AI Receptionist for Home Services: What It Books, What It Costs, Where It Fails
The short answer
An AI receptionist for home services answers every call your shop can’t — 24/7 — qualifies the job, books it into your schedule, and escalates genuine emergencies to the on-call technician. For a 2–6 truck HVAC, plumbing or electrical shop it typically costs $49–$449 a month, and it pays for itself if it books one or two extra jobs a month. It fails on gas leaks, complex diagnostics and angry customers — those still need a human, fast.
If you run a home services business, your phone is your storefront. More than three-quarters of home service bookings still start with a phone call, and for emergency-driven trades like plumbing and HVAC that figure climbs to 85–90% (ServiceTitan, 2024). The problem is that the phone rings hardest exactly when nobody can answer it: on a roof, under a sink, up a ladder, or at 9pm on a Saturday when a furnace dies. This post is the honest version of what an AI receptionist does for a home services shop — what it books, what it costs, and the places it will let you down.
The missed-call problem, with honest numbers
Start with the number that matters most. Invoca’s Home Services Lead Conversion Benchmarks Report 2026 — built on more than 70 million calls and 600 million minutes of conversation across nine home-services trades — found that 52% of inbound calls to home services businesses are answered by a person. The trade-level split is brutal: HVAC answers just 34% of calls, construction 32%, while plumbing answers 74%. In other words, the average HVAC shop misses two out of every three calls.
That is the defensible number. You will also see a much scarier one quoted everywhere: “62% of calls go unanswered and 85% of callers never call back.” Be careful with it. The 62% figure traces back to a 2016 study by 411 Locals, a local SEO agency, covering 85 businesses across 58 industries over 30 days — not home services specifically, and a decade old. The 85% figure has the same problem: it is usually cited without a source at all. We mention both because your competitors’ sales pages quote them; we don’t build our math on them.
What is solid, and worse in some ways than the folklore:
- ServiceTitan’s platform data (3,000+ trade businesses, 2024): the typical shop books just 42% of its inbound calls into actual jobs. HVAC books 38%, plumbing 43%, electrical 41%.
- Weekends are the hole in the bucket: 41% of home service calls on weekends go completely unanswered, even though weekend emergency volume runs at 60–80% of weekday volume (ServiceTitan, 2024).
- Callers don’t wait around: CallRail’s September 2025 survey of 1,000 US consumers found 82% will call a competitor when nobody answers. Hiya’s data puts it more starkly: roughly 80% of callers who reach voicemail leave no message at all.
- Speed decides who gets the job: the MIT Lead Response Management study (Harvard Business Review) found leads contacted within 5 minutes are 21 times more likely to be qualified than leads contacted after 30 minutes. ServiceTitan’s 2025 benchmark puts trade numbers on it: shops responding in about 2 minutes convert 62% of leads, against 28% at the 42-minute industry average.
Put those together and the picture is simple: roughly half your calls never reach a person, the ones that don’t mostly never call back, and the ones you call back hours later have usually already booked someone else. That is the problem an AI voice agent is actually solving. It is not a gadget. It is a second front desk that never goes on a job site.
What an AI receptionist for home services actually books
A good AI receptionist does three things on every call: it qualifies the job, it books the job, and it routes the exception. The qualification is trade-specific, and that is where a generic phone bot falls apart. Here is what it should handle, trade by trade:
| Trade | Routine bookings it handles well | Emergencies it must escalate, not book |
|---|---|---|
| HVAC | Annual maintenance, filter changes, thermostat installs, seasonal tune-ups, quotes for system replacement | No heat in freezing weather, gas smell, carbon monoxide alarm, water leaking from the unit |
| Plumbing | Drain cleaning, fixture installs, leak repairs, water heater service, remodel quotes | Burst pipe, sewage backup, no water, gas leak |
| Electrical | Panel upgrades, outlet installs, EV charger quotes, lighting work, safety inspections | Burning smell, sparking panel, power out to the whole house, exposed live wires |
| Roofing | Inspections, gutter cleaning, repair quotes, storm-damage assessments | Active leak into living space during a storm, tree on the roof |
| Pest control | Quarterly treatments, one-off sprays, inspections, termite quotes | Live wasp nest inside the house, bed bug confirmation (needs a human to set expectations) |
| Cleaning / junk removal | Recurring cleans, deep cleans, move-out cleans, haul-away quotes | Hoarding situations, biohazard — needs a human quote, never a flat rate |
Notice the pattern. The AI is excellent at the 80% of calls that are routine: name, address, what broke, when it started, how urgent it feels, and a slot on the calendar. It is also excellent at the 10% that are emergencies — not by handling them, but by recognising the keywords and getting a human on the line in seconds instead of letting the caller sit in a voicemail queue. The remaining 10% — the weird, the angry, the complicated — is where it hands off. Any vendor that tells you the AI handles all three categories is selling you a demo, not a product. The same pattern holds in other verticals — see how an AI receptionist for dental practices handles bookings, no-shows and reminders.
