Infrakit AI Series – Article 1 of 5 – Decisions in Time
Based on conversations with Teemu Kivimäki, CEO of Infrakit.
Every week I speak with construction company leaders who want to deploy AI. Most of them are not ready — not because the technology is wrong, but because the foundation underneath it is broken. Here is what I mean.
The industry everyone wants to fix
Infrastructure construction has a reputation problem — and it has earned it. Projects run late. Budgets overrun. Contractors fight clients over variations. Subcontractors submit invoices that nobody can verify. And somehow, despite decades of technology investment, the industry keeps producing the same outcomes.
I have spent years working with civil contractors across Europe. The ambition to do better is real. The investment in digitalization is real. But most of it is not working — and the reason is simpler than people want to admit.
Construction companies are trying to deploy better tools on top of broken processes. They add software for reporting, software for machine control, software for surveying, software for project management — and end up with siloed tools, siloed people, and siloed data. Nothing talks to anything else. The project still runs on WhatsApp messages and spreadsheets. And when something goes wrong — when a client disputes an invoice, when a variation is rejected, when a delay needs to be explained — there is no evidence to fall back on.
The problem is not that construction lacks technology. The problem is that technology without a connected foundation just creates more noise.
What an operating system actually means
When I explain what Infrakit does, I sometimes use an analogy that people find surprising. Think about what Windows did for personal computing. Before a common operating system existed, every device ran its own software in its own way. Nothing connected. Nothing shared. The moment a common OS appeared, everything changed — because suddenly all the tools could work together on the same foundation.
Construction needs the same thing. Not another point solution. Not another app for reporting or another dashboard for management. A shared operating system that connects design, field execution, machine activity, and reporting into one continuous workflow — from the moment a tender is won to the moment the final invoice is paid.
Without that foundation, digitalization is not possible. You can add tools indefinitely and still end up with people manually copying data between systems, project managers spending their evenings compiling reports, and company leaders making decisions based on information that is already a week out of date.
The OS enables one connected workflow: design → construction operations → reporting → getting paid. Everything else depends on this being in place first.
Why this matters for AI
Here is the part that most people in the industry are getting wrong right now. Everyone wants to deploy AI. CEOs are under pressure to show they are doing something. Vendors are promising transformation. And companies are trying to layer AI on top of the same fragmented, disconnected data infrastructure they have always had.
It will not work. AI needs data to be useful — real, connected, reliable project data. Machine positions. Survey measurements. Field reports. Design revisions. Photos. Safety observations. All of it captured continuously and connected in the same project context.
If that foundation does not exist, AI has nothing meaningful to analyse. You will get outputs that look sophisticated and mean very little. You will make decisions based on pattern recognition applied to incomplete data. And you will wonder why the technology did not deliver what was promised.
I have seen this play out already. Companies that tried to deploy AI before getting the data foundation right ended up frustrated and sceptical. Companies that built the foundation first — that got their projects running on a connected operating system before asking what AI could do — are now in a completely different position.
AI does not create value on its own. It amplifies what is already there. If what is already there is fragmented and unreliable, AI will amplify that too.
The commercial reality
This is not just a technology argument. It is a business survival argument.
General contractors win public tenders and must deliver data to get paid. In Sweden, Norway, Finland — across Europe — project owners are increasingly demanding verifiable documentation of the work done before releasing payment. A 30 million euro invoice with no supporting data trail is not going to be paid without a fight. And that fight costs time, margin, and relationships.
The contractors who will thrive in this environment are the ones who can prove what they did, when they did it, and how much it cost — in real time, not six weeks after the fact. That requires a connected data foundation. It cannot be assembled retrospectively from emails and spreadsheets after a dispute has started.
The contractors who do not build this foundation are not just leaving money on the table. They are building a fragility into their business that will become very visible very quickly as AI accelerates competition across the sector.
Companies that don’t build the data foundation now risk being out-competed within three to five years by those that did.
Where to start
The answer is not to buy more software. It is to decide, at CEO or COO level, that the project data your company creates will be treated as a strategic asset — captured continuously, connected consistently, and accessible to everyone who needs it.
That means machine control data connected to design models. Survey measurements connected to as-built records. Field reports connected to progress and cost tracking. All of it in one place, in real time, visible to the project team and to company leadership.
When that is in place — when the operating system is running — the question of what AI can do becomes genuinely interesting. Because then AI has something real to work with. Earlier signals. Clearer evidence. Decisions that can still be acted on.
That is the foundation. Everything else follows from it.
Next in the series: Article 2 — The four ways infrastructure projects lose money they already earned.
Decisions in Time · Infrakit AI Series · infrakit.com