When you ask what it costs to make strategy and governance ready for AI, you are actually asking something else: where is the work that has not yet been done, and how much time from whom will that take. That is not a question that can be answered with an amount before it is known what exactly the ambition presupposes. An ambition that involves AI work asks different things of a management team than the same ambition without AI. Not because AI changes the strategy, but because the strategy suddenly presupposes decisions that did not exist before: who approves when an AI system makes a recommendation, who is liable if that goes wrong, and at what level is that discussed.
Strategy and governance are ready for AI when a management team means the same thing about three things: what ambition is set, which decisions shift as a result, and who is allowed to make those decisions. In practice, what is often missing is not the ambition but the translation. A vision speaks of "AI-driven decision-making", but no one has recorded what that means for the meeting in which a human still makes the final call today. That gap between vision and target state is precisely what the ambition test looks at: four layers, from vision to concrete target state, tested on whether the organization can already do what the ambition presupposes.
The noticeable difference is not in a new policy document. It is in the meeting itself. In an organization where governance is ready, there is a list of decisions that AI may prepare, a list of decisions that AI may make with human approval afterward, and a list that remains human work because liability or context requires it. In an organization where that work has not yet been done, that question is asked again case by case, usually under time pressure, usually by the person who happens to be in the room at that moment. That difference is not visible in enthusiasm for AI, it is visible in the agenda.
The difference rarely lies in the technology. It lies in whether there has already been an earlier moment at which a management team was forced to make decision rights explicit. Companies that come out of a merger, out of a heavily regulated sector, or that previously dealt with outsourcing, have often already done that exercise once for another topic. For them, the translation to AI is a matter of adjusting the existing framework. Companies where decision rights have always remained implicit are only now discovering that gap, because AI is the first time that "who actually decides this" cannot be dismissed with a pat on the back.
The work lies in five moments at which a management team must speak out, none of which is technical. First: is the ambition itself sharp enough to be tested, or is it still an intention. Second: is there agreement on what the target state concretely means for roles, not only for results. Third: is it clear who may approve or reject which decision, and for what reason. Fourth: is there a way to see whether the organization is getting closer to that target state, regardless of whether it feels right. Fifth: is there a place where objections to the ambition itself can still land, before implementation has already begun. That is exactly why the question of whether a strategy ever becomes more than a document lies so close to this one: governance is the place where a strategy must for the first time become concrete, or never does.
Once this work has been done, something shifts in the FTE capacity of the management team itself. Time that now goes into re-arguing who decides what becomes available for actually adjusting the course. That is not a promise about how many hours that amounts to, that depends on how many decisions in this organization are still implicit and how often they recur. Nor is it a replacement of people: it is sharpening who is responsible for what, something separate from personnel decisions, for which, if they come up, different legal requirements apply than are at issue here.
Governance does not stand apart from the other dimensions. A decision about who approves AI recommendations immediately touches the question of what it costs to make processes and operations ready for AI, because an approval step only works if the process has room for it. And a decision right can only be checked if there is something to measure, which connects it to the question of what it costs to make performance measurement ready for AI. Anyone who looks at these dimensions separately misses that the readiness of one determines the speed of the other.
The underlying question, which work in this company can genuinely be taken over by AI, is answered by the work scan of FTE TO AI per task, not with an estimate at management level. For the question central to this page, there is the free readiness check: eight short questions, one per dimension, which in a few minutes show where your organization is furthest along and where not yet. The full ambition test, with the four layers and the five confidence gates, is under construction.
Vertel wat u wilt bereiken, dan kijken we samen wat daarvoor moet staan.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.