A realistic call flow looks like this. The phone rings twice. The AI answers in your company’s name, in a natural voice. It asks what the caller needs and listens for the emergency words first — gas, flooding, no heat, sparking, sewage. If it hears one, it says so plainly, offers to connect the on-call technician immediately, and texts the tech the address and the caller’s words at the same time. If it is routine, it asks the four qualifying questions (what, where, when, how urgent), checks the calendar, offers the next two slots, books one, and texts a confirmation with the technician’s arrival window. After the call it logs everything into your field service software — Jobber, ServiceTitan or Housecall Pro — so the dispatcher sees a booked job, not a voicemail to transcribe. The whole thing takes two to three minutes. A human doing the same call takes the same time; the difference is the AI does it at 11pm on a Sunday in July, when your phone would otherwise ring out.
What it costs in 2026
Published prices for AI receptionists built for — or commonly used by — home services shops run from about $29 a month to about $449 a month at entry level, and the honest range for a 2–6 truck shop is $50–$250 a month. Prices below were read from the vendors’ own pricing pages in September–November 2026, and vendors change them, so recheck before you buy. If you want the general pricing picture first, our AI receptionist cost guide breaks down flat-rate vs per-minute pricing across industries.
| Service | Entry price | What you get | Best for |
|---|---|---|---|
| Rosie | $49/mo | 250 minutes included | Cheapest after-hours cover; Spanish on every tier |
| Goodcall | $79/mo | 100 unique callers, then ~$0.50–$0.79 each | Shops with lots of repeat callers; billed per caller, not per minute |
| My AI Front Desk | $79/mo billed annually | 200 minutes included | Voice, text, chat and email on one flat plan |
| Smith.ai (AI tier) | $150/mo | 75 calls, then ~$2.17–$2.50 per call | Field-service integrations with live receptionists as backup |
| NextPhone | $199/mo | Unlimited calls | High-volume shops; the flat rate removes peak-season bill shock |
| Sameday AI | $449/mo | 500 minutes, ServiceTitan scheduling | ServiceTitan shops that want a trade specialist |
| Jobber Receptionist | $29/mo add-on | Built into Jobber | Jobber users — try the built-in option before buying a standalone tool |
Three pricing models dominate, and the model matters more than the headline price:
- Per minute ($49–$449/mo for 250–500 minutes): Rosie, Sameday AI, My AI Front Desk. Cheap in quiet months, painful in July when every air conditioner in the county fails at once. A 300-call peak month at 3 minutes a call is 900 minutes — check what the overage rate does to your bill before you sign.
- Per caller ($79–$249/mo): Goodcall. You pay for unique phone numbers, not minutes, so a long call from a repeat customer costs the same as a short one. Good for shops with a loyal base; bad if your marketing drives a flood of one-off callers.
- Flat unlimited ($199/mo): NextPhone. The simplest to budget. In a 300-call peak month it is often the cheapest of the three models, and it is the only one where a heatwave cannot surprise you with a bill.
Compare that with the alternatives. A live answering service (Smith.ai’s human tier) starts at $300 a month for 30 calls and runs $8.50–$11.50 per extra call — a 300-call month costs over $3,000. A part-time in-house dispatcher costs $18–$25 an hour before payroll taxes, and still doesn’t cover nights and weekends. The AI is not replacing your dispatcher; it is covering the hours and the volume your dispatcher can’t.
One warning before you buy: check whether your field service software already includes one. Jobber, Housecall Pro and ServiceTitan all ship an AI receptionist now. Paying for a standalone tool on top of a feature you already own is the most common way shops overspend on this category.
The ROI math for a 3-truck HVAC shop
Let’s run the numbers with the defensible figures above and no optimism. Take a 3-truck HVAC shop getting 200 inbound calls a month. At the Invoca 2026 HVAC answer rate of 34%, a human answers 68 of those calls and 132 go unanswered. Of the unanswered, most never leave a message and most never call back — so treat them as lost unless you chase them.
Now add an AI receptionist that answers every call. It won’t book all 132. Use ServiceTitan’s 38% HVAC booking rate for the calls it handles: 132 × 38% ≈ 50 booked jobs. Not all of those are incremental — some callers would have called back or booked online anyway. Be conservative and say a third of them are genuinely recovered: about 17 extra jobs a month.
