Revolutionizing Construction: AI Insights from Capmo

AI in Construction: Lessons from Capmo's Florian Biller

Definition: What Is AI in Construction?

AI in construction means software that does the work for you — not software you have to fill and manage. In the context of construction project management software, it should proactively handle routine tasks so teams can focus on site work and decisions.

That distinction comes from Florian Biller, co-founder and CEO of Capmo, one of Europe's leading AI-based construction project management software platforms. After seven years building technology for the construction industry, Florian has a clear and practical view of what AI in construction actually means, why it is different from AI in other industries, how it aligns with broader construction technology trends, and what it will look like in five years.

This article draws entirely on his Foundamental University masterclass.

Watch the full masterclass here: university.foundamental.com/masterclass/florian-biller

If you follow construction technology trends, this session offers practical examples of ai in construction at work.

Why Construction Needs Its Own AI

Construction is the second least digital industry in Europe. Only hunting and fishing ranks lower, according to McKinsey. That sounds like a problem. For AI, it is actually an opportunity, particularly for construction project management software that can surface project knowledge fast.

"Construction is the industry that, after finance, produces the most data in the world," Florian explains. "And up until now, almost none of it was being used."

Every construction project generates enormous volumes of data: contracts, specifications, plans, photos, videos, change orders, daily site reports, emails. If you printed out all the documents for an average project and stacked them, Florian says, the pile would be ten times the height of the building being built.

The problem is not a shortage of data. The problem is that construction teams could not access or use it. Project managers spent roughly 20 to 30 percent of their working day searching for information they knew they had somewhere but could not find quickly enough.

AI changes this. Not by replacing construction professionals, but by making the data they already have actually usable.

How AI Is Used in Construction Today

Florian is direct about how Capmo approaches AI with customers: they do not talk about it.

"We don't label it AI to our customers, because in construction nobody cares about the word AI or artificial intelligence. They care about results."

In practice, Capmo's AI works across several areas of construction project management. In modern construction project management software, these capabilities are embedded so teams get answers without extra effort.

AI Search is the most visible example. Instead of navigating through folders, documents, and emails to find a specific piece of information, a construction manager can simply type a question. The system searches across everything — plans, photos, voice recordings, documents, emails — and returns an answer instantly. Customers who previously spent 20 to 30 percent of their day searching for information now spend around 5 percent.

Voice input and auto-formatting addresses a specific construction reality: site managers are often moving around a busy, loud job site with their hands full, unable to type easily. Capmo's AI lets them fill in daily site reports using voice in any language, which then gets auto-translated and formatted correctly. This feature came directly from a product manager who understood the physical reality of working on a construction site.

Document checking uses AI to verify change orders and plan approvals against what is defined in project specifications — work that previously required hours of manual cross-referencing.

Automated scheduling and meeting preparation means the software can take over coordination tasks that previously fell to already-stretched project managers.

The common thread across all of these: the AI handles tasks that were previously done manually, freeing construction professionals to focus on the work that actually requires human judgment.

Why AI in Construction Is Different from AI Elsewhere

Building AI for construction is not the same as building AI for other industries. Florian identifies two things that make construction AI fundamentally different.

The first is the data environment. Construction data is not clean, structured, or standardized. It comes in every format imaginable: scanned PDFs, handwritten site notes, photos taken in poor lighting, voice messages, emails in multiple languages. AI systems built for construction need to handle all of this — what Florian calls multimodal data — without requiring construction teams to change how they work.

The second is the reliability requirement. Construction is a physical, safety-critical environment. An AI system that is occasionally wrong is acceptable in many consumer contexts. In construction, errors have real consequences: cost overruns, schedule delays, safety incidents, legal disputes. This shapes how Capmo builds its AI. "Let the software do that for you" is the goal — but with the domain understanding to know which tasks can be safely automated and which still require a human to decide. These realities also shape construction technology trends toward trustworthy, domain-specific systems.

The Gap Between AI Prototype and AI Product

One of the most practically useful parts of Florian's masterclass is his distinction between an AI prototype and an AI product.

