The difficulty is that the founder’s job is changing too. You need to understand what AI makes possible, then decide what that means for your customers, your people and the way the business works. Your technical team can help you build what you choose. They cannot choose the future of the company for you.

Your IT team is not coming to save you.

Neither is the person you hired to “handle tech”. Neither is the shiny consultancy deck that promises a “transformation” while your operating rhythm stays exactly as it was.

I keep meeting founder CEOs who have outsourced the AI problem in their heads. They think the technical person will figure it out. They nod in leadership meetings. They fund a pilot. Then they wonder why nothing material has changed in how the company makes money.

That’s not ownership. That’s waiting for someone else to do the founder’s job.

And you will struggle to make those choices if every conversation about AI depends on somebody else’s explanation. You have to learn enough to form a judgement of your own.

Learn from the minority getting results

McKinsey’s State of AI survey, published in August 2026, shows how uneven the results are. Eighty per cent of people using AI say it has made them personally more productive. Yet only about 6% meet the report’s definition of a high performer: they attribute at least 5% of EBIT to AI and describe its impact as significant. EBIT is earnings before interest and tax.

These businesses are seeing the benefit in their results. That should make you curious about what they have worked out.

Great results are skewed. You see it in sport, in sales and in business: a small minority get extraordinary results while the vast majority get something much more ordinary. When that happens, you can spend your time explaining why the exceptional cases don’t apply to you, or you can learn from them.

With AI, the difference takes you straight to leadership. Making a task quicker is useful. Turning that improvement into more customers, better service or higher profit requires decisions about how the business runs. Somebody has to choose where the effort goes, resolve the problems between departments and decide what happens to the time people save.

The same McKinsey report finds that high performers are twice as likely to say their senior leaders demonstrate commitment to AI. They are also more likely to redesign workflows fundamentally. Those are changes in how a company is led and organised, and they explain why handing AI to the CTO is such an inadequate response.

You can buy the same tools as the businesses doing well. The more interesting question is whether you are prepared to lead the changes that make them useful.

Know enough to judge what you are being told

I want you to build something with AI yourself. Choose a job you already do, perhaps preparing for a customer meeting, and try to make it easier. Get help from someone who knows more than you, but stay involved enough to understand what makes the thing work.

You will learn more from trying to use it on a real task than from watching a polished demonstration. The demonstration tends to leave out the moment when a convincing answer turns out to be wrong, or when ten minutes saved producing something becomes twenty minutes spent checking it. Finding those problems yourself gives you a much better basis for judging the next proposal.

In 2026, I would happily see you spend a reasonable chunk of your CEO time learning about AI and applying it to the business. If it could change how customers find you, what you can offer them or how quickly you can deliver, understanding it belongs in your diary.

The Two-Day-Week CEO Blueprint™ protects three days a week for the work only the CEO can do, with two for running the business. Exploring an opportunity that could reshape the company is a good use of those three days.

The difficulty comes when learning quietly turns into doing everybody else’s work. You build one useful automation, somebody asks for another, and before long the easiest way for any department to get its AI problem solved is to bring it to you. You have given the company a new reason to wait for its founder.

Keep asking whether your involvement is helping you make an important decision or whether you are maintaining something a capable colleague could own. As with moving from the seven-day grind to the two-day CEO, you need to develop people who can carry the work as you move on.

Where could AI change how you make money?

The first client who rang me from Israel said, “I found you on ChatGPT.”

A year earlier, none of our inbound leads had come from AI search. In 2026, all of them have. That first call made me want to understand what was happening, and I have become obsessed with learning how people find us through AI and making the changes that help them do it.

We have just had our best month ever. Without that work, the majority of those opportunities would not have come through our front door. Spending time on this has been a good use of my week running Monkhouse & Company, and I would happily spend a reasonable chunk of it doing more.

Your opportunity may be somewhere else entirely. That is why I would begin with the business: what is your number one constraint right now?