CallRail’s September 2025 consumer survey puts the average HVAC repair ticket at $350 (range $100–$3,000). Seventeen jobs × $350 = $5,950 a month in recovered revenue. System replacements, which the AI also books as quote appointments, run into the thousands and are not in that figure at all. Against a $79–$199 monthly subscription, the math is not close — even if you halve the recovery rate, the tool pays for itself many times over.
The honest version of this math: it only works if the AI actually books into your calendar and your dispatchers actually run the jobs. An AI that takes messages and emails them to an inbox nobody reads recovers nothing. The booking integration — not the voice quality — is where the money is. Ask every vendor to show you a booked job in your software, on a live call, before you pay.
Where an AI receptionist for home services fails
This is the section most vendor pages skip. We build these systems for a living, and here is where they break:
1. Emergencies that need judgment, not a script. A gas smell, a carbon monoxide alarm, a burst pipe flooding a kitchen — the AI’s job is to recognise the words and escalate in seconds, not to diagnose. Any system that tries to talk a caller through a gas leak instead of getting a human on the line is a liability. Test this specifically: call the demo number, say “I smell gas,” and time how long it takes to reach a person.
2. Complex diagnostics. “My furnace makes a clicking noise, then the blower starts, then it stops, and there’s a smell like burning dust” is a real call. A human dispatcher with ten years in the trade knows roughly what that is and which tech to send. An AI will book a generic diagnostic visit — which is fine — but it cannot triage the job the way your best dispatcher can. Don’t expect it to.
3. Angry customers. A customer whose basement flooded twice because your tech missed a fitting does not want to talk to a voice agent, however polite. The AI should detect frustration and transfer to a human within seconds, not try to soothe the caller through three more questions. If a vendor can’t show you the escalation trigger, walk away.
4. It can’t dispatch crews directly. Most AI receptionists book appointments; they do not re-route a technician who is already on a job, negotiate a parts run, or decide which truck is closest. That is dispatcher work, and it stays dispatcher work. The AI feeds the dispatcher clean, booked jobs. It does not replace the dispatcher.
5. Language and accent gaps. Spanish support exists on some plans (Rosie includes it on every tier; others charge extra or don’t offer it) and matters enormously in many US markets. But even the best systems struggle with heavy accents, poor cell connections and callers talking over the agent. If a large share of your customers call from a job site on speakerphone, test with real calls, not the vendor’s polished demo.
6. Integration is where cheap tools die. An AI that books into Google Calendar but not into Jobber or ServiceTitan creates double-entry work for your office — the exact admin burden you were trying to remove. Native integration with your field service software is a requirement, not a nice-to-have. Zapier workarounds count only if someone on your team actually maintains them.
7. It can’t quote accurately. “How much to replace a water heater?” depends on the unit, the venting, the code requirements and the access. An AI that quotes a flat price will either scare the customer off or commit you to a loss-making job. The right answer is a range with a clear “final price after the tech sees it” — and the AI must be trained to say exactly that, every time.
AI receptionist vs live answering service vs hiring
| AI receptionist | Live answering service | In-house dispatcher | |
|---|---|---|---|
| Monthly cost | $49–$449 | $300–$3,000+ at volume | $2,800–$4,000+ (part-time, before taxes) |
| Coverage | 24/7, every call | 24/7, per-call charges | Business hours only |
| Books into your software | Yes, if integrated | Usually messages, not bookings | Yes |
| Emergency judgment | Keyword escalation only | Good — a human decides | Best — knows your techs |
| Angry customers | Escalates (if configured) | Handles well | Handles best |
| Peak-season bill | Flat plans: unchanged; per-minute: spikes | Spikes hard | Overtime |
The right answer for most shops is a combination, not a single choice: AI for the first ring and the routine bookings, a live service or your own dispatcher as the escalation path for emergencies and angry customers. The AI is the net; the human is the hands that pull the fish in. Shops that try to run on AI alone discover the failure modes above the expensive way.
The 7-question checklist before you buy
Ask every vendor these seven questions, in this order. A vendor that dodges any of them is telling you something.
- Show me a booked job in my software. Not a screenshot — a live call that ends with a job in Jobber, ServiceTitan or Housecall Pro. If it books into Google Calendar only, it is not a field-service tool.
- What happens when a caller says “I smell gas”? Time the escalation. It should be seconds, with a simultaneous text to the on-call tech. If the AI tries to book a routine visit, stop the demo.
- What does my worst month cost? Price a 300-call peak month, not the quiet month on the pricing page. Ask for the overage rate in writing.
- How does it handle Spanish-speaking callers? If a meaningful share of your customers speak Spanish, this is a deal-breaker question, not a feature checkbox.
