Custom AI receptionists for HVAC companies

Start free pilotFree pilot. About 14 days to a working line, no call limit.

Why an HVAC company needs a custom AI receptionist

Because your busiest day arrives without notice and breaks the phone before it breaks the schedule. AnswerAI answers every caller during a cold snap, triages by who is genuinely at risk, and tells the rest the truth about when you can actually get there.

An HVAC company's demand curve is set by a thermometer, and that makes it unlike every other trade. The first genuinely cold morning of the year generates more calls in a few hours than a normal week produces, all of them urgent, all of them from people whose houses are getting colder, and your crews can serve the same number of jobs that day as any other. The problem on that morning is not answering the phone. It is what the phone promises.

What do people actually say when they call an HVAC company?

On an ordinary day, a mix of maintenance and quotes. On a spike day, one call repeated four hundred times with different amounts of urgency behind it, and sorting that urgency is the entire job.

  • “Furnace is out and it's minus twenty.”

    The defining call of the trade. Emergency, and emergency rates typically run well above standard labour; getting the classification wrong in either direction is expensive.

  • “It's blowing but the house isn't getting warm.”

    Sounds less urgent and frequently is not. A furnace running continuously without reaching temperature on a −30 day is a house that will be cold by evening, and it needs the same triage questions as a dead one.

  • “There's a baby in the house.”

    This changes the queue position and it is a rule, not a judgment. Who gets prioritised on a capacity-constrained day is a decision the business makes in advance: infants, elderly residents, medical equipment.

  • “We're on your maintenance plan - does that get us in faster?”

    Only you know the answer, and the line has to give the same one every time. A plan customer told yes on Tuesday and no on Wednesday is a cancelled plan.

Won't it book every caller into a day we physically cannot serve?

Only if nobody told it what capacity looks like. This is the specific failure that makes a booking-optimised agent dangerous for HVAC, and it is fixable, but it has to be fixed before the cold morning rather than during it.

Consider what a bundled agent is measured on. Booking rate is the metric a platform can show its own customers, so the agent's incentive on every call is to secure a slot. That works beautifully in a trade with elastic scheduling. On the first deep-freeze morning it produces a schedule with forty jobs in it, twelve of which can be served, and twenty-eight households who were told somebody was coming today.

The damage from that is worse than a missed call, and it is worth being precise about why. A caller who could not get through rings the next company and is mildly annoyed. A caller who was promised a technician stays home, keeps waiting, gets colder, and by evening is writing a review that names you. You have converted a lost lead into an actively hostile customer, and you paid for the privilege of doing it.

The alternative is a line that knows what capacity means for your business and says so. That is a small set of decisions: how many emergency slots exist in a day, what happens when they are gone, who gets priority when they are contested, and what an honest wait sounds like when the honest answer is Thursday. None of it is clever, and all of it has to be written down in advance.

The genuinely uncomfortable part is that an honest line loses some of those callers to a competitor who promised same-day. That is a real cost and we would rather state it than pretend otherwise. The trade is that the ones you do book, you serve, and the ones you turned away honestly will ring you first next time, which is a slower and considerably better business.

What does a spike day actually do to a schedule?

The arithmetic is not complicated and almost nobody does it before the morning it matters. Every input below is labelled: substitute your own and the conclusion holds in shape if not in size.

A model: swap in your own crew count and your own worst morning

  1. Technicians available on the day6Replace with your own crew count; this is the number the whole model turns on.
  2. Emergency calls a technician can complete in a long day5No-heat diagnostics and repairs in winter conditions, including travel between them. A business running planned maintenance alongside will be lower; use your own figure.
  3. Jobs the business can actually serve that day306 × 5. This number does not change when the temperature does, which is the entire point of the model.
  4. Calls arriving on your worst morningFrom your logThe one figure worth taking from your own records rather than from an industry average: pull the busiest morning of last winter. Every contractor we have asked already knows roughly what it was.
  5. Callers you cannot serve that dayarrivals − 30Everything above capacity. The only real decision is whether these people are told honestly in the first two minutes, or told a technician is coming and left to find out at nine that night.
  6. What AnswerAI does with all of themAnswers every oneTwenty simultaneous calls for the same monthly fee as one. The thirty who can be served get booked against your real capacity; the rest get an honest window and a callback record, instead of a busy tone or a promise nobody can keep.

The point of running this is that the capacity line does not move. Six technicians serve thirty jobs whether the phone rings forty times or four hundred, so on the coldest morning of the year the only variable you control is what the other callers are told, and that is a decision worth making in October rather than at seven on the morning it happens.

AnswerAI answers every caller on the coldest morning of the year and tells the ones you cannot reach the truth, because a customer promised a technician who never arrives costs a great deal more than the call you never took.

Nick Lovett, Founder, AnswerAI

The morning to find out whether your phone tells the truth about capacity is not the first −30 morning of the year.

Start free pilot

Questions HVAC contractors ask

Every caller is answered: the line handles up to twenty calls simultaneously, so nobody gets a busy tone on the day it matters most. What happens next is your rule set: triage by risk, prioritise whoever you have decided gets priority, and give everyone beyond capacity an honest window rather than a slot you cannot serve.

Yes, and it is one of the more valuable rules to set. No heat with an infant, an elderly resident or medical equipment in the house is a different call, and the line can ask and sort on it consistently. Consistency is the point: the same question asked the same way at 6am and at 6pm, rather than depending on who picked up.

It quotes whatever you tell it to quote, under the conditions you set. The classification questions that separate an emergency from a scheduled visit are written down before the line is built, and where a call genuinely does not fit them it escalates to a person instead of guessing.

If you say so. The important part is that the answer is the same every time: a plan customer told yes on Tuesday and no on Wednesday cancels the plan, and that inconsistency is far more likely from a stretched front desk on a spike morning than from a rule.

The bundled agents are configured from your CRM fields and optimised for booking rate, which is the wrong metric on a capacity-constrained day. If most of your calls are ordinary scheduling, that may be all you need. The case for a built line is specifically the two or three mornings a year when booking everybody is the worst thing your phone could do.

Let's build one before the cold does.

We build a working line for your triage rules, your daily capacity, your priority list and your emergency rates, and you call it yourself. About 14 days, no call limit during the pilot.

Start free pilot