Arch.

Here is a fact about heat pumps that will either bore you or reorganize how you think about climate technology, and there is really no in-between. Residential heat pumps come in roughly five standard sizes. Not five hundred. Five. They range from about 6,000 to 18,000 BTUs, and because modern units can throttle their output up and down, you do not need to size one perfectly. You need to size it approximately. "Approximately" is a word that software is extremely good at and human beings, standing in a stranger's basement with a clipboard, are expensive at.

This is the observation that Arch, a San Francisco startup founded in 2023, decided to build a company around. The pitch, in its original form, was almost aggressively unglamorous: contractors were spending three to five hours per potential customer - driving out, measuring rooms, doing load calculations - and then closing maybe 20 to 25% of them. Do that arithmetic and you find an industry burning most of its most skilled labor on deals it will never win.

Arch's answer was an algorithm that estimates the right heat-pump capacity in minutes using public data a homeowner never has to hand over: square footage, room count, the year the house was built. You do not send a technician. You send a quote. And because the hardware tolerates "close enough," the whole thing works well enough to be useful, which is a lower and much more honest bar than "perfect."

Then the pivot that wasn't really a pivot

If Arch had stayed a heat-pump quoting tool, it would be a fine little business and a forgettable one. What happened instead is more interesting. The founders looked at the contractors they were selling to and realized the sizing problem was one symptom of a much larger disease: home-services businesses are drowning in customers they cannot see. They have years of data sitting inside ServiceTitan, the dominant CRM for the trades, and almost no ability to do anything with it.

So Arch became something broader. It now describes itself as an "autonomous revenue engine," which is the kind of phrase that should trigger your skepticism, so let us translate it into plain English. Arch plugs into a contractor's ServiceTitan account and does three things, on a loop, without anyone asking. It finds - scanning a service territory using property records, permits, and demographics to surface homeowners who probably need work. It protects - watching the existing customer base for signs someone is about to drift to a competitor, then firing off a retention campaign automatically. And it grows - digging cross-sell and upsell opportunities out of data the contractor already owns but never mines.

The reason this matters is a single ugly statistic that Arch likes to cite: the average home-services business churns about 36% of its customers every year. The best operators lose 8%. That 28-point gap is not a marketing problem in the usual sense. It is an attention problem. No small contractor has a team that can watch every customer relationship at once. Software can. That is more or less the entire bet.

The part where a McKinsey guy sells to plumbers

Arch was founded by Phil Krinner and Sacha Schmitz, and the fun detail is that neither came up through HVAC. Krinner arrived via McKinsey and solar installation - he has personally felt the sting of pouring hours into a sales conversation and losing the deal. Schmitz brings fifteen-odd years of software engineering. This is the classic outsider setup, where people who have not been numbed by an industry's daily inconveniences show up and ask why anyone tolerates them.

It also produces a genuinely unusual framing of climate work. Most decarbonization startups want to invent a better machine. Arch's insight is that the machines - heat pumps - largely exist, and the actual bottleneck is distribution: getting tens of thousands of small contractors to sell and install them efficiently. Buildings are responsible for roughly 30% of US carbon emissions. The path to shrinking that number runs not through a single laboratory but through a very large number of family-owned businesses with trucks. Arch's wager is that the fastest lever on that problem is not hardware at all. It is boring, unglamorous back-office software.

Investors bought it. In February 2024, Arch announced a $6.2 million seed round led by Coatue and Floodgate, with Gigascale Capital, ReGen Ventures, and MCJ Collective joining, plus angels including the co-founders of Aurora Solar. That is a notably blue-chip cap table for a company whose customers spend their days crawling through crawlspaces.

Does it work? The evidence so far is anecdotal but pointed. Arch says one customer, Pacific Heating & Cooling, generated about $138,000 in 36 days - roughly an 11x return - largely by reactivating people already sitting in its database. That last part is the whole thesis in miniature. The growth did not come from buying new leads. It came from noticing the ones you already had before they wandered off. If Arch is right, the most valuable thing you can sell a contractor is not more customers. It is the ability to finally see the ones in front of them.