You arrive at a job with your tools, your training and twenty years of knowing what happens when tile goes over an uneven screed. The client meets you at the door with a phone in their hand and a printout from ChatGPT that says the job should take two days and cost 40% less than your quote. If you work in the Hungarian building trades, this scene is now routine.

According to a Mapei Kft. survey reported by Portfolio.hu, 73% of Hungarian construction professionals have encountered clients who challenged their work based on AI-generated answers. One in five run into it frequently. Another 26% see it occasionally, and 26% very rarely. The pattern is now common enough that it has quietly become part of the job, alongside the drill and the spirit level.

What clients are actually arguing about

The Mapei research shows the disputes cluster around a predictable list. Clients most often question the technological solution you propose, the materials you specify and the work process you follow. They also arrive with chatbot-derived opinions on pricing, on what caused a defect, and on how long the job should take.

Béla Markovich, managing director of Mapei Kft., framed the underlying problem plainly. In his words, the root of the issue is that the chatbot cannot see the actual state of the walls, the layers beneath the surface, or the hidden faults that only appear during demolition; it works only with idealised cases, which in construction are the rarest kind. That mismatch, he added, produces unrealistic price expectations, untenable deadlines and technological demands that are simply not workable on that particular site.

One unnamed tradesperson in the survey put it more bluntly: the AI appears to solve everything, but takes no responsibility for the consequences or for the mistakes that follow.

The hidden cost: your unpaid time

The billable hours you invoice cover the work on site. What they do not cover, in most cases, is the growing amount of time you spend explaining why the chatbot is wrong. That conversation now happens at the initial site visit, in the middle of the job when the client checks their phone, and again at the final walk-through.

Interestingly, most professionals in the survey are not particularly upset about it. More than half said clients using AI for information does not bother them at all, or bothers them only slightly. Only 23% described it as distinctly annoying. The friction is real, but for most trades it has become a manageable part of client management rather than a crisis. The question is whether that unpaid explaining time is baked into your quotes.

How to handle the ChatGPT client, in practice

If more than a small share of your jobs now involve debating an AI printout, a few concrete habits help protect both the work and your margin:

• Price the conversation, not just the work. Build a realistic client-communication buffer into your quote. If two client meetings used to take an hour and now take two, that hour belongs in the estimate.
• Ask to see the prompt. When a client says ChatGPT told me it should cost X, ask what they typed in. Nine times out of ten the prompt described an idealised scenario, exactly the gap Markovich flagged, and pointing that out defuses the argument faster than a technical lecture.
• Put the site-specific findings in writing. A short written note, sent by message, that names the hidden faults you found during demolition, the substrate condition and the reason the timeline moved, is much harder to argue with than a verbal update.
• Photograph the surprises. A photo of the damp wall, the rotten batten or the previous botched job under the tile does more to justify a revised price than any explanation. It also protects you if the dispute escalates.
• Decide, in advance, which battles you take. Some clients want a partner; others want a technician who agrees with their chatbot. It is easier to walk away from the second kind before you have started than after you have opened up a wall.

The other side: tradespeople are using AI too

The same Mapei survey found something worth noticing: more than a third of construction professionals are using AI themselves, at least occasionally. 13% use it weekly, 7% daily, and a further 27% have tried it at least once. Roughly one in five use it for cost calculations, materials estimation, quote preparation and troubleshooting.

In other words, the tool that is generating the awkward client conversations is also, quietly, becoming part of the trade's back office. A quote drafted with a chatbot, then corrected by a professional who can actually see the substrate, is a very different document from a client's printout. The difference is your judgement, and that is what you are being paid for.

What this means when you are pricing your next job

If you are self-employed, the practical question is whether your day rate still reflects the real shape of the job. A tiling job in 2026 is not just the tiles: it is the site inspection, the materials, the labour, and now, for roughly three in four Hungarian tradespeople, a client-management layer that did not exist five years ago.

If you are employed by a company, the same logic applies to how you talk to your employer about workload. If you are spending measurable time each week defending quotes against chatbot answers, that is a real change in the job description, and worth raising in your next review.

The chatbot is not going away. But the trades that thrive over the next few years will be the ones that stop treating AI-armed clients as an annoyance and start treating them as a category of work, priced, managed, and documented like any other.