From Chief Executive to Chief Wearer of Hats

AI did give you the time back. Watch where it goes: one small, sensible decision at a time, into work that used to belong to somebody else.

A CEO I work with runs a mid-sized company. He has a CMO, an HR lead who is genuinely good at her job, and a law firm on retainer.

He now does a large chunk of all three jobs himself, and he could not tell you when that started.

What changed is not his appetite for work. It is that the agents he has built over the past year know his contracts, his market and the way his company writes. They hand him a serviceable first draft of almost anything, faster than he could write the email asking somebody else for one. A contract arrives from a customer, and he can have a decent read on it before lunch or a careful one from the firm on Thursday. The campaign copy comes back fine, but he can have a version he likes before the meeting. The job advert needs one pass, and the pass is quicker than the conversation about the pass.

So he stopped waiting. Not as a policy, and not all at once. Each of those was a Tuesday.

Somewhere around the third one he looked up and asked the only question worth asking. Hang on. When exactly did I start doing everyone else's job?

The time you were promised

Here is the part nobody warns you about. The time saving is real. You did get the hours back, exactly as advertised.

The question is what a high achiever does with a windfall of free hours, and the honest answer is that he spends it immediately. Not on rest, and not on the strategy offsite. He spends it on work that is suddenly within reach. The contract he can read himself, the copy he can sharpen, the analysis he would have waited a week for. Every one of those is a good use of an hour. Taken together they are how a chief executive becomes a chief wearer of hats.

It happens one small decision at a time, which is why nobody notices. There is no meeting where you resolve to take on marketing. There is a Tuesday where the pass takes nine minutes and the conversation about the pass would take twenty, and you take the nine.

Picture the street performer with the bass drum strapped to his back, cymbals between his knees and a harmonica on a wire in front of his mouth. He is playing all of it. He is playing none of it especially well. He is also, by some distance, the highest-paid musician in the band.

The second thing that happens is quieter. The standard rises underneath you while you carry it, because the memo that was thorough enough last year is thin now that everyone's memo got deeper. So the hours you saved get spent twice: once on new work, and once on doing the old work to a higher bar. This is why the CEO who is furthest along on his own AI journey is very often the one with the fullest calendar in the building.

The hats do not come off by themselves

Because every single step of it was correct.

Each hat was defensible on the day you put it on. It was faster than the alternative, cheaper than the invoice it replaced, and there is no single one you could point at and call the mistake. That is what makes a run of individually correct decisions so hard to reverse. It helps that the savings looked excellent on a slide, and that no board has ever asked a chief executive to explain why he stopped doing something the numbers liked.

So it does not end on its own. Somebody has to decide it ends.

Your CFO still works for you

There is a particular flavour of delusion available to a chief executive with a good agent, and it is worth naming plainly. You are not now better at marketing than your CMO. You are not reading the monthly numbers better than your CFO. You are producing a passable version of their work more conveniently than they can deliver theirs, which is a different claim entirely, and a much smaller one.

The obligation runs the other way. If you are serious about your own AI journey, the job is to make sure the people around you are equally serious about theirs. Your CFO should be serving you better this year than last, because she has the same tools you do. A CEO who quietly absorbs the first draft of everything has let the organisation off the hook, and paid for it with the most expensive calendar in the company.

Companies still work the way they always did. Clear hierarchy, clear separation of responsibility, everyone accountable for their own patch. None of that was repealed. What changed is only how much each person can now carry inside their patch, which is a reason to raise what you ask of them, not to start doing it yourself.

Alexander Norén, an economics commentator who has made AI documentaries for Swedish television, put me onto the useful distinction here. He points at Daron Acemoglu, who won the economics prize in 2024 and argues for what he calls pro-worker AI. The test is a simple one. Does this make the people already doing the work more powerful at it? That is a choice, and it is made one Tuesday at a time, in exactly the same way the hats accumulated.

The only direction worth going

So there is a fork, and Alexander puts it plainly. Either fewer people end up doing more, or the good people you already have go and do something that was not possible before.

Insourcing is the downstream move. You reach beneath your own role and take work off the people below you, because with an agent it is finally easy. It is satisfying, it is measurable, and it shows up in the cost line, which is why it is the direction almost everyone drifts in without ever choosing it.

Upstream is the harder question, and Alexander asks the biggest version of it. When I put the impact-per-hour argument to him he was unimpressed, and said so. That is scratching the surface. The question that should be surfacing is existential: would we exist today? Would we be organised the way we are, serving the customers we serve, if we were starting now, with these tools? Nobody delegates that question, and it never lands on a calendar by itself.

For a chief executive that is the whole job, and most of it has been quietly out of reach. The bets too complex to evaluate properly, so they got decided on instinct and a two-page paper. The scenarios nobody had time to model. The acquisition thesis that never got past a spreadsheet. There is a version of your role that does all of that, and until recently it was mostly theoretical.

The same gap sits in every seat. There is an A-player version of every title, and it is nearly always the proactive half of the job rather than the reactive one. The CFO who stops assembling the monthly pack and starts running the capital-allocation options nobody had the hours to build. The head of sales who stops reviewing pipeline and starts designing the experiments that tell you which segment is actually worth entering. Alexander's own example is not an executive at all. It is the electrician in front of an installation he has never seen, who used to call a colleague and lose half a day and now solves it on the spot. Same person, same title, materially more powerful inside it.

None of those are new jobs. They are the version of the existing job the good ones always did in the gaps.

Which is why I think this is the more interesting story about AI and work. The common version is that you get through yesterday's job faster, absorb a bit more of it, and wait for the headcount question. The version worth wanting is different. A chunk of your calendar comes back, you spend it on the harder and slower work one level up, and you come out of the year a more capable operator than you went in.

So ask it of your own week, and then ask it of everyone who reports to you. What could you do now that you could not do before? If the honest answer is that you have mostly become faster at other people's jobs, the time did arrive. You just spent it going the wrong way down the org chart.