- What does it do with an angry caller? There must be a frustration trigger that transfers to a human fast. Ask to hear it.
- Who trains it on my business? Your service area, your brands, your pricing ranges, your emergency definitions, your on-call rota. A generic script books generic jobs badly. Ask how long setup takes and who does it.
- Can I listen to the calls? Every call should be recorded and transcribed, with a dashboard you actually check. If you can’t audit it, you can’t improve it — and you won’t know it’s failing until a customer tells you.
How to roll it out without annoying your customers
Don’t switch it on for every call on day one. Start with after-hours and overflow only — the calls your team already misses — and let it run for two weeks. Listen to the recordings. Fix the three things that sound wrong: the greeting, the emergency words it missed, the question it asks twice. Then expand it to lunchtime cover, then to full overflow during peak hours. Tell your regulars: a line on your website and a note in the confirmation text (“we answer 24/7 now — yes, even at 2am”) turns a robot answering the phone from a surprise into a selling point.
Set the expectations honestly. The AI should introduce itself as your company’s assistant, not pretend to be a person — customers forgive a machine that is upfront and get angry at one that deceives them. And keep a human escape hatch on every call: “press 0 or say ‘speak to someone’ at any time” is not a weakness in the system. It is the system working. If you’re weighing which automation to buy first, our roundup of 9 AI agents a small business can actually use puts the receptionist alongside lead follow-up, invoicing and booking agents.
Frequently asked questions
How much does an AI receptionist cost for a home services business?
Published entry prices run from $29 a month (Jobber’s built-in Receptionist add-on) to $449 a month (Sameday AI’s trade-specific plans), based on vendor pricing pages checked in September–November 2026. Most 2–6 truck shops land between $50 and $250 a month. The number that matters is your peak month: a 300-call July can cost $129 on a flat plan or several hundred on a per-minute plan with overage charges. Price the peak month before you sign.
Can an AI receptionist book jobs in Jobber or ServiceTitan?
The good ones can — natively, not through a workaround. Sameday AI and Avoca schedule directly into ServiceTitan; Smith.ai integrates natively with Jobber, ServiceTitan and Housecall Pro; Jobber’s own Receptionist is built in. If a vendor only books into Google Calendar and promises a Zapier fix, treat the integration as unproven until you see a live booked job in your software.
Will customers accept talking to an AI instead of a person?
Mostly, yes — if the AI is fast, accurate and honest about what it is. Customers call a home services shop because something is broken, not because they want a chat. What they won’t accept is being trapped: no way to reach a human, a loop of repeated questions, or a bot that pretends to be a person. An AI that answers on the second ring, books the job and texts a confirmation beats a voicemail every time. A voicemail, remember, is what roughly 80% of callers won’t even leave a message on.
Can an AI receptionist handle emergency calls?
It can recognise them and escalate them — it should not try to handle them. The right design is keyword detection (gas, flooding, no heat, sparking, sewage) that immediately connects the on-call technician and texts them the details. Test this on every vendor demo: say “I smell gas” and time how long it takes to reach a person. Seconds is the only acceptable answer.
Is an AI receptionist better than a live answering service for HVAC and plumbing?
For routine bookings and after-hours cover, yes — it answers every call instantly, books directly into your schedule, and costs a fraction of a live service at volume. For emergencies that need judgment and angry customers, a live service is better. Most shops end up with both: AI on the first ring, a human as the escalation path. A live service alone at 300 calls a month can cost over $3,000; the AI at the same volume costs $129–$789 depending on the plan.
Do I still need a dispatcher if I have an AI receptionist?
Yes. The AI books jobs and takes messages; it does not re-route technicians, negotiate parts runs, or decide which truck is closest. Your dispatcher stops spending the day transcribing voicemails and answering “what time will the tech arrive?” and starts spending it on the scheduling judgment only a human can do. Think of the AI as giving your dispatcher their day back, not as replacing them.
The bottom line
An AI receptionist for home services is not a futuristic purchase. It is a fix for a specific, measurable leak: roughly half of your inbound calls never reach a person, and the callers who don’t reach you mostly book your competitor. For $50–$250 a month, a good system answers every call, books the routine jobs, escalates the emergencies in seconds, and hands your dispatcher a clean schedule instead of a voicemail inbox. Buy it for the booking integration, test it on the words “I smell gas,” price your worst month before you sign, and keep a human one button-press away. Do that, and the phone stops being the leak in your business.
If you want a system like this built for your own shop — tuned to your trades, your service area and your field service software — talk to us. We build AI voice agents and receptionists for home services businesses, and we’ll tell you honestly if your call volume doesn’t justify one yet.