In healthcare, a large part of the time does not go to the care activity itself, but to what surrounds it: schedules that match staffing levels and collective labour agreement rules, record-keeping that is complete and timely, reporting to referrers and insurers, indication assessments and the administrative accountability that comes with it. There is also intake and triage, where speed and diligence must be delivered at the same time. The outcome for a client or patient is driven by three things: how quickly someone is helped, how consistent the quality is across locations and staff, and how well the care fits the specific situation. On these three points, this sector has traditionally distinguished itself, and on these three points the playing field is now changing.
Waiting time for an intake was for a long time a function of staffing levels: more capacity, shorter waiting list. Where triage and file preparation are partly done by AI, with a practitioner assessing and deciding, the waiting time becomes decoupled from staffing on the floor. This does not affect every organisation equally. An institution with standardised intake forms and structured referral data can hand that work to AI; an institution where intake relies on verbal handover and loose notes does not have that work in a form suited to it. The difference, then, does not lie in the willingness to automate, but in how the work was already organised before AI came into view.
The same pattern applies to record-keeping and accountability to insurers. Summarising a file, filling in standardised fields and flagging missing information is work that AI can take over, with an employee approving or returning the summary. Schedule planning with fixed rules around qualifications, rest periods and availability is shifting as well: drawing it up can be taken over, while exceptions and human discussion about shift-swap requests remain human work. The substantive treatment relationship itself, the conversation in which a client must feel seen, remains human work in virtually all cases, and that also remains the part where healthcare organisations distinguish themselves the least from AI and the most from each other.
The speed at which this work shifts depends on how structured the underlying data is, on the type of care, and on what regulators and professional bodies allow in terms of automated decision-making. An organisation that has its systems in order sees freed-up hours return as extra FTE capacity for direct care or for reducing waiting lists. An organisation where files are scattered across systems that do not communicate with each other does not see that gain, even if it wants to. This is a question about the organisation of work and data, not about workforce policy; if a shift in tasks has consequences for staffing levels, separate statutory requirements apply that are not addressed here.
If waiting time and file quality are increasingly less a function of staff size and increasingly more a function of how well an organisation has arranged its tasks for takeover by AI, then what clients, referrers and health insurers judge a provider on shifts as well. Delivery time of an intake becomes less a capacity question and more an organisation question. Consistency of reporting becomes less a matter of individual diligence and more a matter of system choices. Whoever recognises this early shifts the comparison along with it; whoever holds on to the old assumption that more staff automatically means a shorter waiting time is comparing themselves on a dimension that is disappearing.
This shift is not limited to healthcare. In wholesale, the comparison shifts as AI takes over inventory and order work, in professional services, the comparison changes as AI takes over reporting and advisory preparation, and whoever wonders why competitors all sound the same in their promises sees the same underlying cause: claims that are no longer substantiated on the dimension on which they once won.
The question of which work in a specific healthcare organisation can genuinely be taken over by AI cannot be answered in general terms; the work scan by FTE TO AI answers that question per task, with an assessment along the lines of takeover, human oversight, or human work.
To know what your organisation currently wins on, it is necessary to know whether that claim still matches how the work is organised, or how it used to be organised. For competitors without published figures, a dedicated approach exists to still arrive at a substantiated picture, as described in how to obtain information about competitors who do not publish figures. A first step is the free dimension check: you name what you believe you win on, and you see which of those claims can be defended with evidence. The full benchmark, with the peer group of your market and the evidence matrix per dimension, is under construction.