Two AI-backed PE firms confirmed the same thesis. They want your firm. Zero RFI raised $13M with three day-one buys. E3 Tech rolled up $1B+ in trade contractor revenue. Plus: a humanoid robot lands on a UK tier one site for the price of a small van.

THE EXECUTIVE BRIEFING
THIS WEEK’S KEY TAKEAWAYS

Full episode write-up at the bottom↓

A few weeks back, Zero RFI raised $13M from General Catalyst and made three day-one acquisitions. E3 Tech, backed by Andreessen Horowitz, bought two trade contractors with over $1B combined revenue. The thesis: roll up the firms, embed AI tools internally, sell outcomes.

Fifty-three percent of US construction workers are expected to retire by 2036. Around 25,000 specialty contractors form E3's acquisition pipeline. Owners want a succession plan, kids don't want the firm, AI-backed PE has the checkbook.

Tilbury Douglas became the first UK tier one to deploy a humanoid on a live site. The robot, on a Unitree base with an Nvidia compute backpack, cost roughly £15k. Mark Buckle's rule: pick the task first.

7 THINGS WORTH YOUR ATTENTION
ON THE RADAR THIS WEEK

POWERED BY:

I spoke to two AI-backed private equity plays in construction who confirmed the same bet. They are buying construction firms outright, embedding their own AI tools inside, and reselling the entire thing as an outcome to the owner.

That changes the question every construction executive should be asking right now. The conversation has shifted from whether AI is coming for your business to which category you belong to: target, acquirer, or differentiated. And what you plan to do about it before someone else makes the call for you.

Move one. Zero RFI. Founded by KP Reddy. Backed by General Catalyst. Roughly $13 million in seed funding. Three day-one acquisitions: an owner's representative firm called Brookwood Group, a building data company called Building Works, and KP's advisory practice, KP Reddy Co. The thesis is direct. Roll up the owner's rep space, embed AI tools internally, and resell the entire thing as a service.

Move two. E3 Tech. A joint venture between PE veteran Rudy Adolf, who took Focus Financial Partners public and then private again for over $7 billion, and Andreessen Horowitz. Two day-one acquisitions, both trade contractors: Sylvan and Andy J Egan. Combined revenue over $1 billion. The playbook is the same as Zero RFI's, applied further down the value chain at the mechanical, electrical, plumbing and conveyance levels. Zero RFI works at the owner's representative level above them. 

The team at Zero RFI was direct when asked whether owner's rep firms are acquisition targets or roadkill. Their answer: if you are not able to adopt tech, you will become roadkill. Owners will pay more for certainty. Firms that can deliver that certainty will eat the firms that cannot.

E3 frames the same idea with a softer edge. Anita at E3 Tech put it like this:

E3's strategy: incubate startups, embed them inside the trade contractors they acquire, then let those startups sell to the wider market. Around 25,000 specialty contractors operate across the United States. That is the acquisition pipeline. Target profile: 50 employees and below, owners ready for a succession plan, family with no interest in running a plumbing business, and a firm without the resources to build its own technology.

There’s a broader population trend underpinning all of this that few people talk about openly. By 2036, experts expect 53% of US construction workers to retire. The same generation that built these mid-market firms is hitting retirement age with no succession plan. AI-backed private equity has walked in with a checkbook and a tech stack and started picking them off, one by one.

PE rollups in construction services have been tried before. Renovo Home Partners, General Catalyst-backed, folded last year. Pitchbook reported VCs were skeptical AI rollups even worked.

So why is this round different? Two reasons.

First, neither Zero RFI nor E3 Tech is building total proprietary AI infrastructure. They are not training their own large language models. They are using what is available in the market today, building the bits that do not exist for the interstitial spaces, and embedding the tools inside services that operate within the industry's existing contract structures. As David at Zero RFI put it, the strategy is to help the general contractor perform better while leaving the contractor structure intact. That is a far more conservative thesis than most AI rollup plays.

Second, the cost of building the tool you actually need has collapsed. David built an estimator tool for Building Works in two days, using Claude and modern AI development tools. He has, in his own words, never coded in his life. His view: construction tech software as a standalone business model is almost dead. Mid-sized firms can now build for themselves what they used to license from a vendor. Personally vibe-coded apps are usually a long way from commercial-grade, but the direction of travel is clear.

Confirming the same point from another angle: AEC Magazine published a piece this week arguing that the agentic BIM tools released by Autodesk, Bentley and Trimble sound smart but are missing the behind-the-scenes plumbing (reliable data flows, permissions, integrations, and process support) needed to run real day-to-day project work. Some firms are now being charged to access their data through the platforms they pay for. Some are starting to move outside the vendor ecosystem entirely. We covered Trimble's Document Crunch acquisition and Autodesk's close on Rumbix earlier in April; the platform consolidation thread keeps lengthening.

While the PE players are busy redrawing the firm side of construction, the worker side is moving too. Tilbury Douglas became the first UK tier one to deploy a humanoid on a live construction site. He is called Douglas. Unitree base unit, customized with an Nvidia AI compute backpack, LiDAR, photogrammetry, collision avoidance. Total cost: roughly £15,000. About the same as a small van.

Mark Buckle of Tilbury Douglas walked us through the deployment (video coming soon). Most firms approaching new technology start with the technology. Mark's team did the opposite. They identified three repetitive tasks eating one to two hours a day per site team member:

Lock the tasks down first. Then backfill the technology to solve them. 

As Mark put it on the show: "Don't start with the robot. Start with the task."

A few weeks ago, we cited the Zacua Ventures construction robotics report in this newsletter, which argued that humanoids were hype and task robots were real. So how do we square that thesis with a UK tier one putting a humanoid on a live site? The honest answer: Douglas is a “Scout” in the Zacua taxonomy. His job is data collection; he is not concerned with bricklaying or weilding. The form factor matters because humanoids navigate stairs, doors, and tight site logistics better than dogs do. The vision system sits higher up, which improves LiDAR scans and photogrammetry. Same task work. Different chassis.

Two more developments from this week, both pointing the same way. London-based startup All3 raised $25 million to build a production system around a different legged robot called Mantis, with a German first deployment targeted for later this year. North America's Building Trades Unions just announced a partnership with Microsoft to bring AI tools to tens of thousands of trade workers, focused on code interpretation, paperwork automation and compliance updates. Robots on site. AI in pockets. Both are moving in the same direction.

And the workforce reaction? Mark says the first thing every site worker wants is a selfie with him.

So which of the three buckets are you in?

Targets are firms with revenue, real client relationships, and a thin technology layer; the kind of business an AI-backed PE play wants to buy outright, plug its own tools into, and resell as an outcome. Acquirers have the balance sheet and the platform to do their own buying. Differentiated firms are smaller and rarer, holding a specific edge (a niche, a method, a relationship) that an acquirer would prefer to partner with on its terms.

Most firms quietly assume they are acquirers. Few actually are. Two AI-backed PE plays just started making that call from the outside, and the demographic clock is running in their favor.

Next week we go deeper into the humanoid playbook. What does it actually take to deploy one? What works? What doesn't? We also sit down with the people putting Palantir inside construction businesses. The line one of them gave us was simple: "ERPs are in trouble." And we drop our episode with the CEO of Nemetschek Group on the HCSS acquisition, the next chapter in the platform consolidation story running through this edition. 

Subscribe so you don't miss it.