The test
Trades businesses miss calls. After 6 PM, during a job, on a Sunday — every missed call is a job that goes to whoever picks up next. The pitch behind every AI receptionist product is that software can catch those calls, qualify the caller, and book the work. We're going to find out whether that's true for a real shop, not a demo.
One Detroit-area plumbing company. One AI receptionist, configured by us, answering only the calls the humans don't. Fourteen days. We run a baseline week first so the comparison is honest.
What we measure
After-hours and overflow calls answered. Callers qualified versus lost. Jobs booked that would otherwise have gone to voicemail, priced at the shop's own average ticket. Cost to run the system, all-in. False positives — the robocallers, the wrong numbers, the callers who hung up on the robot.
The verdict rubric is fixed before we start: BUILT DIFFERENT means the system paid for itself at least three times over and the owner chooses to keep it running. WAIT means it works with caveats we'll spell out. VAPORWARE means it didn't survive contact with a real front desk.
The deal for the business
The pilot business pays nothing. We build, deploy, and babysit the system for the full run, and they keep whatever it books. In exchange, we publish the numbers — the business stays anonymous by default ('a Detroit plumbing co.'), named only if the owner wants the press.
Run a trades or service business in metro Detroit and want in? Email us. Serious inquiries get a call back within a day.
Publishing commitment
This page ships with the full profit-and-loss table, anonymized call log excerpts, the system configuration we deployed, and a straight verdict. If the thing flops, that's the story. Field Reports exist because every AI tool demos well; the only interesting question is what happens in week two at a business that doesn't care about AI.