The trade press just blessed voice AI for HVAC. Here's the part it didn't price.

When a technology gets explained in ACHR News — the publication HVAC contractors have read for decades — it's no longer a startup pitch. It's trade consensus. Their recent piece, "Missed Calls, Lost Jobs," walks through why HVAC companies are adopting voice AI agents: how the technology works, how to deploy it, and the dos and don'ts. We build in this space, so we read it the way a buyer should. Here's what's solid, and the one question the article never asks.

What the article gets right

• Hold time is the silent killer. The article cites Call Centre Helper's finding that 50% of callers hang up after 45 seconds on hold, and 70% after 90 seconds. A two-minute hold on a 100-degree day isn't a service hiccup — it's a $2,000 install handed to the next company on Google. Answering in seconds isn't a luxury feature; it's the whole game.
• Start with one job. Their first deployment rule: pick a single use case. For most shops that's after-hours and overflow answering — the calls you're currently losing outright — not a rip-and-replace of your whole front desk.
• Try to break it before you commit. The best advice in the piece: call the thing yourself, throw it an angry customer, a weird address, a 3am emergency — and never sign a long contract before a pilot. Any vendor who won't let you test first has told you what you need to know.
• Don't pretend it's human. Their don'ts list is blunt: be upfront that callers are talking to an AI. In our experience that's right — customers care whether their problem got handled, not whether the voice was born.

The question it never asks: what does a booked job cost?

For all its deployment detail, the article never touches pricing models — and for a 1-to-5 truck shop, the model matters as much as the technology. There are three ways this gets sold: a flat monthly seat (you pay the same whether the calls were jobs or spam), per-minute bundles (the meter runs on wrong numbers and tire-kickers too), and per qualified lead (you pay only when a call becomes real work).

Before you compare vendors, run your own number: our missed-call cost calculator shows what unanswered calls cost your shop per month. That's the budget any of these models has to beat — and the yardstick for judging whether a pilot actually paid.

A buyer's checklist, distilled

• Write your rules down first. Hours, fees, what counts as an emergency, when a human takes over. The article calls this "your manual" — a page of bullets is enough, and a good vendor builds to it exactly.
• Test with curveballs, not the demo script. Interrupt it. Ramble. Ask for a price it shouldn't invent. Call at 3am.
• Review the recordings for the first two weeks. Every call, then spot-checks. If a vendor can't show you recordings and transcripts, walk away.
• Keep a human escape hatch. High-value or complex calls should warm-transfer to a person — without looping the customer back to the same busy line.
• Judge the pilot on booked jobs. Not on how natural the voice sounded in a demo — on work that landed on your calendar.

Where SpeedToLead lands on that advice

We built SpeedToLead for HVAC and the other trades around exactly this checklist: trade-specific triage (a no-cool call in a heat wave is not a routine ticket), warm transfer when a human should take it, recordings and transcripts with every lead, and a pilot that's actually a pilot — your first 20 qualified leads are free, no card, no contract. After that you pay per qualified lead, so a month of junk calls costs you nothing.

The trade press is right: this technology is mainstream now. The shops that win won't be the ones who bought first — they'll be the ones who priced it right and tested it hard. Start with the test.