Before a board decides to do something with AI, employees are often already doing it. Someone in administration has a language model write draft texts. An analyst uses AI to produce a first version of a report. A team leader has summaries generated that used to take an hour. This does not happen as a project with a start date. It happens because it works, and it happens unnoticed by the layer that manages the strategy.
That is the signal: not whether AI is being introduced, but how much of the existing work is already shifting in practice from fully human work to work that is partly or largely done by AI. A strategy that assumes a task costs x hours, or that a team of a certain size is needed to deliver a service, rests on an assumption that may already have been altered by this kind of use, without anyone having established that.
It does not sit in a single system. It is spread across several sources, each of which gives part of the picture:
None of these sources on its own gives a complete picture. Together they show whether a task is shifting between the categories "AI takes over", "AI with human oversight", or "remains human work", and in which direction.
The speed of this shift differs strongly per company and per task. For some tasks, little changes, year after year. For other tasks, the situation can be different within a quarter, for example because a new version of a tool has just crossed the threshold at which oversight becomes unnecessary for part of the work. Fixed annual cycles do not match that pace. A fixed review frequency works better than a fixed annual moment: a short, recurring moment at which it is checked whether the assumptions underlying the strategy still hold, separate from the annual plan.
Whether that is quarterly, or more often for tasks that move quickly, depends on how much of the primary process touches work that AI can take over in parts. A company where that is limited can manage with less frequent checks. A company where that is substantial sooner sees a difference between companies that already track this and companies still running on the old cycle: that difference rarely comes from unwillingness, usually from the absence of a fixed review moment.
If it turns out that a task that was fully human work last year is now partly done with AI and human oversight, that means the assumption underlying the strategy on that point is no longer confirmed. It means the capacity calculation, the turnaround time, or the cost price that was based on that task needs to be reviewed again before a decision continues to rest on it.
It does not automatically mean that fewer people are needed, that a role disappears, or that a reorganization follows. What an employer does with its staff falls under its own legal requirements and its own judgment; those requirements do not follow from a shift in AI use and are not addressed here. The signal says something about the work, not about the people currently doing it. It is also no guarantee that a task will actually be taken over: some tasks remain human work despite technical possibilities, for reasons that lie outside the technology.
The decision tied to this signal is not a personnel decision. It is the decision whether or not to revise an assumption in the strategy. Is capacity planning still based on the old picture of the task, while practice has already shifted? Then that is a reason to revise that assumption, separate from whatever else happens with staffing. The same logic applies to assumptions that do not depend on AI use within one's own company, but come from outside: what competitors' AI announcements mean for one's own position, how to track new entrants without staff, and what the difference is between a strategy and a plan when it comes to which assumptions are actually kept up to date. What customers themselves are going to do with AI also touches on the same question, worked out on the page about what customer use of AI means for one's own strategy.
The underlying question — which work in this company can genuinely be taken over by AI, per task, with the degree of oversight that goes with it — is answered with the work scan that determines per task whether AI takes it over, does it partly with oversight, or leaves it as human work.
A board can start today with a simple question to its own team leaders: which task are we doing differently this year than last year because of AI use, and who noticed that before it became a decision. That conversation alone already exposes which assumptions in the strategy still rest on the old picture.
Anyone who wants to test this without first setting up an entire system 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 proof press, with sector data, regulatory clocks, and the own management team's self-plot 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.