Broccoli AI

Here is a fact about the American economy that does not get enough attention: a lot of extremely valuable work is done by people who physically cannot answer the phone while they are doing it. A plumber with both arms inside a wall cannot take your call. Neither can an electrician on a ladder or an HVAC tech in a crawlspace. And the call they are missing is, very often, another customer with another emergency - which is to say, another job, which is to say, money that rings twice and then goes to a competitor.

Broccoli AI, a San Francisco company founded in 2021, has built a business on this specific, unglamorous gap. Its product is a set of AI voice agents that answer inbound calls 24 hours a day, book and schedule jobs in real time, follow up on leads, and alert technicians when something is genuinely on fire. The company describes itself as "the AI-native operating system for home service businesses," which is a mouthful, but the underlying pitch is refreshingly plain: never miss a call again.

The trades are a good market for this in the way that boring markets are often good markets. They are enormous, they are fragmented into tens of thousands of small businesses, and the software they run on - the CRMs, the dispatch tools, the scheduling systems - is real and widely adopted. Broccoli's agents plug directly into ServiceTitan, the CRM that a large share of home service companies already use, so a booked call becomes a scheduled appointment without a human retyping anything. That integration is not a garnish. It is the whole point. An AI that answers the phone but can't put the job on the calendar is a very expensive answering machine.

Two engineers who picked the unglamorous door

Broccoli was founded by AJ Jain, its CEO, and Aditya Ketkar, its CTO - both BITS Pilani alumni who, by their own account, had worked together for roughly nine years and built multiple startups before this one. Jain came out of Adobe; Ketkar came out of Microsoft, with a research stint at the CVC computer vision lab at UAB Barcelona along the way. This is a fairly standard pedigree for founders who then go and build, say, a developer tool or a productivity app for other people exactly like them.

They did not do that. They went and built software for plumbing dispatch. There is a version of the AI boom in which the smartest engineers all pile into the same handful of shiny, crowded categories, and there is another version in which some of them notice that the most defensible markets are the ones nobody wants to talk about at a party. Broccoli is a bet on the second version.

The growth, and the money that followed it

The numbers that make investors sit up are less about the technology than about adoption. Broccoli grew from zero to more than 800 contractors in under two years. In 2025 it scaled revenue roughly 20 times and grew the team from about five people to twenty. Then Khosla Ventures led a Series A of $25 million, with participation from Y Combinator, whose Winter 2022 batch Broccoli came through.

A 20x revenue year sounds like the kind of thing that happens because of a viral moment or a clever launch. It usually isn't. In a business like this it happens because the product does a boring thing reliably - it answers a call that would otherwise have rung out - and because each contractor who tries it and keeps it represents recurring revenue that compounds. The unsexy version of hypergrowth is just a lot of small, correct decisions by a lot of small businesses. That is roughly what appears to have happened here.

It is worth being precise about what is and isn't public. The exact annual recurring revenue is not disclosed; "multi-million dollars" and "20x" are the figures the company and press have put forward. The $25M Series A led by Khosla is well documented. The valuation is not public. When a company is growing this fast the temptation is to round every number up and add an adjective; the more useful thing is to note that the trajectory is real and the specifics are partly estimated.

There is also a structural reason a company like this can grow quickly, and it is worth spelling out because it explains why investors like Khosla are willing to write a check into what sounds, on its surface, like plumbing software. The home services market is huge and it is diffuse. There is no single dominant buyer; there are tens of thousands of independent shops, each of which loses real money to missed calls, and each of which will happily pay a monthly fee to stop doing so. That is close to an ideal shape for a software business: a large total market, a clear and quantifiable pain, and a product whose value shows up on the very first missed-call-turned-booked-job. You do not have to convince a plumber of an abstraction. You have to show him a job on his calendar that he would otherwise not have had.

The bet underneath all of this is a bet on where AI is actually useful right now, as opposed to where it is most talked about. A lot of the loudest AI products ask a human to change how they work - to open a new app, learn a new interface, trust a model with a creative task. Broccoli asks the contractor to change almost nothing. The phone still rings the same number. The jobs still land in the same CRM. What changes is that the ringing no longer requires a human, and the after-hours calls no longer go to a $200-a-month answering service that takes a message and gets half the details wrong. The AI is doing something narrow, repetitive, and verifiable, which is exactly the kind of task current models are good at and exactly the kind of task a small business is glad to hand off.

None of this is risk-free. Voice AI in a live customer conversation has to be right, or at least gracefully wrong, because the person on the other end is often stressed, sometimes standing in an inch of water, and rarely in the mood to be misunderstood by a robot. A booking that gets the address wrong or double-schedules a crew is not a rounding error; it is a lost afternoon and an annoyed customer. The companies that win in this category will be the ones whose agents fail softly - that know when to escalate to a human, that confirm the details, that do not confidently invent an appointment slot. Broccoli's framing of its agents as "employees" rather than "chatbots" is partly marketing, but it also points at the right standard: you judge an employee by whether the work actually got done.

What you can actually do with it

For a contractor, the practical offer is straightforward. Broccoli's agents answer every inbound call, including the after-hours and weekend ones that used to go to voicemail or a paid answering service. They talk to the customer, capture the details, book the job into the schedule, and follow up on leads that would otherwise have gone cold. When a call is a genuine emergency - a burst pipe at 2 a.m., no heat in January - the system can flag a technician rather than letting it wait until morning. The framing the company uses is "AI employees for the trades," which is marketing, but it is honest marketing: the agents are meant to do the front-office labor that a small shop can rarely afford to staff around the clock.

The name, incidentally, is Broccoli. Not a fierce animal, not a Latin verb, not a made-up word with a missing vowel. A vegetable. There is something fitting about software for an unglamorous industry choosing the most unglamorous possible mascot - the thing on your plate you're supposed to eat because it's good for you, not because it's exciting. The market Broccoli serves is a lot like that: not exciting, extremely good for the economy, and easy to ignore right up until your water heater dies.

Broccoli operates in a category - vertical AI for the trades - that is getting more crowded, with names like Avoca, Rilla, and Sameday all circling the same contractors, plus the traditional call centers and answering services it is trying to replace. The company's advantage, for now, is a head start on adoption, a deep ServiceTitan integration, and a founding team that chose this market deliberately rather than stumbling into it. Whether that is enough to stay ahead is the open question that a $25 million Series A is meant to answer. For the 800-plus contractors whose phones it already answers, the question is more immediate and more settled: the call gets picked up, and the job gets booked, whether or not anyone is free to reach for the receiver.