Higharc just raised a $95 million Series C, pushing its total past $170 million, and bolted on a supply-chain deal with US LBM. The same week on the show, OpenSpace CEO Jeevan Kalanithi walked us through his push to give construction AI agents “eyes.” Two different companies, one bet: turn the physical building into structured data, and the software (and the money) follows.
Two things landed around the same table this week that look unrelated until you squint. Marc Minor came on to talk about Higharc’s fresh $95 million round. Then Jeevan Kalanithi from OpenSpace explained why his team is building AI agents that can look at a job site and actually see what’s there. Line them up and it’s the same story told twice.
Here’s what both companies worked out. A home, or a job site, isn’t really a pile of drawings. It’s data pretending to be drawings. Once you represent it as data the way software understands data, everything downstream gets faster, cheaper, and a lot less manual.
The smart money is chasing that boring, hard-won layer underneath the tools: a structured model of the physical thing that an algorithm can actually trust. Not the flashiest demo. The plumbing.
The companies winning construction AI aren’t the ones with the slickest launch video. They’re the ones who turned the building itself into data an algorithm can rely on.
The $95 million bet on homes as data
Why Higharc raised so much
Higharc builds software for home builders that collapses three usually separate jobs into one. Design, estimating, and the buyer’s sales experience all run off a single 3D model of the house. Marc put it plainly on the show: they’ve had the joy of building three companies at once, which he happily admits nobody should ever set out to do on purpose.
The round was led by Insight Partners, with Wellington Management and a long list of existing backers joining. It takes Higharc’s total funding past $170 million. Alongside the raise, the company announced a deal with US LBM, the largest privately owned distributor of lumber and building materials in the US, pushing its estimating tech into the materials supply chain.
The mechanism is the interesting bit. Higharc’s AutoTranslate AI takes any floor plan image and turns it into a live 3D spatial data model. Change the home, and the drawings, the estimates, and the shopping experience all update together, because they’re reading from the same source. We broke down an earlier version of this, a sketch-to-buildable-home demo, a couple of months back, and this cheque is the market voting yes.
- Disconnected by default. Today a home gets drawn, then PDFs get passed to the estimator, the sales team, and the subs, and everyone rebuilds their own version of the data by hand.
- Change is constant. Homes aren’t stamped out once like a phone. Every tweak ripples through, and disconnected files bake waste in deep.
- The payoff. Higharc says its builders have cut product cycles from months to weeks, opened communities 25 to 50 percent faster, and lifted margins 10 to 15 percent.
OpenSpace wants to give AI agents eyes
Turning the job site into data an agent can use
OpenSpace comes at the same problem from the field side. Founded back in 2017, it got known for reality capture: strap a 360 camera to a hard hat, walk the site, and get a searchable visual record that maps to the plans. It plugs into Procore and Autodesk, so nobody has to change how they already work.
Now Jeevan and his team are building what they call agent space internally, an early-access set of AI agents that do jobs for builders. His framing is sharp. An agent is only as good as the data it can reach, and in construction the money gets made or lost in the field, not in a spreadsheet. So the agents have to be able to look at the site and see what’s really going on.
That’s also why OpenSpace bought Disperse late last year, a progress-tracking outfit it had worked alongside for a long stretch before the deal closed. We dug into that acquisition, and the birth of “spatial AI,” in an earlier issue. The shape of it matches Higharc exactly: own the trustworthy record of the physical thing, and the useful software finally has somewhere to stand.
Builders’ agents need eyes. They need to be able to see the field.Jeevan Kalanithi, CEO, OpenSpace
The construction tech moves that matter, before everyone else clocks them
Deals, tech, and honest takes for people who build, fund, and sell into the industry. One email, no fluff.
Join 3000+ ReadersWhat it means if you build, fund, or sell into construction
The pattern the smart money is following
Strip away the two logos and the lesson fits on one line: whoever owns the trustworthy data model of the building owns the workflow that runs on top of it. That’s why a $95 million cheque lands on design automation that spits out build-ready data, and why the reality-capture crowd are suddenly best placed to ship agents.
If you build, this is the good kind of boring. Real numbers show up earlier, rework drops, and the tools finally reflect the thousand homes you’ve already put up instead of starting every job from scratch. If you fund or sell into the space, the tell is usage, not demos. Jeevan was refreshingly blunt about the trap of raising money on deals that nobody actually opens twice.
And if you’re watching from the outside, here’s the quiet shift worth clocking. Construction is turning into a data business. Jeevan reckons this visual-data-plus-AI thing gets big well beyond construction, across the whole real-world economy, with new foundation models being built natively around spatial data. The buildings are becoming readable. The software is learning to read them.
Higharc vs OpenSpace at a glance
Same thesis, two ends of the build
Both are turning a physical thing into data software can act on. They just start from different ends of the process, one at the design desk, one out on the site.
| Company | Turns into data | Core product | Built for | Latest milestone |
|---|---|---|---|---|
| Higharc | The home itself | Design, estimating and sales off one 3D model | Builders putting up 20+ homes a year | $95M Series C, US LBM deal |
| OpenSpace | The job site | Reality capture plus AI agents (“agent space”) | GCs and owners tracking live builds | Disperse acquisition, agent early access |
A $95 million Series C led by Insight Partners, taking total funding past $170 million, plus a partnership with US LBM to bring its AI estimating into the building-materials supply chain. (Source)
Instead of storing a house as static drawings, Higharc represents it as a live 3D data model. The drawings, the cost estimate, and the 3D sales experience all get generated from that one model, so a change in one place updates everywhere. (Source)
Insight Partners led, with Wellington Management joining alongside existing investors including Fifth Wall, Spark Capital, Lux Capital, SE Ventures, Simpson Strong-Tie and others. (Source)
OpenSpace captures the job site as a visual record and is now building AI agents (internally “agent space”) that act on that data. The thesis is simple: agents have to see the field to be worth anything. (Source)
Because in building, the money is made and lost in the field, not in documents. An agent that can’t see what’s actually built can only guess. OpenSpace’s job is to hand agents that sight. (Source)
Not yet. It’s purpose-built for US single-family homes, both detached and townhomes. Try a skyscraper or a bridge and it won’t work. The depth comes from the narrow focus. (Source)
The US LBM deal is aimed partly at giving small builders, of which Marc counts roughly 40,000 in the US, access to the kind of pricing and planning power only the big builders have had. (Source)
Related Articles
Higharc Just Automated Architecture’s Most Repetitive Job
Inside OpenAI’s Construction Power Move
