Construction AI: a builder ships software that runs the whole company, and Procore's record quarter shows who's paying for the old way

On 30 July a Las Vegas outfit called Sledge came out of stealth with something the category has been circling for a while: not a tool that helps with one task, but an AI operating system meant to run a contractor's entire company. Its own words are "bid to build to paid". The agents write the bid, run the job, raise the invoice, chase the late payment, and keep the back office ticking, with the contractor approving and the software doing. It says it has nearly 20 million dollars of project volume under management while still in private beta, which is the combined value of the jobs its early users are running through it. That's a vendor number and beta is beta, so read it as a signpost, not a track record.

What makes me take it seriously is the founder. Raz Danoukh still runs Ferrocrete Builders, a concrete contractor in Los Angeles he describes as north of 100 million dollars in turnover. So the software is being built and tested against a live job book, by someone who's spent years doing the very admin he's now trying to hand to a machine. That's the opposite of a generalist AI team looking for a vertical. It's a builder who got sick of the paperwork and wrote his way out of it.

This is the shape I mean by challenger construction software, the broad AI-first platforms taking on the big incumbents the way the challenger banks took on the high street. Sledge clears the two tests that matter most: the AI carries real workload rather than sitting on top as a chatbot, and it's broad by design, covering how a whole project runs instead of one slice. The two I can't verify from a beta are the ones that decide whether it's a true challenger or just a clever demo: is it simple to buy with pricing you can see, and are you free to leave with your data. Those are the questions I'd put to Sledge before anyone in the UK got excited, and it's US-focused for now regardless. But the direction is right, and it's another name in a category that only becomes real when it holds more than one company. Worth watching whoever brings this model to a British site first, because the compliance load here is heavier than in Nevada.

Two days earlier, on 29 July, Procore reported the strongest quarter in its history. Revenue of 375.2 million dollars, up 15.8 per cent on the year and a touch ahead of what the market wanted. Its first-ever quarter of GAAP operating profitability. Free cash flow up sharply. Gross retention at 95 per cent, and 2,871 customers now handing over more than 100,000 dollars a year each, a number up 14 per cent. It lifted its full-year guidance to between 1.51 and 1.514 billion dollars. By any normal reading, that's a company executing well, and I'm not going to pretend otherwise.

But look at the engine underneath the headline. The growth is coming less from a rush of new firms signing up and more from getting more money, year after year, out of the base already inside the walls. That's the logic of Procore's pricing, which scales with the construction volume you run through it and with the modules you switch on, so your bill climbs as your business does whether or not you use more of the product. Long-standing customers have been saying for a couple of years that their effective rate has roughly doubled over five to seven years (customer-reported, via published pricing analyses). A record profit built partly on that is exactly the incumbent behaviour that makes the case for the challengers.

The comparison only goes so far, because Procore genuinely does a lot and does it at scale. But the contrast with Sledge is the useful bit. One model sells you a job finished and, in principle, lets you walk. The other sells you a platform that gets more expensive the more successful you are on it, and makes leaving hard by design. For the person doing the paperwork, and the one signing the renewal, that difference is the whole game. Read the pricing model and the exit terms before the feature list, every time.

Step back from the two announcements and the pattern is clear. The category is being defined this year from both ends at once. From below, builders like Danoukh are shipping AI-first platforms that do the work rather than describe it. From above, the incumbents are posting record numbers that reveal, if you read past the headline, how much of the old model runs on lock-in and rising bills. Both things happening in the same week isn't a coincidence, it's the market sorting itself out.

For a UK contractor none of this is abstract. You don't have to bet on which company wins to benefit. What a real category gives you is choice and negotiating room: more than one credible way to get the admin off your team, and a stronger hand at renewal when the incumbent's price creep shows up in your inbox. The discipline is the same one these briefs keep coming back to. Whatever you buy, make sure the AI finishes real tasks you can check, make sure you can see what it costs before you commit, and make sure your data walks out with you if you leave. Those three questions do more work than any feature comparison.

The procurement filter: at your next software renewal or trial, ask for one thing: show me a job this ran end to end, then show me the exit. If a vendor can't do both, you've learned what you needed to.