Financial services consist to a large extent of work that can be described before it is carried out: acceptance according to fixed criteria, transaction processing, document checks, reporting in a fixed format, customer questions that recur in patterns. That is not a coincidence. Supervision and regulation have for decades forced this work to be repeatable and traceable. What is repeatable and traceable is also the work of which a part is first eligible for AI. Alongside that stands work that revolves around explanation, exceptions and responsibility: an advisor guiding a client through a difficult decision, an underwriter deviating from the rule, a compliance officer assessing a grey area. That work shifts much more slowly, and sometimes not at all.
The circumstances that steer the outcome are therefore not the same everywhere. They depend on how standardised a task is, how much oversight is legally required, and how well the underlying data are in order. A company with outdated systems and files that are partly still on paper experiences a different shift than a company that has worked digitally for years.
Within this sector, three patterns run alongside each other. For part of the work, AI can take over the task: simple transactions, standard checks, summarising files. For another part, the work is done partly by AI, with an employee approving or rejecting it and substantiating that with a reason — think of acceptance of more complex policies or initial triage of customer questions. For a third part, the work remains human work: everything in which explanation, trust or an individual judgement forms the core of the task.
This is not a division that lies ahead. In some companies the first category is already running in production and the second category has been set up with a clear review process. In other companies the same task is still done entirely by hand, not because it could not be done otherwise, but because no one has explicitly reviewed it. The difference rarely lies in the technology. It lies in whether someone has recently looked at the task again.
A strategy in financial services often rests on assumptions about lead times, staff deployment per process, and the place where competitive advantage comes from — speed of acceptance, quality of advice, cost per policy or per file. Those assumptions were fixed at a certain point and not tested again afterwards. When AI takes over part of a task, the calculation underlying those assumptions changes without the plan itself being adjusted. The assumption is not contradicted; it simply stops being confirmed.
That is the pattern this page points to: not that AI is a project that gets implemented somewhere, but that AI continuously invalidates assumptions that a strategy silently uses as a fixed point. A board that holds on to a cost advantage that was based on human processing speed is steering on a figure that has meanwhile been overtaken by competitors with a different process. Similar shifts are playing out in other sectors, as can be seen in how strategic pressure in construction is changing now that AI is taking over work via contracting and estimating work partly prepared by AI, in the installation sector where planning and material estimation are shifting, and in the cleaning sector where scheduling and inspection are partly automated. The sectors differ, the mechanism does not: an assumption that no one has withdrawn stays in the plan while the reality beneath it shifts away.
This is not a staffing question and not advice on who still performs which task. Which consequences a company attaches to the fact that a task is capable of being taken over is up to that company, and touches on legal requirements that are not addressed here. What this is about is the question that precedes such a decision: which assumption in the strategy is standing still on work that is by now being done differently. That is a factual question, not a staffing question, and the two are not mixed here.
Nor is it a question that can be answered with an estimate per sector. How much of the acceptance work, reporting or customer contact within a specific company currently falls under which of the three categories differs per organisation, per system and per process. A board that reasons at sector level misses precisely the difference that is relevant to its own strategy.
Whether a strategy still holds true is not something you can wait for the annual plan to clarify. The way to track that continuously is described in how you measure whether a strategy works without waiting for the annual cycle, to be found via a method for measuring strategy between the regular evaluation moments, and connecting to the question of how you recognise that an assumption has become outdated, elaborated in signals that show you a strategy has started to shift. The underlying question — which work in your own company can genuinely be taken over by AI — is answered task by task with the work scan of FTE TO AI.
You can start without reopening the entire plan. The free assumption check is a short round in which you identify the key assumptions underlying your strategy and see, per assumption, when it was last confirmed — not whether it was ever true, but whether it still holds now. The full strategic pressure test, with an outside view alongside a self-assessment from the board, is under construction.
Stel uw vraag. Vaak zit de echte vraag een laag dieper — daar mag ik naar vragen.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.