The cleaning industry consists largely of operational work on location: reaching spaces, assessing contamination, choosing materials and methods, and checking the result against a contractual standard. Alongside this sits a layer of planning and administration: building rosters around sickness absence and object changes, hours registration, recording quality checks, quotations and contract management. These two layers respond differently to AI. The physical work on location is location-bound and dependent on the circumstances on site; planning and administration run on data that can be structured, compared and processed automatically.
The circumstances that steer the outcome are not the same everywhere. An object with a fixed daily routine lends itself to support differently than an object with varying contamination and incidental assignments. An organisation with many small contracts has a different planning load than an organisation with a few large multi-year contracts. That difference in starting position is exactly why the same technology already works in one company and not yet in another.
Within the industry, three types of work run alongside each other. Roster planning based on historical occupancy, absence patterns and object characteristics is work that a system can largely take over; a planner then assesses the outcome instead of putting it together from scratch. Quality control and complaint handling more often sit in the middle category: a system flags deviations based on photos, sensors or recurring complaints, but a supervisor assesses whether the deviation is justified and what action follows. The cleaning itself, assessing a space that is not contaminated according to pattern, and contact with a client about custom work remain largely human work, because the variation on site is too great to lay down in advance.
This division is not a snapshot that stays still. As more objects get standardised sensor data or fixed rounds, part of the middle category shifts towards full takeover. This happens per object and per contract, not all at once for the entire organisation.
The pace at which AI takes over work does not depend on the will to innovate, but on what is already in place. A company with standardised object profiles and digital hours registration can have planning supported faster than a company still working on paper or in separate Excel files. A company with a lot of its own staff at fixed locations needs a different assumption about oversight than a company that mainly works with subcontractors. Contract form, object mix and the degree of digitalisation together determine where the boundary between the three categories currently lies, and that boundary is slightly different for each company.
A strategy in this industry often rests on assumptions such as: the planning department needs a certain size to keep occupancy balanced, or: quality control requires a fixed number of rounds per week per supervisor. As long as AI is only mentioned in other companies, those assumptions remain standing on paper. As soon as roster support or deviation detection is actually used in one's own company, the assumption is no longer true, even though it still stands in the strategy document. No one has decided to revise it; it has simply been overtaken by what is already happening on the floor. This is the pattern that also plays out in similar form in other sectors, as can be seen in the descriptions of the strategic shift in construction through AI and how AI is changing strategic pressure in healthcare.
The difference between a plan and a strategy plays a direct role here: a plan describes what needs to happen, a strategy rests on assumptions about how the world works. Anyone wanting to get that distinction sharp will find an explanation in the difference between a strategy and a plan. And to recognise when an assumption is no longer confirmed, the explanation of what a strategic assumption is and how you check whether it still holds helps.
The question of which work in a specific cleaning company can actually be taken over by AI cannot be answered in general terms; the work scan from FTE TO AI answers that question per task, with a judgement about the category the task currently falls into.
This page describes a shift in work, not a personnel decision. Whether and how an organisation does something with freed-up hours or changing capacity needs is up to the employer, and if that choice touches on dismissal or reorganisation, its own legal requirements and its own advice apply. What happens here is more limited and concrete: checking which assumption underlying the strategy still holds and which has quietly become invalid.
The first step is not a fully revised strategy document, but a short inventory of the assumptions that already underlie the plan. The free assumption check is intended for this: a short round in which you name your key assumptions and see for each assumption when it was last confirmed. The full pressure test, with the external sector view alongside the management's own plot, 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.