A board wants a number. Seventy percent ready, or forty, or ninety. A number compares, ranks, justifies a decision in a one-hour meeting. The problem is that readiness for AI work does not exist in that way. Readiness is not a property of an organisation the way revenue or headcount is a property. It is the answer to a series of questions that turns out differently per dimension, and that outcome differs again per ambition.
The reason lies in what actually happens when AI takes over work. That does not happen at once and not equally everywhere. Some tasks AI can carry out independently. Other tasks AI can prepare, after which a human approves or rejects with reason. Yet other tasks remain human work, not because AI cannot do it, but because the organisation has not set up the conditions for it. These three categories run straight through every department, every process and every ambition. A number that summarises this diversity into a single percentage hides exactly the information a board needs: where does the difference lie, and why.
Suppose a readiness score of seventy percent is presented. What does that mean? Seventy percent of processes are ready? Seventy percent of employees are trained? Seventy percent of the data is usable? A score without an underlying question leaves the reader guessing, and the guessing usually happens in favour of the ambition, not in favour of the truth. A board that wants an ambition happily reads a high number as confirmation. A number invites that reading, even when the underlying substantiation does not support it.
The eight dimensions that together determine readiness — from data and systems to decision rights and legal frameworks — do not relate equally to every ambition. An ambition that calls for faster customer contact leans heavily on different dimensions than an ambition that calls for automated reporting. A single readiness number averages away those differences. The answer to the question "are we ready" is therefore almost always: it depends on which part of the work you mean, and which layer of the ambition you are testing.
In one company, the question of whether AI can take over a task has long been answered: oversight has been set up, decision rights have been assigned, someone knows who approves an outcome and on what grounds. In another company, the same task still rests entirely with a human, not because the task requires it, but because no one has recorded who is allowed to reject it if AI makes a mistake. The difference rarely lies in the technology. It lies in whether the organisation has determined in advance what needs to be in order before work shifts: which data is complete enough, which role bears the oversight, which mandate is needed to let an outcome stand.
Those questions are exactly why an ambition test consists of four layers — vision, vision state, target, target state — and why each layer is held separately against the eight dimensions. A vision that states that AI takes over the largest part of reporting asks something different of the organisation than a target that wants this operational within a year. The questions per layer and per dimension do not produce a score, but a map: where is the organisation already far along, and where not yet.
Even within that map, not every answer is equally certain. A confidence gate marks where the substantiation is solid and where it rests on assumptions. A dimension can look "ready" because no one has critically asked the question, not because the answer holds up under testing. Five of these gates run through the test, precisely to prevent a false certainty from being taken for a real one. Where a gate scores low, that does not mean the ambition is unachievable. It means that the outcome on that point says nothing until it has been examined more closely.
That is uncomfortable in a boardroom that wants an answer, but it is the most honest description of what testing can deliver at this moment. An outcome that gives five stars across the board is more suspect than an outcome that says on three dimensions: unknown, further investigation needed.
The eight dimensions and five gates are not an end in themselves. They translate back into capabilities: what must a role be able to do, who decides, who approves or rejects. That is also where many ambitions in practice run aground — not on the technology, but on the question of how you prevent an AI project from getting stuck at the IT department instead of with the roles that actually carry out the work. Even an ambition that is technically achievable can stall on a question that has nothing to do with technology, such as what you do when the ambition is achievable but compliance does not allow it. And not every ambition belongs in the first year: sometimes the answer to the readiness questions is that it is better to look at which ambition you are better off postponing a year than at what needs to happen this quarter.
The underlying question — which work in this company can genuinely be taken over by AI, task by task — is answered with the work scan of FTE TO AI, separately from the ambition test itself.
Giving a number on readiness is therefore not the point. The point is knowing the questions and knowing where the answer still rests on an assumption. Anyone who wants to see where their own organisation is furthest along and least far along can take the free readiness check: eight short questions, one per dimension, giving a picture of where the ambition already has support and where it does not. 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.