Perhaps you do excellent work but too few of the right customers find you. Perhaps the enquiries are there, but the sales team struggles to turn them into business. Work might be piling up because you cannot deliver quickly enough, or you might be waiting for people you cannot recruit.

Find the biggest constraint, then work out what today’s tools could do to make a significant difference to it. That gives your learning a purpose and a way to judge whether the effort is paying off.

It also gives you a reason to question things you have always accepted. What has always seemed impossible in your business, or even across your entire industry?

If customers have always waited 72 hours for an answer, what would happen if you could give them a useful answer in 72 minutes? Or 72 seconds? What would that do to their experience, their willingness to buy and the work your team has to do afterwards?

An outrageous result like that would be worth putting your time and energy into, wouldn’t it? You need enough understanding of the tools to see the possibility, and enough ambition to ask whether the old limit still needs to exist.

Follow the work far enough to find the problem

Once you have chosen the constraint, resist the temptation to stop at the first task you can automate.

Take proposals. Producing a first draft faster sounds useful, until you realise that it still waits three days for pricing approval, promises something delivery cannot fulfil and needs rewriting after the next customer call. You have saved some typing time while leaving most of the delay intact.

Follow the enquiry all the way through to signed work. You begin to see questions about which customers to pursue, what the team needs to know before quoting and who can approve the price. You also see whether delivery finds out what has been promised early enough to do anything about it.

AI might help qualify the enquiry, find relevant examples, produce a proposal or spot missing information. Getting the benefit means changing the decisions and handovers around those tasks. If sales, finance and delivery cannot agree on how the work should move between them, the founder has a leadership problem to resolve.

McKinsey’s article on the CEO’s role in making AI succeed makes this responsibility explicit. The CEO must choose the business priorities and make decisions that reach across functions. A technical team cannot settle those choices on your behalf.

The same thinking applies when delivery is the constraint. Better estimates, earlier warnings about overruns or fewer mistakes in the handover may let you finish work sooner. Choose an improvement the customer will notice, then track quality alongside speed so you know whether you have achieved it.

Somebody needs to own that result, with enough time and authority to change the work. Review what happens to the constraint, rather than how impressive the latest demonstration looks. The five core principles of successful scale-ups apply here too: agree the priority, put somebody in charge and review the result.

Put what your business knows to work

As you follow a problem like that, you will discover how much useful knowledge the company has and how little of it is easy to find.

Your competitors can get access to the same models. They don’t automatically have your record of which proposals won, why those customers chose you, which estimates were accurate or which handover mistakes cost you money. That experience can help you make better decisions, provided you can find it and learn from it.

For God’s sake, please tell me you’re transcribing all your meetings and phone calls, and all your sales team’s calls. That should be the minimum.

Personally, I use Granola. Its killer feature for me is that it works on my Apple Watch, so I can record an in-person meeting without getting my phone out and putting it on the table. It makes it easier to keep a record of conversations I would otherwise be trying to remember later.

The record gives you something to work with. You can go back to what a customer actually said, compare it with what sales understood and see whether the proposal answered the problem they described. Across the sales team’s calls, you can look for questions they handle well and the ones they keep avoiding.

There is still judgement involved. People need to check what the system has understood and correct it when it is wrong. But you can now work with the conversation itself, rather than whichever parts somebody remembered to put in the CRM.

Decide what the time saved is for

Suppose your team can do the same work in less time. What do you want to happen next?

You could take on more customers, improve service, develop a new offer or reduce costs. Each choice asks something different of the rest of the company. Faster sales can overwhelm delivery, while cutting costs too quickly can leave customers waiting. If the aim is better service, people may need to spend the time they have saved with customers.

In a business with 30 to 250 people, there is plenty to explore in the work nobody enjoys: re-keying information, chasing updates or assembling a first draft. People are more likely to help you change it when they understand what you intend to do with the extra capacity.

