ServiceNow down 28%. Salesforce down 26%. Intuit down 34%. The software stocks that seemed like safe, recurring-revenue machines are getting revalued fast.

Wall Street is finally catching up to something that’s been clear to anyone paying close attention: most enterprise software is a database with a nice interface. And AI doesn’t need the interface.

What “system of record” software actually is

A plain description of most project management software, including the platforms your GC and architect are using on your projects right now:

Tables. Relational connections between tables. Forms to put data into the tables. Reports to pull data out of the tables.

That’s it. The design is elegant, the marketing is sophisticated, the pricing is not cheap, but underneath, it’s a place to store and retrieve information. The actual work still happens in email, texts, meetings, and spreadsheets. The software is where you record what already happened.

This was always true. It just didn’t matter as much when the alternative was paper.

The alternative now is AI that can take unstructured information (conversations, documents, photos, voice notes) and produce the structured output you actually need, without requiring anyone to manually enter data into tables. The fundamental value proposition of most SaaS software just collapsed.

Why this matters for your project technology

The construction tech landscape is littered with platforms that promised to transform project delivery and ended up as expensive archives. Procore. Autodesk Build. Dozens of point solutions layered on top of each other, each adding a subscription cost and a training burden, none of them actually changing how decisions get made.

The firms clinging to those platforms as their technology strategy are about to face a painful reckoning. Not because the platforms are being shut down, but because competitors who’ve moved to AI-native workflows will be able to do more, with smaller teams, at lower cost. The efficiency gap will become visible in proposal pricing, in staffing structures, in what they can offer you for the same fee.

The “vibe coding” moment

A venture capitalist at a conference challenged a CRM consultant on the future of his business. Two days later, the consultant came back with a functional CRM he’d built himself using Replit, designed specifically for general contractors, doing exactly what he needed. It took him twenty hours.

He wasn’t a software engineer. He was a domain expert who understood the problem. The tool landscape made it possible for him to build his own solution in less time than most software implementations take to scope.

This is the actual competitive threat to legacy SaaS: not a better platform, but the ability for people who understand the problem to build exactly what they need, without paying for what they don’t need.

For building owners, this creates an opportunity. The technology you need to run your capital program (your specific workflows, your reporting requirements, your data integration needs) no longer has to come off a shelf. The question isn’t which platform to buy. It’s what outcome you need and who can build it.

What survives

Some categories of software will be fine. Tools with genuine technical depth, network effects that make them more valuable with more users, or proprietary data that can’t be replicated by a domain expert with AI assistance. Those have real moats.

Most construction technology doesn’t meet that bar. It meets “this is the tool everyone uses so everyone uses it.” That’s not a moat. That’s inertia.

The technology that survives the current shift will be tools that deliver outcomes rather than interfaces. Not “track your RFIs” but “ensure your project doesn’t get surprised by a late RFI that triggers a change order.” Not “store your documents” but “surface the specific document risk that’s going to cost you money in the next thirty days.”

The distinction sounds subtle. The business model implications are enormous. And the owners who understand the difference will choose better tools and better partners.

The software you’re paying for was already becoming obsolete. Now the timeline just got shorter.