A construction company consists of several types of work that do not all move in the same direction. Estimating and work preparation require calculation and working through specifications and standards. Project control runs on planning, progress monitoring and tracking additional and reduced work. On the construction site itself, it is about execution under changing circumstances: weather, ground conditions, deliveries, other parties on site. In addition, there is a layer of administration and accountability tied to permits, quality assurance and completion dossiers. Each of these layers has a different ratio between what is fixed in rules and drawings, and what depends on the situation on site.
That ratio determines where AI is already taking over work today, where oversight remains necessary and where it stays human work. Work that relies on fixed standards, repeatable calculations and structured documents lends itself more easily to being taken over than work that requires decisions on site about a situation that does not fit a template.
In estimating and tendering processes, a large part of calculating quantities and comparing them with previous projects can be done by AI, with an estimator assessing the outcome before it goes out the door. In permit dossiers and quality assurance, it is visible how AI can compile documents and check them against checklists, while a project manager approves or rejects them with reason. On the construction site itself, in execution and resolving unforeseen situations, it remains predominantly human work: the combination of ground conditions, weather, other contractors and the physical location does not lend itself to being summarized into a task that a system takes over.
This shift does not proceed at the same pace across companies. A construction company with standardized specifications, a fixed calculation model and digitized dossiers sees more quickly where AI frees up estimating or administrative hours than a company where those same steps still sit separately in people's heads and loose files. The difference therefore does not lie primarily in the sector, but in how the work within that company was already organized before AI came into view.
A management team that has built a multi-year plan on a certain staffing requirement, a certain lead time per project or a certain margin on estimating, fixed those assumptions at a particular moment. AI that takes over work does not change the plan itself; it changes the ground on which part of that plan stood. An assumption about how many estimating hours a quote requires, or about how much FTE is needed to deliver a dossier, may still hold true today and no longer next quarter, without anyone on the management team having explicitly established that.
That is the point at which strategy quietly ages: not because the plan was wrong, but because one of the assumptions underlying the plan has lapsed without anyone withdrawing it. With AI in construction, the question is then not whether automation can handle something, but which assumption in the company's own plan no longer holds as a result.
Within a single management team, the picture of this shift can diverge: someone close to estimating already sees AI taking over tasks; someone responsible for the construction site mainly sees work that cannot be transferred. Both pictures can be correct for the part of the company that person oversees, and still produce a distorted overall picture as soon as they are not combined. How such pictures within a management team can diverge and what that means for decision-making is described in an explanation of why the pictures within a management team of the same company can diverge.
Construction shares this pattern with other sectors where physical work and document work occur together. How the same shift plays out in the installation sector, where breakdown service and administration exist side by side, is discussed in the explanation of strategic pressure in the installation sector. For companies that work more with document flows than with a construction site, such as wholesale, the ratio between takeover and oversight is different again; that difference is worked out in the piece on strategic pressure in wholesale.
This shift is not a statement about what a construction company should do with its staff. Whether freed-up hours lead to different deployment, other projects or something else, is a choice that lies with the employer and is bound by its own legal requirements as soon as it concerns personnel decisions. What is described here is where the nature of the work itself changes: which part is transferable, which part requires oversight with assessment and which part remains human work. Figures on exactly how much that is differ per company and depend on how standardized the work already was before AI came into view; a fixed percentage for the entire sector does not exist.
The question of which work in this company, with these projects and these dossiers, can genuinely be taken over by AI, cannot be answered from a distance. The work scan from FTE TO AI answers that question per task, based on how the work is organized now. Anyone who first wants to see which assumptions in their own strategic plan are affected by this shift can gain clarity with outside-in thinking as a counterweight to an inward-looking plan, which describes how external perspective and internal perspective are placed side by side.
Anyone who wants to start can do so with the free assumption check: a short round in which you identify the key assumptions underlying your own strategy and see, for each assumption, when it was last confirmed. The full pressure test, with sector data, a self-plot for the management team and a running assumption list, 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.