The AI decision a chief executive cannot delegate

A leader can delegate the technology but not the judgement: which parts of the business AI should touch first, what to stop doing, what to say to the workforce, and how much to spend before there is proof. AI coaching closes the leader's own knowledge gap so those decisions can be made rather than deferred.

Almost every senior leader has now sat through an AI vendor pitch, read the headlines and been asked by a board or a regional office what the plan is. Rather fewer have had a single hour in which they could ask a basic question without it costing them something in front of their own team.

That is the actual constraint, and it is not intelligence. It is exposure. The people being asked to set a direction on AI are the people with the least unstructured time in the organisation, and the material available to them is written either by vendors with something to sell or by commentators with nothing at stake.

What cannot be handed to somebody else

Plenty of AI can be delegated. Tool selection, integration, security review and the mechanics of a pilot all belong with people who do that work. Four things do not move off the chief executive's desk.

• Which parts of the business this touches first, and which are deliberately left alone this year.
• What the organisation stops doing. AI programmes that only add work fail quietly, because nobody has capacity for the new thing.
• What is said to the workforce, and when. Left too late, the rumour arrives before the position does.
• How much is spent before there is proof, and what evidence would justify spending more.

None of those are technical questions. All four are harder to answer if the person answering them has never used the tools on their own work.

Why the market has moved to coaching

Consulting produces a recommendation, which the leader then has to carry, defend and update without the consultant. With most subjects that is fine. With AI it ages badly: the capability moves every few months, so the ability to think about the decision outlasts any particular answer to it.

The Adecco Group's July 2026 whitepaper on AI and hybrid labour markets makes the same point structurally. It finds that adoption diffuses quickly at the consumer level while organisational deployment lags, because of workflow rigidity, governance constraints and the complementary investment required. The gap between what AI can do and what a firm can absorb is where the value sits, and closing it is a leadership job rather than a procurement one.

The same paper is blunt about uncertainty: it sets out four plausible futures rather than a forecast, and observes that which one arrives depends less on what the technology can do in principle than on the choices firms and institutions make. A leader cannot outsource a choice like that to a deck.

What it looks like in practice

Work of this kind is usually one to one, sometimes with a small group of three or four leaders from the same business who can be candid with each other. That second shape works best where the real problem is that the top team is not aligned rather than that any individual lacks knowledge.

The ground covered is normally the landscape in plain terms, which tools are worth this particular organisation paying for, where AI changes its economics specifically, a first plan with a budget and stop conditions, and how to scale what works past the first team. That last one is where most AI programmes stall.

Because these are conversations rather than workshops, they work as well remotely as in a room, which is why this kind of work is rarely bounded by geography. We coach leaders online across the region and beyond, and in person in Ho Chi Minh City, in Vietnamese or English.

One test worth applying

Whoever you work with on this, ask what they run themselves. Advice about AI from someone who has only advised on it tends to be optimistic in exactly the places it should not be: about how quickly people adopt, about what quietly gets abandoned, and about what costs more than it saves. We have been running AI inside our own business for two and a half years, including the parts that did not work.

The whitepaper cited is The Adecco Group, On the threshold: AI, the future of work and the rise of hybrid labor markets, July 2026.

Back to all insights