A construction company consists of work that is difficult to summarize in a single process. Estimating, work preparation, planning, procurement, execution on the construction site, supervision, revision and handover run into each other and depend on a drawing that can still change, a supplier that delivers late, or weather conditions that upend the schedule. Many hours go into aligning these parts with each other: an estimate that matches a specification, a schedule that matches material delivery, an execution that matches what is actually in the ground or in the structure. That makes construction a sector in which ambitions are quickly put on paper but land more slowly in daily work, because the work itself is fragmented across roles, locations and moments.
AI is already taking over tasks in parts of this work, not as a future vision but as something that happens today in some companies and not yet in others. An estimate that largely comes about automatically based on previous projects, a schedule that recalculates itself when a delivery time changes, an environmental permit that is pre-checked for completeness: these are tasks where software can do the largest part of the work. Other tasks, such as assessing a deviation on the construction site or estimating a risk with a difficult subsoil, remain partly human work with AI providing an initial assessment and a project leader approving or rejecting it with reason. Other work still, such as negotiating with a subcontractor or steering a team on the construction site, remains human work without AI intervening there in the short term.
The difference between companies that already organize this way and companies that do not yet do so rarely lies in the availability of the technology. It lies in the question of whether the underlying data is in order: an estimating system that learns from previous projects needs consistent project data, and that is not automatically present in construction, because each project is recorded differently. It also lies in the question of whether roles and decision rights are set up for the new work: who assesses an AI proposal for a schedule, and on what criterion may that person deviate. Without that answer, an ambition remains an idea.
A management team that talks about "AI in estimating" does not mean the same thing everywhere. One ambition assumes that AI produces a draft estimate that an estimator still checks; the other assumes that AI independently delivers estimates that are only reviewed in case of deviation. Those are two different levels of readiness, with different requirements for data quality, for the estimator's trust in the system, and for how responsibility is divided. Those who do not make that difference explicit only discover it once the project is already underway and the estimate turns out to be wrong.
This is where the ambition assessment comes in: ambitions are recorded in four layers, from the broad vision to the concrete target, and tested for readiness across eight dimensions, with five confidence gates that indicate where the ambition does and does not stand on solid ground. That assessment is translated back into capabilities: which roles need to be able to do something new, and who gets which decision right in the changed process. This is not personnel advice and not a basis for a decision about jobs; separate legal requirements apply to that. It concerns the question of what an organization must be able to do before work can actually shift, and where it currently turns out that it cannot yet.
Construction shares this pattern with other sectors where work is fragmented across locations and roles: the installation sector faces comparable issues around planning and fault diagnosis, wholesale struggles with comparable data requirements for inventory and procurement, and manufacturing runs into similar limits in quality control and process planning. The question is always not whether AI can take something over, but whether the organization around it is set up in such a way that the takeover also holds up. For a management team, that is often first a conversation about language: how do you get a management team on the same page when everyone interprets "AI in planning" differently, and how do you prevent a strategy from remaining a document when the ambition has indeed been established but nowhere translated into what an estimator, planner or executor does differently tomorrow.
The underlying question — which work in this company can genuinely be taken over by AI — is answered per task by the work scan from FTE TO AI, separate from the broader ambition assessment.
A management team that wants to know where its own ambitions on AI in construction stand does not need to start with a fully worked-out plan. The free readiness check consists of eight short questions, one per dimension, and produces a picture of where the organization is furthest along and where the least. The full ambition assessment, with the four layers and the five confidence gates, is under construction.
Vertel wat u wilt bereiken, dan kijken we samen wat daarvoor moet staan.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.