology. It has to be the judgment.
Here is what the 5% do. They place the lever where leverage compounds, not where it is visible. They keep their hand off what should not be multiplied yet, especially bad data and decisions that still need a human. And they treat every move as reversible and informational, so they learn fast instead of betting big and praying.
That is MOMENTA, lived rather than listed. It is not a smarter tool. It is a leader who decides better, every week, until better placement is just how the company runs.
You can run this on your own. The full playbook, with the worked examples and the way I coach CEOs through their first placements, is the next step inside the AI Executive Mastermind, and the complete thinking sits in the book, AI Leadership Mastermind.
Frequently asked questions
What is an AI playbook for CEOs?
An AI playbook for CEOs is a decision lens you run repeatedly, not a list of tools or tasks. The point is placement judgment: deciding where AI creates compounding leverage and what you should not multiply yet. Mine is the MOMENTA framework, three questions a CEO answers every week to turn scattered AI activity into compounding advantage.
What is the MOMENTA framework?
MOMENTA is a weekly decision lens built over more than nine years of running AI as a Canon executive and a founder. It asks three questions in order: where does leverage actually compound here, what must NOT be multiplied yet, and what decision will this let us revisit faster? It treats AI as a force multiplier and every AI move as reversible and informational.
How are CEOs using generative AI for strategic planning?
The best use is shrinking the loop between a changing reality and the next decision, not producing a slicker deck. CEOs use gen AI to draft and pressure-test scenarios fast, while keeping the final call and the accountability with a human. Run through the MOMENTA lens, it lets you re-run your assumptions the moment a number changes, so the plan becomes a living thing.
Why do most CEO AI initiatives fail?
Because the judgment is missing, not the technology. MIT found 95% of AI initiatives fail to turn a profit while 5% win with the same tools. Most failures come from pointing a force multiplier at the wrong thing, often a broken process or bad data, which scales the mess instead of fixing it. An operating system like MOMENTA closes that judgment gap.