If every conversation about efficiency immediately becomes a conversation about headcount, you give them a good reason to keep quiet. You are asking people to volunteer evidence against their own jobs, then wondering why they aren’t excited to show you what they have found.

Of course jobs will change, and you need to be honest about that. You also need to understand what else a task was doing before you remove it. If producing a first draft helped a junior colleague learn, where will they now get that experience? Who will help them develop the judgement the business still needs?

These choices belong with the leadership team because the consequences reach across the company. Letting each department improve its own numbers without considering the others can make the whole business harder to run.

Make curiosity part of the working week

Your people will judge how seriously you take AI by what happens when somebody brings you an idea.

If you ask to see how it works, give them time to improve it and listen to what they have learnt, they have a reason to try again. If you tell them to take it to IT, they have learnt something about you instead.

BCG’s 2025 AI at Work research found that only about a quarter of frontline employees felt they received strong leadership support for AI. What would your people say about you?

Mediocrity is invisible until passion shows up and exposes it.

When somebody finds a better way, a task that supposedly couldn’t change has changed. The process everybody tolerated starts to look unnecessary. That can be exciting, but it can also be uncomfortable for the people who have spent years explaining why it had to work that way.

You set the tone for what happens next. Make room in the week for people to try things, ask what happened and help useful ideas reach everyday work. Give credit to whoever found the improvement, including when they are junior to everybody else in the room.

Apply the same standard to your own learning. A regular session with someone who can teach you is useful if you bring real work and expect to do some of it yourself. You are showing your team that learning continues at the top of the company, which is part of building a Scaling Up mindset.

Help the company learn without waiting for you

If you become the only person who understands AI, important decisions will keep coming back to you, however much automation you introduce.

Each leader needs enough understanding to make decisions about the work they own and recognise when a problem reaches into another department. Give them responsibility for learning and applying it, then review the business results alongside cash, pipeline and delivery.

Keep learning yourself while developing people who can carry the work with you. The aim is a company that gets better at making decisions as its understanding grows, rather than one that needs the founder in every AI conversation. It is part of building a business that can move without you.

Do you want to learn from the minority, or bumble along with the majority?

Frequently asked questions

Can I hand AI to IT or my CTO?

Your CTO can take responsibility for the technical work. You still need to choose which business problems matter, where to invest and what you want to change for customers and staff. Agree who owns each decision, so a useful project does not stall between departments.

How much time should I spend on AI personally?

Enough to understand the possibilities and act on the opportunities that matter. In 2026, that may take a reasonable chunk of your three CEO-only days. Keep asking whether your involvement helps the business make progress, and develop people who can take over work that no longer needs you.

Where should we start if we’re dabbling?

Start with your number one constraint. That might be finding customers, converting sales, delivering work or recruiting. Look at the whole process, choose a result you can measure and give somebody responsibility for it. Judge the effort by whether that constraint begins to move.

What if only I understand AI in the leadership team?

Make learning part of each leader’s job. Give them time, help from somebody who knows more and a real problem to work on. Review what changes in the business as they learn. The company needs several people who can make sound decisions without waiting for you.

What should you do next?


Choose the constraint you want AI to change and the result you want to achieve.

Four ways to take this further

  1. Book a call. Talk to Dominic about the constraint you want to change and the decisions your team needs to make. No obligation, no pitch.
  2. Read the book. Mind Your F**king Business explores how to develop a leadership team and build a company that can scale without you in every room.
  3. Explore the Two-Day-Week CEO Blueprint. Make space in your week for the choices only you can make.
  4. Subscribe to the newsletter. One useful tool each week for running and scaling your business.

Your move. Put your number one constraint on the next leadership-meeting agenda and decide what today’s AI tools could change.

About the author

Dominic Monkhouse is the founder of Monkhouse & Company. He coaches founder CEOs of UK and European scale-ups on Scaling Up and leadership teams. He was previously Managing Director of Rackspace UK and Peer 1 Hosting.