NOSO LABS Built the AI That Chases the Estimate Nobody Had Time For

Ask any home-services owner where the money goes, and after a beat they will tell you it does not go anywhere - it just sits there. A technician drives out, opens a panel, writes a $4,200 estimate for a new furnace, and the homeowner says they need to think about it. Then the tech drives to the next call. The estimate is never mentioned again. Multiply that by a full board of jobs and you get the number NOSO LABS likes to put on its dashboard first: $47,000 in unsold estimates, sitting quietly, waiting for a follow-up that almost never comes.

NOSO LABS, a San Francisco company in Y Combinator's Summer 2025 batch, was built to send that follow-up. Its product is an AI agent that works alongside field technicians in the trades - HVAC first, then plumbing, electrical, and pest control - handling the parts of the job that have nothing to do with turning a wrench. It follows up on estimates over text, keeps notes, syncs the CRM, and when something gets complicated, it steps aside and hands the job to a human. The pitch on the company's site is blunt: "10x your field techs. 0 extra headcount."

The ProblemDiagnose, sell, or file paperwork - pick two

The trades have a staffing math problem that no amount of hiring solves. A field technician is expected to be three people at once. They have to diagnose the issue, which takes a mechanic's instincts. They have to sell the fix, which takes a salesperson's read of the room. And they have to document all of it, which takes the patience of a clerk. NOSO's founders started from a simple observation: almost nobody is good at all three, and the person who can diagnose a failing compressor by the sound it makes is usually the least interested in filling out forms afterward.

The gap between those three jobs is where revenue goes to die. A great diagnosis with no follow-up is a lost sale. A sale with no paperwork is a scheduling headache. NOSO's bet is that AI does not need to replace the technician - it needs to cover the two-thirds of the role the technician did not sign up for. That framing matters, because it is the difference between a tool a crew resents and one they quietly rely on.

It also explains why the trades have been slow to adopt software that promises to fix this. Most tools assume the technician will change their behavior - learn a new app, tap through a checklist, remember to log the outcome of a call. In a job where your hands are dirty and the next appointment is fifteen minutes away, that assumption breaks on contact with reality. The forms go unfilled, the data goes stale, and the expensive system of record becomes an expensive filing cabinet. NOSO's founders seem to have concluded that the winning move is to ask the technician for nothing at all beyond a reply they were going to send anyway.

The ProductAn agent that lives in the text thread

The design decision that makes NOSO interesting is what it refused to build. The obvious move for a field-service startup is another app - a login, a dashboard, a set of forms for the tech to fill in between jobs. NOSO went the other way. Its agent works over plain text messages. A technician replies the way they would text a dispatcher, and the AI does the rest: it drafts the follow-up to the homeowner, tracks who responded, updates the CRM, and books the job when the answer comes back yes. The company says the agent handles roughly 95% of that administrative loop, and that connecting it to an existing CRM takes under five minutes.

The remaining 5% is the part worth paying attention to. When a request is complex, high-value, or simply beyond what the agent should decide on its own, it escalates - routing the conversation to a back-office dashboard where a human can review and step in. That escalation model is the quiet admission that keeps the product honest. NOSO is not claiming the AI closes every deal. It is claiming the AI does the tireless, repetitive chasing that a busy shop never gets around to, and knows when to raise its hand.

The NumbersSelling recovered revenue, not seats

Most AI startups sell a subscription and hope you feel the value. NOSO sells a number you can already see: the dollar amount of work you quoted and never closed. That is a shrewd position. A five-person HVAC shop does not want to hear about large language models. It wants to hear that the $47,000 on its board could become booked jobs without hiring a follow-up coordinator. In one six-month example the company points to, that recovered figure reached $676,000 - revenue that came not from new leads but from estimates the business had already earned the right to.

There is a second reason this framing works. Recovered revenue is measurable in a way most software value is not. A shop can look at its own board before NOSO and after, and count the jobs that got booked from estimates that would otherwise have gone cold. That makes the buying decision less an act of faith and more an act of arithmetic. For a small operator who has been burned by software that promised efficiency and delivered another login, "here is money you already earned" is a far easier sentence to say yes to than "here is a smarter way to work."

The company itself is still small - reported revenue of roughly $330,000 with a team you could fit around one table. That is not a knock; it is the point. The people building NOSO have run the experiment where a tiny team ships something real before it raises a war chest. Winston Chi, the co-founder and CEO, is a two-time founder whose previous company, Butter Technologies, was acquired by GrubMarket. His co-founder, Alex Xi, worked on AI at ByteDance and Meta and holds a PhD. The unusual thing is not that two technical founders built an AI agent. It is that they pointed it at the least glamorous industry they could find.

The MarketEnterprise-grade AI for the strip-mall contractor

Field service is a large, fragmented, deeply un-digitized industry. The incumbents - the ServiceTitans, Housecall Pros, and Jobbers of the world - are systems of record. They store your customers, your invoices, your schedules. What they have historically not done is act on your behalf. That is the seam NOSO is working. It does not want to replace the CRM; it wants to be the thing that reads the CRM and then goes and does something about it. In practice that means NOSO can sit on top of the software a shop already pays for and turn its dormant data into outbound follow-up.

There is a credibility detail that tells you who NOSO is really selling to. The company carries SOC 2 Type I and II certification, GDPR and CCPA compliance, 256-bit encryption, and a 99.9% uptime commitment - the kind of security posture you would expect from a company selling to banks. It is instead selling to pest-control operators and HVAC crews. That mismatch is deliberate. Larger service businesses and franchise groups have compliance requirements, and a startup that can pass a security review gets to sell to the accounts that actually have budget.

Where NOSO sits

Systems of record

CRM

Systems of action

NOSO

The technician

Text

What You Can Do With ItThe tireless coworker nobody has to manage

For an owner, the practical value is a coworker who never forgets to follow up, never gets awkward about asking for the sale a second time, and never lets a Friday estimate go cold over the weekend. For the technician, it is one fewer thing to carry - reply to a text, keep driving. For the back office, it is a queue of only the conversations that actually need a person, instead of a spreadsheet of everyone who ever got a quote. The appeal is not that the AI is clever. It is that the work gets done whether or not anyone remembers to do it.

NOSO frames the sales side carefully, describing its approach as ethical and measured rather than a machine for squeezing every homeowner. That is a smart line to hold in an industry with a reputation for pressure tactics. An AI that pesters customers into a corner would poison the brand it is attached to. An AI that politely reminds someone about the furnace quote they asked for is just good service. Which of those NOSO turns out to be in the field is the thing worth watching as it scales.

The name is a small tell of the ambition. NOSO describes itself as "the source of intelligence for field technicians" - not the follow-up tool, the intelligence layer. Estimate follow-up is the wedge, the thing that pays for itself on day one. The longer arc is an agent that rides along on the whole job: diagnosis support, notes, scheduling, and whatever else the person in the van should not have to hold in their head. For now, it earns its keep by chasing the estimate nobody had time for. That turns out to be plenty.