"In the days of tools like Lovable and Cursor, you can build an AI prototype in minutes or hours. It's easy to show in a sales demo. The problem is the gap from prototype to product — that's a huge step."

An AI prototype that processes 2,000 documents may work well in a demo. An AI product that processes 200 million documents and trillions of emails for more than 2,000 customers every day requires a completely different technical foundation.

"The big myth: the gap is not in building the initial AI setup — that's easy. The hard part is making it scalable."

For construction specifically, this scalability challenge is compounded by the need to handle offline functionality. Construction sites often have poor or no internet connectivity. An AI-enabled mobile app that does not work offline is not useful on a job site. This is not a feature. It is a prerequisite.

What AI in Construction Will Look Like in Five Years

Florian's view of where AI in construction is heading is specific and grounded in what he has already seen happen with his own engineering team.

"Looking five to ten years ahead, it's very much about what the software can automate, what the software can do for a construction company, rather than just what the software can help an individual do better."

The shift he describes is from AI as a tool that helps people do their jobs, to AI as something that does significant parts of the job itself.

As a concrete benchmark: an average construction manager today runs around seven million euros of construction volume per year. Florian expects that with AI tools, the same person will be able to run fifteen to twenty million euros in parallel within five years.

He draws the comparison directly from what has already happened with software engineers using AI coding tools. "Their productivity went up enormously. We kept the same headcount and want to hire more, but the output per person has increased dramatically. I expect the same to happen in construction." This direction reflects broader construction technology trends toward automation across project delivery.

Why Domain Understanding Is the Key to AI in Construction

AI tools are only as useful as the domain knowledge built into them.

This is one of Florian's most consistent themes across the masterclass. Capmo spent seven years building deep understanding of construction workflows before applying AI to them. That foundation is what makes the AI useful rather than generic within construction project management software.

"To use AI well, you need both the data and the domain understanding. We have them."

The practical implication for anyone building or evaluating AI in construction: a general-purpose AI tool applied to construction data is not the same as an AI system built specifically for construction workflows. The difference shows up in the details — knowing that a site manager fills out daily reports while walking around a noisy job site, knowing the difference between a general contractor's workflow and a specialty contractor's, knowing what a change order approval process actually involves.

Without that domain knowledge, AI in construction produces outputs that look impressive but do not fit how construction actually works.

Frequently Asked Questions About AI in Construction

What is AI in construction? AI in construction means software that automates tasks construction teams previously did manually — searching for information, filling out reports, checking documents, scheduling work. Rather than software that needs to be filled and managed, it is software that actively helps run the project.

How is AI used in construction project management? AI in construction project management software is used for document search, voice-based site reporting, automated scheduling, change order checking, and plan approval verification. The goal is to reduce the time construction managers spend on administrative tasks and give them faster access to the information they need.

Why is construction data good for AI? Construction generates more data than almost any other industry — contracts, specifications, plans, photos, videos, voice recordings, emails, and more. This volume and variety of data gives AI systems a rich foundation to work with, though the data is often unstructured and multimodal, which requires AI systems built specifically for construction contexts.

What is the difference between an AI prototype and an AI product in construction? An AI prototype can process a small number of documents and work well in a demo. An AI product needs to handle millions of documents reliably, work offline on construction sites with poor connectivity, and scale to thousands of customers without breaking. The technical gap between prototype and product is large.

Will AI replace construction workers? Based on Capmo's experience, AI in construction is more likely to multiply what individual construction managers can handle than to replace them. A manager running seven million euros of construction volume today may be able to run fifteen to twenty million with AI tools within five years — the same person doing significantly more with software handling the administrative and analytical load.

What does "vertical AI" mean in construction? Vertical AI in construction means AI built specifically for construction workflows, with domain knowledge embedded in how the system works, often inside construction project management software. This is different from applying a general-purpose AI tool to construction data. Vertical AI understands the specific processes, terminology, and constraints of the construction industry.