Every other AI receptionist company we look at is chasing enterprise.
Smith.ai talks about legal and professional services. Synthflow's homepage advertises "65M+ calls per month" across BPO, healthcare, and financial services. The YC batches keep producing AI voice startups whose first customer is a 500-seat call center.
We're going the other direction. We're building for the roofing company in Knoxville and the veterinary clinic in the Chicago suburbs — the service businesses we help every day. Here's why that's a deliberate choice, not a consolation prize.
The TAM is bigger than people think
There are roughly 420,000 HVAC companies in the United States [VERIFY]. Tens of thousands of veterinary clinics [VERIFY]. Hundreds of thousands of roofing, plumbing, and electrical contractors. Add the rest of home services and you're already in the millions globally.
Even at a low average revenue per customer — call it $20,000 to $30,000 a year, fully loaded — this is a multi-billion-dollar market. It doesn't look like a billion-dollar market because it's fragmented across a thousand small towns, but the dollars are real and the dollars add up.
The mistake people make is assuming "small business" means "small market." It doesn't. It means distributed buyers, which is a sales problem, not a TAM problem.
The enterprise market is crowded; the SMB service market isn't
Walk into the AI voice space today and the enterprise lane is full. Every well-funded competitor is fighting for the same Fortune 500 logos. The differentiation there has to be raw model quality or some narrow compliance feature, and both of those get commoditized quickly when the underlying model providers (Anthropic, OpenAI, Google) keep releasing better foundations.
The service business lane is almost empty. There are call answering services that have been around for 40 years and don't book appointments. There's voicemail, which has roughly an 85 percent miss rate when a customer calls a contractor and gets a voicemail box. There's hiring a $40,000-a-year front-desk person, which most small operators can't justify or can't keep staffed.
None of those work well. And none of the AI vendors are building specifically for this buyer.
Why small service businesses are underserved
It's worth understanding why this gap exists.
The buyer is hard to reach at scale. There's no list of "all the HVAC owners in America who are ready to buy software." You find them through local SEO, trade associations, agency partnerships, and referrals. That's slower and less defensible-looking than landing one big enterprise logo, so VCs steer their portfolio companies toward enterprise.
The product needs to be vertical-specific to actually work. A roofing call flow is different from a veterinary call flow is different from a plumbing call flow. Generic horizontal AI receptionists fail on the first real call because they don't know the difference between a leak emergency and a quote request, or between a vaccine appointment and an urgent triage.
The combination — distributed buyers, vertical product depth required — is exactly the kind of work that takes patience and operational discipline. Most companies aren't set up for it. We are.
Service businesses are price-tolerant when the math works
A roofer doing $14,000 storm-damage jobs treats an AI receptionist subscription as cheap if it captures one extra job a quarter. The math isn't close.
A veterinary clinic that loses one new client a week to a missed call is leaving roughly $1,500 to $4,000 of lifetime value on the floor every week. Same story.
The pricing question for these buyers isn't whether the monthly number is a lot of money. It's "does this thing pay for itself in the first 60 days." If it does, they buy. If it doesn't, no amount of discounting saves the deal. We design our offer around that fact instead of trying to compete on monthly subscription cost.
The moat is vertical depth
Generic AI voice is becoming a commodity. Bland AI, Retell AI, and VAPI all sell raw voice infrastructure cheap. Anyone can spin up a chatbot that picks up the phone in 2026.
The defensibility isn't the model. It's the workflow.
It's knowing that a roofer needs the call to capture insurance carrier and storm date in the first 90 seconds. It's knowing that a vet practice needs the AI to ask about the breed and weight before suggesting an appointment slot. It's the integrations into the field service software each vertical actually runs on. It's the post-call analysis tuned to flag the four or five specific signals each industry cares about.
That kind of depth doesn't come from a clever prompt. It comes from spending months with each vertical, breaking the system on real calls, and rebuilding it. That's the work. That's the moat.
Owner-level relationships beat enterprise procurement
There's one more reason we like this market.
With service businesses, you talk to the owner. They make the buy decision in the first conversation. There's no procurement department, no vendor security review, no 9-month sales cycle, no committee that requires three references and a SOC 2 audit before they'll consider a pilot.
A roofing company owner can decide on a monthly subscription in a 20-minute call. The same decision at a Fortune 500 takes 9 months and 14 people. We can run the small-business sales motion at speed and learn faster than the enterprise-focused competitors can.
The bet
We'd rather be the obvious choice for 1,000 roofers than option #6 in an enterprise RFP.
That's the bet. We think the math works, we think the moat is real, and we think the buyers are underserved by every category of vendor currently chasing them.
Interested in being a founding customer? Book a 20-minute demo and let's talk.