Fixseer Early prototype

See the job before the truck rolls.

Before the visit, your customer sends a photo of the unit and a short voice note. Fixseer identifies the equipment, sorts the problem, and gives your technician a one-page brief, so the job gets finished on the first visit.

Email us

For small residential HVAC companies. You keep your customers, your brand and your data.

Example brief · illustration

No cooling, upstairs unit

Needs a technician
Unit
Outdoor condenser, identified from the customer's photo
Customer said
"Fan runs, air isn't cold, started yesterday"
Check first
Likely causes, ranked, with what to bring
After visit
Brief right or wrong? Cause was…

How it works

Being built now. What exists today is listed further down.

  1. Your link, your logo

    Your customer opens a link from your company and sends up to three photos of the unit and a 20-second voice note.

  2. Sorted before anyone drives

    The AI identifies the equipment and sorts the job. Fixed safety rules override the model.

    Safe to fix yourselfNeeds a technicianDo not touchEmergency

  3. The technician gets the brief

    Unit, what the customer said, and what to check first. After the visit, two taps: was the brief right, and what was the cause. That answer is how it gets better for your company.

Who it is for

  • Small residential HVAC companies, roughly three to fifteen technicians, on Housecall Pro or Jobber.
  • HVAC first. Plumbing and electrical use the same link later.
  • Not a marketplace. We never sell your customers as leads.

What exists today

  • The first prototype, HVAC Copilot, built by Rayhan Patel at the YC × Cactus × Google DeepMind Voice Agents Hackathon in April 2026, placed in the top 10 of 200+ builders. Watch the pitch.
  • It identifies a unit from a phone camera and produces a structured job record. Both videos are below. Code: github.com/Rayhanpatel/Ycombinator-Cactus-Deepmind (MIT).
  • Next: first trials in Maryland. If you run a small HVAC company or a weatherization program, we would like to show you.

See it

Two videos from April 2026. What they show is unit recognition and the original pitch. The intake-first version described above is what we are building now.

60-second demo. The prototype recognises the unit from a phone camera and starts the job record. Open on YouTube.

Two-minute pitch on stage at YC. Where this started: the hackathon team's original framing was an assistant for the technician on site. Open on YouTube.

What we want to build

For anyone who wants to understand the whole goal, not just the first step.

The problem we care about. Small service companies lose money in the gap between a customer who cannot describe the problem and a technician who has to drive out to see it. Wasted trips, second visits, and customers who feel nobody listened.

Step one, HVAC. A link the company gives its customers. Photo and voice note in, one-page brief out, technician arrives ready. We will measure one number with the first companies: how often the job is finished on the first visit, and whether the technician says the brief was right.

Step two, the record. Every job pairs what the customer said and showed with what the technician found. No field-service software keeps that pairing today. It is what makes the next brief better, and what tells a company which parts to keep on the truck.

Step three, the same companies, more trades. Most small HVAC companies also do plumbing and electrical. Same link, same customers, second trade. Then appliances, then the other trades where a customer cannot describe the problem and a technician has to drive out to see it.

The technician's side. Today the brief arrives on a phone. Later it talks: hands-free in the ear, working in a basement or on a roof with no signal, because in the US there is no junior standing next to the technician handing over tools and answers. Glasses when they are ready for a job site, not before.

Parts, through partners. Once the record knows which units are out there and what fails on them, it can tell a company what to keep on the truck, and tell the distributors and manufacturers who already sell the parts what is failing where. We do not want to be the warehouse.

Every size of company. Small shops first, because they buy fast and nobody serves them. Then the larger companies and the groups that own many of them, who need the same thing across many technicians and many brands.

The long goal. Every service business gets its own expert, one that talks to their customers before the visit and to their technicians during it. The business sells it to its customers as part of the service, under its own name, and both sides trust the visit more. The way Shopify let a small shop look and work like a big one, without becoming the shop.

What we are not building. Not a marketplace. Not another scheduling system. Not a homeowner app. The company keeps its customers, its brand and its data.

Links