A management team talks about "next year this process will largely run on AI" and everyone nods. What happens next differs per head around the table. One director thinks of a task that AI takes over entirely. Another thinks of a task that AI prepares, with a human who approves or rejects it. The third actually thinks of something that remains human work, only supported more quickly. That difference is rarely named before the budget, the planning, or the workforce planning has been fixed.
That is the core of an ambition the organization can't handle: not that the ambition is pitched too high, but that no one has established what the ambition actually presupposes. An ambition that presupposes AI work requires different capabilities than that same ambition without AI. Without that translation, the ambition is a sentence, not a plan.
AI already takes over parts of tasks today, more so in some companies than in others. That difference does not lie primarily in the technology, which is often just as accessible to competitors. It lies in whether an organization can sharply distinguish three categories: work that AI can take over independently, work that AI partly does with human oversight that approves or rejects with reason, and work that remains human work. Companies that have made this classification per task can test an ambition before voicing it. Companies that have not done so only discover the classification while trying to execute the ambition, and by then the delay has already been incurred.
An ambition that fits on half a page usually conceals four different things. The vision: where does the organization want to stand in the long term. The vision state: what part of that is already reality and what is still aspiration. The target: the concrete goal for the coming period. The target state: what must be in place before that goal is achievable. Without naming these four layers separately, a management team keeps talking about the vision while thinking about the target, or vice versa. Our vision is too vague, how do I make it concrete addresses exactly that first layer problem: a vision that inspires but is not testable.
The ambition test places the four layers alongside eight dimensions of organizational readiness: among others, data quality, process maturity, decision rights, and the capacity to organize oversight over work that AI partly takes over. Each dimension is tested along five confidence gates, from "we have no visibility into this yet" to "this has been demonstrated in practice, not just on paper". An ambition is not unsuitable because a dimension scores low. It is unsuitable if the target presupposes that a dimension already scores higher than the gates demonstrate.
The order in which dimensions are addressed is itself a choice, and not always an obvious one: sometimes decision rights must be settled first before oversight of AI work makes sense, sometimes it's the other way around. Which order do you follow when three dimensions lag behind at once addresses precisely that bottleneck.
The ambition test does not produce a score that approves or rejects the ambition. It provides a series of questions per dimension, and the answer to those questions is an estimate, not a measurement. For two dimensions the estimate is fairly solid, because the organization already has experience with comparable work. For other dimensions, especially those involving oversight of AI work not previously tested, the estimate is based on a brief round of questions, and the outcome then mainly says: more investigation is needed here, not: here is the answer. An outcome resting on a single guess from a single manager says little. The test only becomes meaningful when multiple people with different positions in the organization work through it and compare answers. That comparison, and valuing what an improvement actually delivers, is a separate question: how do you measure whether an improvement has truly landed addresses that. The reason readiness is not a single score is explained in why readiness is not a score but a series of questions.
If the eight dimensions and five gates show something about where the organization stands, the next question is who must do the work to close the gap. The ambition test translates the outcome into capabilities, and those in turn into roles and decision rights: who decides when AI work is approved, who is responsible if oversight fails, who records what has changed. This is not a workforce question in the sense of who stays and who goes; that is governed by its own legal requirements, which are not addressed here. It is about who may make which decision. Who is needed to make work by AI succeed and what do you record about decision rights in a change elaborate further on that translation.
The ambition test says something about the organization: is it ready to carry the ambition. It says nothing about which tasks in this specific company can actually be taken over by AI. That question, answered per task, is the domain of the work scan of FTE TO AI.
The full ambition test is under construction. Pending that, there is a free readiness check: eight short questions, one per dimension, resulting in a picture of where the organization is furthest along and where it lags most. No score, no guarantee, but a first view of what the conversation in the management team should actually be about before the next ambition is voiced.
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.