Construction is a sector with two faces. On one side is the work on the building site: physical, weather-dependent, full of custom work and on-the-spot improvisation. On the other side is the office or the site cabin: calculation, planning, procurement, permits, hours accounting and communication with clients, subcontractors and inspectors. These two worlds run into each other, and that makes the question "what can AI take over here" less simple than in office-bound sectors.
On the building site itself, physical work is dominant: bricklaying, installing, pouring, connecting. AI does not change much directly about the action itself for the time being. What does change is everything surrounding that work: recording progress, checking drawings, flagging deviations, keeping track of materials and hours.
In the office environment of a construction company, the largest part of the hours goes into recurring administrative and coordinating tasks: drawing up quotes and calculations, updating planning, processing invoices and purchase orders, communication with suppliers and subcontractors, and keeping track of permits and inspection files. In addition, there is a layer of work centred on regulation: building codes, safety regulations, collective labour agreements (CAO) and certification requirements. This regulation changes regularly and differs per project, which makes it harder to fully automate tasks without oversight.
A third characteristic is the multitude of separate systems: one company works with a planning package, another with Excel, a third with an ERP system that does not talk to the accounting software. This fragmentation strongly determines how much AI can take over in practice, because a task that looks easy to automate on paper often gets stuck in practice on a system that has no connection.
Some tasks within a construction company can be clearly delineated and have a clear, repeatable structure. Think of transferring receipts and invoices into the accounting system, summarizing project statuses from progress reports, or making old drawings and files searchable. Companies that still work with paper construction files will recognize this type of work in the question of whether AI can digitize and unlock paper archives: scanning, recognizing and indexing drawings and permits is a task with a fixed pattern, where AI can do the largest part of the work.
Keeping basic personnel administration is also suited to this. The question of whether AI can keep track of leave balances applies in construction just as much as elsewhere: leave days, ADV hours and sick leave follow rules that can be well captured in a system, as long as the underlying data is correct.
A large part of construction work falls into the middle category: AI can provide a proposal, concept or signal, but a person assesses and decides. Think of calculations: AI can draw up a cost estimate based on previous projects, but a calculator assesses whether the assumptions are correct for the specific situation, the ground conditions, the planning and the current market prices of materials.
The same applies to hours registration at project level. The question of whether AI can register hours is relevant for construction companies that need to allocate hours to projects for invoicing and post-calculation. AI can recognize hours from time registrations or photos of work tickets and link them to a project, but a project manager or administrative employee must check the allocation, especially in the case of additional work or unclear descriptions.
Also in assessing photos and videos from the building site for quality control or safety, AI can signal deviations, for example a missing fence or a crookedly placed element. The assessment of whether this is actually a problem and what action is needed remains with a person, if only because the consequences of a wrong assessment on a building site can pose direct safety risks.
The physical executive work on the building site remains human work: bricklaying, carpentry, installing, welding. Negotiating with clients and subcontractors, resolving conflicts on the building site and making decisions in unforeseen circumstances also require experience, adaptability and responsibility that cannot be captured in a task.
In addition, the final responsibility for safety and quality remains with people, even when AI signals or supports. Legal liability for construction defects or accidents does not shift to a system.
This breakdown says something about tasks, not about persons or job functions. Whether and how a company adjusts its staffing based on this information is a decision subject to its own legal requirements, for example regarding dismissal, employee participation and collective labour agreement provisions. This page provides no basis for that.
The ratio between what AI can take over, what requires oversight and what remains human work differs strongly per company: a contractor with a lot of custom work and little repetition will be able to automate fewer tasks directly than a company with many standard projects and well-structured systems. Similar differences can be seen in other sectors, as shown by what AI can take over in the installation industry at /wat-kan-ai-overnemen-in-de-installatiebranche and what AI can take over in healthcare at /wat-kan-ai-overnemen-in-de-zorg, where physical presence and regulation also play a major role.
If you want to know how this applies to your own company, a good starting point is the free quickscan from ftetoai: twelve questions, no account required, giving an indication of what part of the hours in your profile can be taken over by AI today. The full work scan, which goes deeper into individual tasks, is still under construction.
Vraag maar. Ik ken de kennisbank van deze site; wat ik niet weet, zeg ik erbij.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.