An operations director steers on capacity: how much work there is, how many FTEs are allocated to it, and what turnaround time goes with that. That planning rests on assumptions about who does which work. As long as those assumptions hold, the planning holds. The problem is not that AI undermines those assumptions. The problem is that the assumptions lapse silently while the planning remains intact. A task that cost three days last year now costs an afternoon, with human oversight that approves or rejects with reason. Nobody has adjusted the planning, because nobody has formally withdrawn the assumption.
That is the core of what is changing: not that a project is running under the name AI, but that somewhere in the operation a task changes category. From human work to oversight with reason. From oversight to full takeover. In one team that is already the case, in another not yet, and the difference rarely lies in the technology. It lies in who took the trouble to test the assumption.
The operations director's question is not whether AI is taking over work. That question has been answered: it is happening, in parts, unevenly distributed across departments and processes. His question is more specific: which task in this company, at this moment, falls into which category. Can AI take over the task, is oversight with reason required, or does it remain human work. That classification determines whether the current FTE deployment still matches the work that exists, or whether capacity is freed up somewhere that is needed elsewhere.
The answer he does not accept is an estimate at sector level. A percentage cited elsewhere in the sector says nothing about the process that has been set up this way in this company. Work that is identical on paper differs in practice in data quality, in exceptions, in who bears responsibility when it goes wrong. A statement without that precision is not an answer, it is an assumption still waiting to be tested.
The companies where this shift is already visible in the figures have done one thing that others have not yet done: they have broken the work itself down into tasks, and determined per task which category applies and when that was last checked. Companies where the planning still runs on last year's assumptions have not done that work, usually not out of unwillingness but because nobody owned that check. It is not an IT task, nor an HR task alone: it is an operational question about what work actually consists of now.
The underlying question — which work in this company can genuinely be taken over by AI — is answered task by task with the work scan from FTE TO AI, with a category and a date attached to each.
To gain: capacity that is freed up, demonstrably and per task, to be deployed wherever the operation needs it. Freed-up hours are not an abstraction for this role; they are the raw material of every workforce plan. To lose: credibility, if the planning turns out to rest on assumptions that have not held for months and this comes to light during an audit or a reorganization. What an employer then does with that capacity is not up to this role; that is governed by its own legal requirements, and those do not belong in a planning document.
The HR director looks at the same shift from a different angle — what an HR director looks for when AI takes over work shows that this role primarily safeguards the conversation with employees and the duty of care, while the operations director safeguards capacity. Those two angles clash the moment one speaks of hours and the other of people: the same number, two meanings. The supervisory board member asks yet another question, more at a distance and focused on risk rather than planning, as described in what a supervisory board member looks for when AI takes over work. And when a private equity partner is watching, freed-up capacity quickly counts as value creation on paper, while the operations director knows that capacity only has value if the assumption behind it still holds — a difference worked out in what a private equity partner looks for when AI takes over work.
That clash is precisely why our management team disagrees on priorities, what now is a question that comes up at the table more often than a strategy document would suggest: not because the management team disagrees about direction, but because each role holds a different assumption to be true.
The first step is not a reorganization and not a project. It is checking which assumptions carry the current planning and when they were last verified. That can only be done with the right people at the table; who should take part in a strategy review describes how that group is put together so that the operational perspective is not missing alongside the financial and personnel perspectives.
Anyone who wants to start without immediately setting up a full process can take the free assumption check: a short round in which you name your key assumptions and see, per assumption, when it was last confirmed. The full strategic pressure test, with sector data, regulatory clocks and a self-plot of the management team side by side, is under construction.
Stel uw vraag. Vaak zit de echte vraag een laag dieper — daar mag ik naar vragen.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.