Ask any contractor why their AI tools aren’t delivering and they’ll tell you that the data is a mess. It’s in text messages, emails, handwritten site notes, different software platforms, or a thousand other locations. Construction doesn’t have a shortage of information. It has a crisis of fragmented, unstructured, and unmanageable information, and until that gets solved, no construction firm can utilize the full capabilities of AI.

The data problem in construction didn’t start with AI. It’s been building for decades as the industry layered tool on top of tool without ever unifying what lives underneath them. Scheduling software, procurement spreadsheets, and RFIs all having to be juggled because there is no unification. The result is an industry that generates enormous amounts of data and can access almost none of it coherently.

AI has made this problem impossible to ignore. Up to 80% of enterprise information is locked in unstructured formats across disconnected systems, and this fragmentation is the primary obstacle to feeding AI initiatives the coherent contextual information they need to succeed. Construction sits at the severe end of that spectrum. A typical commercial project generates data across dozens of parties in formats ranging from clean digital files to photos of handwritten notes.

Gartner predicts 60% of AI projects lacking AI-ready data will be abandoned through 2026, and that rate is already at 42% of companies (source: sranalytics.io/blog/why-95-of-ai-projects-fail). For construction specifically, those numbers are likely conservative.

58% of organizations admit they don’t have a well-defined data foundation, yet they’re investing in AI tools anyway. They hope the technology will somehow compensate for the chaos underneath it. It won’t. If a metric is defined in multiple ways, AI will produce multiple answers. Feeding a fragmented construction project into an AI tool doesn’t produce insight. It produces faster confusion.

The firms making real progress on AI in construction share one trait that has nothing to do with which software they bought. They fixed their data foundation first.

Companies that centralize their workforce data around experience, skills, and availability are seeing 3x higher growth rates than those that don’t, even when facing similar conditions. The same principle applies across every data type in construction. Centralized, verified, governed data doesn’t just make AI work better, it makes every decision on a project better.

The technology industry has started building specific tools to attack this problem. Trunk Tools, one of the most watched construction AI startups, takes a different approach from most platforms. Rather than requiring clean structured data before it can function, it’s specifically designed to pull answers from fragmented sources such as emails, documents, or contracts, and reconcile them into a single response. It’s an acknowledgement that the data problem in construction is real and that waiting for perfect data before deploying AI isn’t a viable strategy for most firms.

Procore and Autodesk are both investing heavily in AI layers that operate within their ecosystems precisely because they understand that unified data is the prerequisite for everything else. Their long term competitive advantage isn’t their features, it’s the structured data their platforms have accumulated over years of industry use.

The data fragmentation problem hits differently depending on where you sit on a project. For project managers it means spending hours hunting for information that should take seconds to find. For estimators it means basing bids on incomplete pictures of project history. For executives it means making strategic decisions from reports that reconcile three different versions of the same numbers. And when disputes arise, it means fighting a $60 million claim with evidence scattered across a dozen disconnected systems.

Gartner confirms that 57% of organizations report their data is not AI-ready. In construction, that number is almost certainly higher. The implication is that most firms investing in AI tools right now are building on a foundation that will undermine those investments before they deliver meaningful returns.

The firms that will extract real value from AI in construction over the next five years aren't necessarily the ones buying the most sophisticated tools. They're the ones doing the unglamorous work of getting their data organized, centralized, and trustworthy first. That work is slower and less exciting than buying a new platform. It's also the only thing that makes the new platform worth buying.

Construction generates more data per project than almost any other industry. The problem has never been a shortage of information. It's that the information is everywhere and nowhere at the same time. Fragmented across systems, formats, parties, and platforms, it resists every AI tool thrown at it and produces the disappointing results that are fueling construction's AI skepticism.

Only 27% of business leaders say their data, processes, and applications are well connected enough to support AI, even as 94% agree that connected data is essential to making it work. The firms that make the most out of AI are not the ones making headlines by purchasing the newest software. It’s those that solidify the foundation before building on top. If you want to get the most out of AI, fix the data. Then buy the tools. The results will show naturally.

New issue every Tuesday. Subscribe to get it delivered straight to your inbox.

Has your firm made it a priority to centralize project data? Let us know in the comments.

Krish Sule & Justin Ranisate