A buyer sooner or later asks the same question: if part of the work is done by AI, what remains that depends on people, and what remains that depends on you. The answer you give determines whether a saving is read as something structural or as something temporary that disappears the moment the owner leaves.
In some companies, AI already takes over tasks today that were on a schedule last year: first drafts, classification, repetitive checking work. In other companies, almost nothing changes yet. The difference usually does not lie in the sector, but in how the work is broken down. Where tasks are already described, repeatable, and organized independently of the person, software can take over part of it. Where knowledge sits in people's heads and decisions run undocumented, little changes, regardless of which model is purchased.
Three categories run through every process. A task is fully taken over, a task is partially taken over with a person approving or rejecting it and who can act as a motivator, or a task remains human work because judgment, context, or responsibility cannot be transferred. Most companies have a bit of all three, often without anyone ever having written that down.
A buyer does not buy the number of people on the payroll, but what keeps running when part of those people are no longer there. Work that AI demonstrably takes over therefore changes the weight of virtually every value driver. Owner dependency weighs the heaviest, because if knowledge and decisions rest with you, any gain from AI adoption becomes uncertain again the moment you leave. But earnings quality also changes in nature: a saving that arises because you happened to find a cheap supplier counts differently than a saving that is anchored in the process itself and therefore travels with the company. How you substantiate a margin improved by AI determines whether that margin stays in the valuation or gets stripped out.
A work scan that assesses per task whether AI takes it over, partially takes it over, or does not take it over produces evidence based on how the work is currently organized: how repeatable a task is, how much context is needed, how much oversight an outcome requires. That is a snapshot of today, not a prediction of what a vendor's model can do in two years. Models improve, regulations shift, and a task that is still human work now may not be in a year. The reverse also holds: a task that is now partially taken over can, in the event of a staff turnover, become fully human work again if the knowledge was never recorded.
The outcome also means nothing without the context of the process the task sits within. Ten hours that AI takes over in a process with one customer and one contract is something different from ten hours in a process spanning a hundred customers — which is why a buyer also looks at what customer concentration means for weighing a saving. And a saving that rests entirely on a single software license is more fragile than a saving built into multiple systems, which partly explains why a due diligence examines your software licenses. A scan that does not account for that produces a number without meaning.
The method says nothing about personnel. Whether a task can or cannot be transferred to AI is a fact about the work, not advice about who does the work. If an outcome touches on personnel decisions, its own legal requirements apply, separate from what a value scan shows.
A buyer who probes carefully sees through a story that is only optimistic. Credible is a story that keeps three things separate: which tasks have already been taken over today with evidence from the process, which tasks have been partially taken over under oversight and therefore still cost human hours, and which tasks deliberately remain human work and why. Whoever cannot make that distinction is often offered a lower price or an earn-out proposal to leave the gap between promise and evidence with the seller.
The underlying question — which work in this company can genuinely be taken over by AI, per task, with evidence — is precisely what the FTE TO AI work scan answers. And because AI adoption is not an endpoint but a process that continues as the market changes, it is also worth knowing how you maintain your value while the market shifts, and which agreements help with that, because certain contracts increase a company's value more than others, especially when part of the work shifts from people to systems.
A good answer to the buyer's question starts with knowing where you stand yourself. The free value check consists of eight short questions, one per value driver, and gives a picture of which driver weighs most heavily on your price today. The full value scan, with maturity scores per driver, the owner dependency index, and a two-year calendar toward the exit moment, is under construction.