Cactus Is the AI Receptionist That Plumbers Actually Need

The Macro: Missed Calls Are Killing Home Service Businesses

There is a number that should terrify every HVAC company, plumber, roofer, and electrician in America: roughly 60% of inbound calls to home service businesses go unanswered during peak hours. The owner is on a job site. The office manager is handling another customer. The after-hours voicemail picks up. The customer hangs up and calls the next company on the list. The job is gone.

This is not a problem that lacks solutions. It is a problem where every existing solution has serious drawbacks. Hiring a full-time receptionist costs $35,000 to $50,000 a year, plus benefits, plus the reality that one person cannot answer phones 24/7. Answering services like Ruby Receptionists and Smith.ai work, but they charge per minute or per call, and the operators are handling multiple clients simultaneously, which means they cannot deeply qualify leads or answer detailed questions about your specific services.

The AI voice agent category has exploded in the last year. Bland AI, Retell AI, Vapi, and a dozen others are building the infrastructure for AI phone calls. But these are platforms, not products. They give you the tools to build an AI phone agent. They do not give you an AI phone agent that already knows how to qualify a homeowner who needs their AC fixed.

That distinction matters in home services. The customer calling is often stressed, sometimes dealing with an emergency (burst pipe, no heat in January), and needs to talk to something that understands the context immediately. A generic AI voice agent that asks the wrong questions or stumbles over service-area boundaries will lose the call just as fast as a voicemail would.

The market is enormous. There are over 900,000 home service businesses in the United States. Most of them are small operations doing between $500,000 and $5 million in annual revenue. Most of them are terrible at answering their phones. And most of them know it.

The Micro: A Second-Time YC Founder Who Has Done This Before

Cactus was founded by Ajith Govind and Avinash Joshi. Govind is the CEO, and his background is the most relevant thing about this company. He is a second-time YC founder. His first company, Turing Labs, went through YC’s Winter 2020 batch and raised a Series A north of $20 million. At Turing Labs, he worked with major consumer packaged goods companies including Procter & Gamble, Unilever, Nestle, Kraft, Coca-Cola, and Mars. The man knows how to sell to businesses that are not tech companies.

Joshi is the CTO, a veteran Rails developer with experience shipping products across multiple languages and frameworks. They are based in the San Francisco Bay Area and came through Y Combinator’s Spring 2025 batch as a two-person team.

The product does exactly what you would hope. When a homeowner calls, the AI answers. It sounds human. It asks the right questions: what service do you need, what is the address (to check if it is in the service area), when do you need it done, what is your budget. It qualifies the lead and books the appointment directly into the business’s calendar. It follows up automatically if the customer does not confirm. It integrates with existing CRM systems so the business owner sees everything in one place.

The ROI pitch is aggressive but credible. Cactus claims a $500 per month operational cost, and their ROI calculator shows a potential 56x return in the first month. Those numbers assume you are missing calls that would have converted to jobs. For a plumbing company where the average job is $300 to $800, capturing even two or three extra jobs per week more than covers the cost.

The target customer is clear: home service businesses doing $1 million to $20 million in annual revenue, receiving 20 to 40 or more inbound calls per day. That is the sweet spot where call volume is high enough to justify automation but the business is not large enough to staff a full call center.

What I like about the vertical focus is that it constrains the problem. Cactus does not need to handle every type of phone call. It needs to handle calls from homeowners who want HVAC repair, plumbing, roofing, landscaping, or electrical work. That is a finite set of conversation patterns, and getting really good at those specific patterns is more valuable than being mediocre at everything.

The Verdict

I think Cactus is one of the clearest product-market fit stories I have seen in the AI voice space. The pain is obvious (missed calls equal lost revenue), the buyer is identifiable (home service business owners), the ROI is easy to calculate, and the founder has already built and scaled a B2B company.

The competitive risk comes from two directions. From below, the AI voice platforms (Bland, Retell, Vapi) could enable competitors to spin up similar products quickly. From above, the answering services (Ruby, Smith.ai) could integrate AI and defend their existing customer base. The moat for Cactus is domain specificity: knowing home services deeply enough that the AI handles edge cases a generic solution would botch.

Thirty days, I would want to see churn numbers. Are businesses trying Cactus for a month and leaving, or are they sticking? Sixty days, whether the product handles emergency calls well enough that owners trust it with their most valuable leads. Ninety days, the question is expansion. Can Cactus move into adjacent verticals like dental offices, auto repair shops, or property management companies without losing the depth that makes it work for home services? Govind has scaled a company before. He knows the playbook. The question is whether this market is as ready for AI as the metrics suggest, or whether small business owners will resist handing their phones to a robot longer than the optimists think.