If you are a CEO worrying about what your individual AI agents are doing, you have gone back to micromanaging, and it will fail for the same reason it always has. The job is not to stop paying attention. It is to pay attention at the right altitude: delegate the trench, and own the operating model.

The statistic that gives it away

CEOs are now directing more than half of this year’s AI investment into agents, and 72 per cent say they are the primary decision-maker on AI, roughly double the year before. Nearly half believe their own job depends on getting AI right. So the most senior person in the company is leaning down into the trench of individual agents. That is the wrong altitude. The blunt version of that mistake, releasing money and effort without direction, is the argument in You Didn’t Lead Your AI Project. You Just Bought Shovels.

The consultancy line for the year is that ROI will be the acronym of 2026. Return does not come from a CEO supervising agents one by one. It comes from a CEO running the company.

Micromanaging versus setting up

Let me pre-empt the obvious objection, because there is a real difference between micromanaging the workforce and setting it up.

When a company hires 500 people, the CEO does not interview all 500 or check each one’s daily tasks. But the CEO does own the decision to build the division, the budget, the risk appetite and the line of accountability. That is setting up the workforce, and it is the CEO’s job. Agents are a new class of worker, a new resource. Worrying about what each individual agent does is the work of a manager, a project lead, a director. Owning the operating model those agents run inside is the work of the CEO. Delegate the first. Own the second. Choosing who genuinely leads that model is its own discipline, which I set out in Who Should Actually Lead Your AI Initiative.

The man at the door

I once knew a CEO who stood at the building entrance every morning at eight o’clock. Not to greet anyone. To scream at whoever arrived after eight for stealing company time, and to write them up over ten seconds. He spent his mornings humiliating latecomers instead of building revenue or meeting a client.

That is precisely what a lot of CEOs are now doing with agents. We have decades of evidence that micromanaging fails, which is exactly why serious leaders stopped doing it. Just because it is called AI, did you lose your focus? The opposite error, handing the whole thing off and admiring it from a distance, fails too, which I covered in The Proud Leader Who Handed Off His Own Future. The point is the altitude, not the distance.

What to do on Monday

Ask yourself one question and answer it honestly. Why am I micromanaging something I already know will not work this way? Treat agents as employees. Run the business: the vision, the future profitability, the client relationships, the traps to avoid. Stop supervising every individual worker, and set the operating model they work inside. If your instinct instead has been to lower the ambition to something safe and mundane, that is its own failure, and I wrote about it in AI Leadership Failure: Set Low AI Goals, Automate the Obvious, Fail Quietly.

Sixty per cent of leaders have deliberately slowed their agent rollouts over fear of errors, and only about 6 per cent say they would scale back if AI failed. The instinct to hover is understandable. It is still the wrong altitude.

While you are at the door with a stopwatch, someone else is upstairs running the company. In the AI era, that is the whole difference.

If you are leading your organisation through this, I work with a limited number of senior leaders each quarter. Get in touch at Anglero.com.

Thomas Anglero is a Strategic AI Advisor, keynote speaker and author of Intro to Artificial Intelligence. He has delivered over 450 keynotes across 30 countries for organisations including IBM, the WHO, the World Government Summit and the European Commission. He founded the IBM Watson AI Lab for Cancer at the Oslo Cancer Cluster and closed over $500 million in enterprise transformation deals as CTO and Chief Innovation Officer at Cognizant.

Frequently asked questions

What are CEOs getting wrong about AI agents?

They are micromanaging them. A CEO worrying about what individual agents do has dropped into the trench, which is a manager’s job. The CEO’s job is to run the company and own the operating model the agents work inside.

Is delegating AI agents the same as ignoring them?

No. There is a difference between micromanaging the workforce and setting it up. A CEO owns the decision to build the division, the budget, the risk appetite and accountability, and delegates the day-to-day supervision.

What does the wrong altitude mean for AI?

It means paying attention to the wrong level. The CEO should be setting the operating model, the vision and the accountability, not supervising individual agents one by one, which is where much of the current effort is wrongly going.

What is the one thing a CEO should do on Monday?

Stop micromanaging agents, treat them as a new class of employee, delegate the trench to managers, and own the operating model. Then return to running the business the AI is meant to serve.

More than half of my leadership team will be AI services, working around the clock for less than $100 per month. This is not an experiment or a stunt; it is the operating model of my AI-first venture, run with AI agents. Here is what the AI seats actually do, what stays human, and why the economics are impossible to ignore.

The model: a leadership team that is half AI

A traditional leadership team is a set of standing functions: someone who challenges your strategy, someone who checks your numbers, someone who reviews what leaves the building, someone who watches the market. Each seat costs a salary, works office hours, and takes holidays.

My plan is for 50%+ of those seats to be AI services. The AI seats work 24/7, providing feedback, reports and analysis at any hour without hesitation, for a combined cost of less than USD $100 per month. They never tire of a question, never need the context re-explained, and never let a document leave unreviewed because it arrived at midnight. This is the foundation of my AI-first venture, run with AI agents as the operating team.

What the AI seats actually do

This is not theory; the seats are already working. My venture’s SEO is 100% done by AI, and the results are steady traffic growth with our most important keywords ranking high in Google, achieved at a speed no human team could match. The AI seats draft and pressure-test documents, produce analysis on demand, prepare positions before negotiations, and act as a standing counsel that challenges my thinking before decisions are made, which is precisely the discipline a real AI initiative demands from its leader.

What stays human

The seats that remain human are the ones the model cannot hold: final judgement, accountability, and relationships. An AI can prepare the negotiation; it cannot sit across the table and build trust. It can flag the risk; it cannot own the consequence. The point of the model is not to remove people, it is to stop paying leadership-team prices for work that no longer requires a leadership-team salary, and to redirect the human seats to the work only humans can do.

Why most companies cannot copy this yet

The barrier is not the technology or the $100. It is culture and mindset. A business only benefits from an AI leadership seat when its leader actually trusts the seat, uses it daily, and leads the cultural change rather than delegating it. Organisations that invest in cultural innovation before AI investment gain the advantage; those that buy the tools without the mindset get an expensive subscription and no leverage. It is also why I speak on When Every Employee Becomes a Leader, Because of AI: the same model that transforms a leadership team eventually reaches every desk in the company.

Frequently asked questions

Can AI really hold seats on a leadership team?

Yes, for the standing functions that are work products rather than relationships: feedback, reports, analysis, document review and strategic challenge. The AI seats work around the clock and never need context re-explained. Final judgement, accountability and relationships remain human seats.

What does an AI leadership team cost?

Less than USD $100 per month for the AI services in my model, against the salary cost of the equivalent human functions. The economics are not an incremental saving; they are a different category of operating model.

What results has this model actually produced?

My venture’s SEO is run entirely by AI and has delivered steady traffic growth with top keyword rankings in Google, at a speed no human team could match. The AI seats also draft, pressure-test and analyse on demand, functioning as a standing counsel before every significant decision.

Why can’t most companies just copy this model?

Because the barrier is culture, not technology. The model only works when the leader genuinely uses and trusts the AI seats daily and leads the cultural change personally. Companies that buy the tools without changing the mindset get a subscription, not leverage.

Thomas Anglero is a Strategic AI Advisor, keynote speaker and author of Intro to Artificial Intelligence. He has delivered over 450 keynotes across 30 countries for organisations including IBM, the WHO, the World Government Summit and the European Commission. He founded the IBM Watson AI Lab for Cancer at the Oslo Cancer Cluster and closed over $500 million in enterprise transformation deals as CTO and Chief Innovation Officer at Cognizant.