A management team states the ambition: to become less dependent on key employees. Everyone nods. And then every manager sets to work with a different interpretation. One manager has processes documented. Another starts a training programme. A third calmly waits, because the key employee in his team was only hired two years ago and is unlikely to leave soon. All three are busy, none of the three are measuring the same thing.
The statement itself does not explain that difference. "Less dependent" is not a task and not a goal, it is a direction. To be able to do anything with it, that direction must first be broken down into what exactly is currently locked into a key employee, and then into what the organisation needs to be able to do to release that. That is a different conversation than planning a knowledge transfer. It is a conversation about capacity: which work can continue without that one person, and who or what takes it over.
If you unpick why an organisation is dependent on a few people, that usually falls into three types of work. First, routine execution that is not documented anywhere, only in the head of someone who has been doing it for years. Second, judgement work: someone who, based on experience, assesses whether a file is correct, a quote works, a risk is acceptable. Third, relational work, where the key employee is themselves the connection with a customer, supplier, or team.
These three types of work differ fundamentally in what can be done with them today. The first category -- documented or documentable routine -- is often the work that a system or AI application can take over, provided it has once been described. The second category, judgement work, can usually be partly taken over: an AI application can make a proposal or flag a deviation, a human approves or rejects it and gives the reason. The third category, the relational part, generally remains human work, although the preparation for it can be eased.
The ambition "less dependent on key employees" therefore does not mean one intervention, but three different movements, each with its own pace and its own risk.
The organisations where this shift is already visible have generally first mapped the work of their key employees before starting on transfer. They knew which part was routine, which part judgement, which part relationship. This allowed them to make targeted choices: automate this part, have this part done with oversight by a broader team, deliberately leave this part with the person.
At organisations where the dependency persists, what is often missing is not the will but the overview. There is no up-to-date picture of who is indispensable where, let alone why. Documentation is outdated or does not exist. Decision rights have never been made explicit, so no one knows whether a replacement is actually allowed to decide what the key employee currently decides. The difference, then, does not lie in ambition, it lies in whether the organisation knows itself at the level of tasks and decisions.
The ambition assessment from FTE TO AI first records this ambition in four layers: the vision itself, the vision state in which the dependency has been reduced, the target that lies in between, and the target state that concretely describes which work sits where. It then tests against eight readiness dimensions -- among other things whether processes are documented, whether decision rights are assigned, whether there are systems that can support judgement work, and whether managers are willing to organise oversight differently.
Five confidence gates determine whether the organisation can proceed to the next step, or must first put something in place. An organisation that has no process documentation does not get through the gate that requires it, however much the management team wants to push ahead. That is not a delay for form's sake, it is the point where it becomes visible what still needs to happen before work can actually shift.
The outcome is translated back into roles: who gets which decision rights when the key employee steps back, who assesses the exceptions that a system cannot handle, who is responsible for maintaining the knowledge that was previously only in one head. This affects how roles are structured; what consequences that has for individual employees falls under the employer's own applicable legal requirements, which is for the employer to assess.
A year from now this ambition will not have been achieved because someone says so, but because there are figures that show it: how much fte capacity that was previously with a single person is now spread more broadly or supported by a system; how many hours that used to stall when that person was absent now simply continue. That is a different kind of evidence than a satisfaction score or a statement of intent. It is measurable because it traces back to the four layers and eight dimensions with which the ambition was recorded.
This ambition, incidentally, does not stand apart from others. Anyone who also wants to grow without hiring extra people runs into the same question of which work sits where. And because every ambition assumes a different pace of change, it is also useful to look at how quickly an organisation can actually change, regardless of how fast the management team would like it to go.
The underlying question -- which work in this company can genuinely be taken over by AI -- is answered per task with the FTE TO AI work scan. For the ambition itself, it is wise to first get clear on what exactly an ambition is and what a goal is, since the two are often confused; that distinction is explained on the page about the difference between a goal and an ambition.
Anyone who wants to know where their own organisation is already ready and where it is not can take the free readiness check: eight short questions, one per dimension, resulting in a picture of where you are furthest along and where you are least far along. The full ambition assessment, with the four layers and the translation into roles and decision rights, is under development.
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.