He sold his first startup to GrubMarket, then went looking for a harder problem. He found it in HVAC vans and service calls - and named it NOSO Labs.
Most people building artificial intelligence right now are pointed at the same targets. Coders. Lawyers. Marketers. The knowledge workers who already sit in front of a screen all day. Winston Chi looked at that crowded room and walked out the back door, toward a hot attic in the middle of a July afternoon, where a technician is kneeling next to a dead compressor trying to figure out what went wrong.
That technician is the customer Chi cares about. As co-founder and chief executive of NOSO Labs, a company that came through Y Combinator's Summer 2025 batch, he is building AI agents for the field - the HVAC and home-services workers who keep houses warm, cool, and running. It is not the obvious place to plant a flag. That is more or less the point.
The premise behind NOSO is a small observation with big consequences: a field technician is quietly expected to be three people at once. They have to diagnose the problem correctly. They have to explain it and sell the repair honestly. And then they have to finish the paperwork so the business actually gets paid. Almost nobody is excellent at all three.
To understand why Chi ended up here, it helps to look at where he was before. His first company, Butter Technologies, was founded in 2020 and aimed squarely at another industry that tech founders tend to skip: food wholesale. Butter built AI-powered software for distributors - order entry, a cloud ERP with real-time inventory, e-commerce, and payments - the unglamorous plumbing of how food moves from warehouse to restaurant.
In 2024, GrubMarket acquired Butter to strengthen its AI capabilities across the American food supply chain. For a lot of founders, that would be the finish line. Chi treated it as a data point. Both companies share a signature: pick a large, old-fashioned industry where the software is genuinely bad, then bring modern AI to the people doing the work.
"Our mission at NOSO is to empower the trades through technology."
Winston Chi, Co-Founder & CEO, NOSO LabsThat word - empower - matters to how NOSO is designed. The pitch is not that AI replaces the technician. It is that AI makes one technician as effective as several. NOSO's own shorthand for it is blunt: 10x your field techs, with zero extra headcount. The company is small, roughly six or seven people working out of San Francisco, which makes the ambition of aiming at a national trade workforce feel appropriately audacious.
The product is built to fit the day of someone who spends it in a truck, not at a desk. Before a technician even leaves for a job, NOSO can hand them a brief on what they are walking into. On site, it can listen to the conversation with the homeowner, then help turn that into a clear, measured proposal - the repair, the options, the rebates - rather than a rushed guess scribbled on a clipboard.
Underneath it sits the part that is hard to copy: a library of roughly 20 terabytes of equipment manuals, so a technician never has to have memorized the spec sheet for every furnace and heat pump ever made. NOSO connects into the systems these businesses already run on, like ServiceTitan and Housecall Pro, and follows up automatically on estimates that never got a yes.
That last decision says a lot about how Chi thinks. Plenty of software fails in the field not because it is weak, but because nobody downloads it. NOSO works over text. No new app, no training day, no fighting with a technician who is already busy. The best tool is the one people actually use.
Chi is not building this from a distance. In October 2025, NOSO signed on as a platinum sponsor of the Bring Back the Trades Skills Expo at Gillette Stadium - an event drawing thousands of students, parents, educators, and trade professionals, with scholarships handed out to kids weighing a future in the trades. It is a telling choice of stage. Not a developer conference. A stadium full of the people NOSO exists to serve.
He is doing it with a co-founder whose resume points the other way, deep into research: Alex Xi, a former principal engineer who led work on code-generating language models, with time at ByteDance and Meta AI and a PhD. It is a classic pairing - a builder who keeps choosing overlooked markets, next to a researcher who can make the hard AI actually work. Chi himself studied at Tufts before any of this.
The goal is not to replace the technician. It is to make one technician as good as three.
There is a bet buried in all of this, and Chi is making it out loud. The consensus says the big AI money is in white-collar work. He keeps walking the opposite direction - toward wrenches, refrigerant, and service vans - on the theory that the industries with the worst software have the most room to improve. He was right about food wholesale once. NOSO is the wager that he is right again, this time in your basement.