In brief
In 2015 Luke Hansen was working in his family's roofing business in Nebraska. He spent months looking for software to manage jobs properly, didn't find any, and said something that every operator in an unglamorous industry has thought at least once:
In 2015 Luke Hansen was working in his family's roofing business in Nebraska. He spent months looking for software to manage jobs properly, didn't find any, and said something that every operator in an unglamorous industry has thought at least once:
"I ultimately just decided that if I couldn't find something that could do what I wanted, I would just create it myself."
The app he built was called CompanyCam. It does something that sounds almost too simple to be a company: contractors photograph the job site, the photos get tagged with location and time, and everyone — office, crew, customer — sees the same record of what happened.
In November 2025 CompanyCam was valued at $2 billion. B Capital led a Series C that PitchBook put at $415 million. It became the first unicorn in Nebraska's history.
I find this story interesting for a reason that has nothing to do with construction.
What actually got funded
Strip away the vertical and CompanyCam is a bet on one idea: in industries where the work happens outdoors and away from a desk, the photograph is the record.
Not the form someone fills in afterwards. Not the note the foreman types from memory in the evening. The photo, taken at the moment, attached to the place and the time it was taken.
Every downstream benefit follows from that. The homeowner who says the crew never fixed the flashing can be shown the flashing. The insurance adjuster gets evidence instead of assertion. The office knows what happened on a roof forty miles away without calling anyone.
It's worth sitting with how unglamorous this is. There's no model in the original product. There's no marketplace, no network effect in the classic sense. It's a camera, a timestamp, and the discipline to make that the source of truth instead of somebody's recollection.
Ten years and two billion dollars.
Why I keep coming back to it
I own a chain of auto repair shops in Ukraine — AvtoAtmosfera, in Kharkiv and Cherkasy. For years I did exactly what Hansen did: bought the available software, found it didn't survive contact with a working shop, and eventually built my own.
The failure mode in my industry is identical to his, and I think it's identical in every industry where the valuable work happens away from a keyboard.
A mechanic is next to a car with his hands covered in grease. He's found that the customer also needs a tie rod end. To record that, he must stop, clean up, walk to a terminal, find the work order, and type. The repair takes four minutes. Recording it takes six.
He doesn't record it. He tells the service advisor at lunch, and most of it makes it into the system.
Then a month later the customer comes back and says we didn't do the work. Open the system: there's a line item, there's a price, there is no evidence. My word against his. And in that argument the shop loses by default, because the industry has given customers decades of reasons to assume they're being cheated.
Same structure as the roof that may or may not have been repaired. Same solution: make the physical event produce the record, instead of asking a busy person to produce it afterwards.
The difference between a feature and a foundation
The obvious objection, and I've heard it from software people many times, is: fine, add photo upload to the existing system.
CompanyCam's valuation is the argument against that. Photo upload has existed in construction software for fifteen years. What got funded wasn't the ability to attach an image — it was reorganising the entire workflow around the image being primary.
In a record-first system, a missing photo is a data-quality nag. In an evidence-first system, a missing photo means the work was never documented, and everything referencing it is provably incomplete before anyone signs off.
Those are different products, and you can't get from one to the other by adding a button. Which is, I suspect, why incumbents in these categories keep losing to newcomers who look, at first glance, like they built something trivial.
What's different about auto repair
Two things make the automotive version harder and, I'd argue, more valuable.
The event is machine-captured, not human-captured. A roofer still has to take the photo. In a repair shop, the car physically enters a bay past a fixed camera. Recognition happens whether or not anyone is paying attention. That removes the last human decision from the top of the funnel — the point at which most shop systems already have holes.
The dispute is transactional, not structural. A roof is one job. A car comes back four times a year for a decade. Every visit is a new opportunity for the record to disagree with someone's memory, and every dispute compounds into whether the customer returns at all.
That's the system we built at MECH Orbit — the car arrives, the camera reads the plate, the work order opens itself, the mechanic photographs and dictates from the phone already in his pocket, the customer approves specific line items and the system stores what he approved and when. It has been running my own shops for three years.
I want to be precise about limits, because overselling this is how you lose credibility with anyone who actually runs a shop. The camera knows a car stood in a bay for four hours. It does not know why. It can't tell waiting-for-a-part from waiting-for-the-customer from the mechanic being on another job. Anyone claiming their camera diagnoses lost revenue is selling a story.
The pattern worth noticing
Hansen's story has a structure that keeps repeating in vertical software, and it's the opposite of the one startup culture usually celebrates.
He wasn't a technologist who identified a market. He was an operator who had a problem, couldn't buy the solution, and built it because the alternative was continuing to lose money. The industry knowledge came first; the software came second.
That sequence produces a specific advantage that's hard to compete with: a competitor can copy the features in a quarter. He cannot retroactively acquire ten years of standing on roofs.
There are a lot of industries still waiting for their version of this — where the software on offer was designed by people who have never done the work, and where the operators who have done it don't think of themselves as the kind of people who build software.
Some of them are wrong about that.
Sources
- Silicon Prairie News — CompanyCam reaches unicorn status with $2 billion valuation
- Euclid Ventures — The Vertical Report 2026
What to check in your workshop
- Verify the primary source and date behind every figure
- Separate reported facts from the author's interpretation
- Compare the conclusion with data from your own workshop