A supervisory director is not at the controls. He oversees who is. That makes his position on AI and work different from that of a board member: he doesn't need to know how a model works, but he does need to know whether the executive board knows what it still bases its strategy on. What he stands to lose is not a project that fails. It is the liability of a board that approved a plan whose underlying assumptions had already lapsed at the moment of approval.
What he stands to gain is sharper: a board that asks early rather than checks late retains its grip on a shift that would otherwise happen unnoticed. AI taking over work is not a topic that sits on the agenda as a separate item. It is something that shifts underneath existing assumptions without anyone having formally withdrawn that assumption.
The supervisory director's question is not "what is AI doing in our company." That question is for the executive board. His question is: on which assumption about cost, capacity or lead time was this plan built, and is that assumption still true today. That is a different question from a technology question. It is a question about the durability of decision-making.
He also probes for distinction. Of the work included in the plan: which part can a system already take over, which part happens with human oversight that approves or rejects with reason, and which part remains human work for reasons that don't disappear the moment a supplier releases a new version. Without that distinction, any statement about "AI impact" is an estimate that no one can verify.
A supervisory director does not accept a status update stating that the strategy is "on track" without showing which assumptions have been tested and when. He also does not accept an enthusiastic story about automation that gives no answer to the question of what that means for the FTE capacity the plan assumes. And he does not accept AI being presented as a complete project with an end date, while in practice the takeover of work happens in parts, faster in one company than in another, depending on how structured the work already was before a system looked at it.
Where the topic touches on personnel decisions, that is explicitly a different domain: separate statutory requirements apply there, and these should not be mixed with a strategic assessment of assumptions.
The first clash is with the CEO. The CEO is responsible for execution and has an interest in a story that shows progress; what a ceo looks at when ai starts taking over work shows the pressure under which that role reports, and that is not the same pressure as that of oversight. A supervisory director who only tests the CEO's story for persuasiveness misses the question of whether the underlying assumption still holds.
The second clash is with the shareholder with a shorter horizon. A private equity partner often assesses an AI shift on the speed of cost reduction within a holding period; what a private equity partner looks at when ai takes over work shows that different time horizon, and a supervisory director must know that his own responsibility — continuity beyond that holding period — raises a different set of questions than those of the shareholder at the table.
The third clash is with the bank or financier, who tests cost structure based on ratios that do not always move in step with the speed at which work actually shifts; what a banker looks at when ai changes your cost structure shows exactly where that view rests, and a supervisory director who does not know that difference risks holding two different assessments for one and the same development.
In companies where this topic is well organised, it is not on the agenda as an incident but as a recurring item, with a fixed question: which assumption has been tested this quarter and which has not. How to get that structurally onto the agenda, including for regulation which suffers from the same silent obsolescence, is described in how do you get regulation structurally onto the executive agenda. The role of the board in this is not incidental; what role does the supervisory board play in ai decisions describes where oversight begins and where execution ends.
In companies where this does not yet work this way, AI is usually still treated as a one-off project with a project leader and an end date, rather than as an ongoing shift that must be reassessed per task, per department and per quarter. The difference rarely lies in the technology. It lies in whether someone at the table keeps asking the question structurally.
The underlying question — which work in this company can genuinely be taken over by AI, partly or fully — is answered per task by FTE TO AI's work scan, regardless of which role at the table needs the outcome.
A self-assessment by the executive board next to an outside view quickly shows where the gap lies between what the board assumes and what the sector, regulation or competition is already showing. If you want to know where your own assumptions stand, you can take a free assumption check: a short round in which you name your key assumptions and see, per assumption, when it was last confirmed. The full stress test, with sector data, regulatory clocks and your executive team's self-assessment 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.