A multi-year plan is set at one point in time and then carried out for months on the basis of assumptions that were correct at that moment. Halfway through that term, there is no natural trigger to look at those assumptions again. There is no annual planning process open to it, no shareholders' meeting asking for it, no quarterly figure pointing at it. The plan simply continues to run. That is exactly what makes this moment difficult: the question of whether the assumptions still hold does not surface on its own, while the time that has passed has become long enough to make a difference.
With AI, that passage of time is relevant, because what a task can or cannot handle is not a fixed property. A task that was still entirely human work twelve months ago may now fall under the second type: partly taken over, with someone approving or rejecting the outcome for a reason. And a task the plan labelled as "human work for years to come" may by now have largely been taken over. Not everywhere, and not at the same pace — which is exactly why the interim period is well suited to taking a look.
When the plan was drawn up, the assumptions about AI were partly an estimate: what would be a reasonably workable scenario for the coming years. Halfway through, that is no longer necessary. By then, it is simply possible to see what has happened: which tasks within the company are now performed by AI, which fall under supervision, and which have remained unchanged human work. The difference between then and now is the difference between an assumption and an observation.
That difference does not arise at the same pace in all companies. In some organisations, that shift is already visible on the work floor — a team that no longer does part of its reporting work itself, a process in which a review step has been added in place of an execution step. In other organisations, with comparable plans and comparable markets, little has changed so far. That difference is rarely found in the technology itself. It lies in what has been picked up, tried out and implemented within the company, and that varies by organisation — even within the same sector.
Halfway through a term, it is possible to establish which assumptions in the plan still match what is happening within the company and in the market, and which have quietly become outdated without anyone recording it. This does not require a new strategy or a revision of the plan. It requires an assumption list: for each assumption, noting when it was set, whether it still matches today's picture, and if not, what has changed in the meantime. This is not a judgement on the plan, but a check on the foundation on which the plan rests.
It is also useful to place the outside view and the inside view side by side. From the outside: how sector data, regulatory clocks and signals about what competitors are doing have developed since the plan started. From the inside: how the company's own management team assesses the current situation, independent of what the plan assumed at the time. Differences between those two views are themselves already information — they show where the assumption and the observation have drifted apart.
At this point, it is not possible to establish what will still happen during the rest of the term. A mid-term assumption check describes a current state, not a prediction for the remaining period. Nor can it be established whether a task that is still human work now will still be so in a year's time — that depends on developments that cannot yet be seen. And it is explicitly not a basis for personnel decisions: if an assumption about work being taken over has consequences for roles or staffing levels, separate legal requirements apply to that, independent of this check.
What can serve as a guide for how this moment relates to other checkpoints during the year is described in the explanation of how you test the same assumptions before a shareholders' meeting, during annual planning and after a disappointing quarter — each moment calls for a different kind of check, even though it concerns the same underlying assumption list.
All these moments revolve around the same question: which work in this company can now genuinely be taken over by AI, which work falls under supervision with review, and which work remains human work. That question is answered per task with the work scan from FTE TO AI. Anyone who would rather first look at the distinction between a plan and a strategy — and why a plan becomes outdated faster than the strategy behind it — will find that explained on the page about the difference between a strategy and a plan.
At this point, you can place the assumption list of your current plan alongside the current state, without revising the plan itself. A first step in this is the free assumption check: a short round in which you name your key assumptions and see, for each one, when it was last confirmed. The full proof, with the outside view alongside your management team's self-assessment, 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.