Trades and clinics

AI receptionist for trades and small clinics: when it pays, and when to keep the human

By Babar Lal  ·   ·  11 min read

It is 10:40 on a Thursday. You are on your back under a kitchen sink with both hands on a compression fitting. Your phone buzzes in your pocket. You cannot answer it. Twenty minutes later you wipe your hands and call back. The woman with the leaking water heater says thanks, but she already booked someone. She found them on the next line of the search results.

That call was not lost because you were slow. She had a problem now and you were busy now. Dental offices have the same story. The phone rings while the front desk is checking in a patient.

This post is about whether an AI receptionist fixes that. Sometimes it does. Sometimes a cheaper tool will do. Every number below links to a source you can open, and there is a sum you can redo with your own figures.

Two outcomes for the same call to a plumber. On the left, a missed call at 10:40am, no voicemail, and the customer has already booked someone else by the 11:00 callback. On the right, an AI receptionist answers at 10:40am, takes the details and books the job for 2pm.

The short answer

An AI receptionist answers your line when nobody else can. It talks in a natural voice, answers common questions, books the job into your calendar, and texts you a summary.

It does not replace your judgement. It should not quote a complex job, calm an angry customer, or manage an emergency on its own. It suits busy owners who miss calls while working, and offices whose front desk cannot answer every ring.

QuestionShort answer
What it does wellAnswers every call, collects name, address and problem, books into open slots, answers routine questions, texts you a summary
What it should hand to a personEmergencies, angry callers, complex quotes, insurance and treatment questions
Who it suitsOwner-operators on job sites, small crews, dental and medical offices with a busy front desk, anyone with lots of after-hours calls
Who should skip itOffices with a person already answering fewer than about 20 calls a day, and businesses whose callers are happy with a text back
What it costsRoughly $1.70 to $3 per call on published AI plans, against about $7 to $11.50 per call for a human answering service

If you want to hear one before reading further, our live AI receptionist demo lets you call it yourself.

The real reasons calls get missed

The first reason is simple volume. CallRail looked across its customers and reported in July 2026 that 28% of business calls go unanswered on average. Invoca found a similar figure for home services. In its 2024 analysis, 27% of calls to home services businesses were not answered. That is more than one call in four.

The second reason is the job itself. Nobody answers from a ladder. Invoca's 2026 home services benchmarks, built on more than 70 million calls, found that only 52% of callers to home services businesses speak with a person.

The third reason is timing. ServiceTitan's analysis of residential HVAC calls in 2025 found that 14.1% of inbound calls in June came outside 8am to 5pm on weekdays. The busiest off-hours window was 5pm to 9pm.

The fourth reason is voicemail. It is not a safety net. Invoca found that fewer than 3% of callers who reach it leave a message.

Clinics see the same pattern. Patient Prism analysed 11.5 million calls to 8,280 dental locations for its July 2026 report. Of every 100 patient calls, 31 hung up before reaching anyone. Only 21 booked. The report says that 70% of the time the lost booking is not the front desk's fault.

What an AI receptionist handles well, and where it needs a human

An AI receptionist is good at the calls that follow a pattern. Most calls to a trades business or a clinic do.

It takes new job requests: what is wrong, where, and when the caller is free. It books into real open slots if it can see your calendar. It answers routine questions about hours, service areas and insurance. It works after hours, because it never goes home. And every call ends with a written summary on your phone.

Emergencies need a person. A burst pipe, a house with no heat in January, a smell of gas, a patient with facial swelling. The AI should spot these words and ring the on-call person at once, or send an urgent text. For a gas smell it should tell the caller to leave the building and call the gas company or 911. It should never try to talk someone through the fix.

Angry callers need a person who can make it right. The AI can take the details and promise a call back within the hour. Complex quotes need a person too. For a remodel, a rewire or a treatment plan, the AI books the visit. It should not give a price.

Clinics have one more limit. Insurance and treatment questions often need the practice system and a trained person. Mike Derlatka, writing on Dentaltown, makes the same point from the vendor side. He says the setups that work today are after-hours answering, overflow and basic intake.

Here is what that looks like in a real build. For AHM Services, a plumbing business in France, we built a voice agent that picks up every call straight away. It takes the caller's name, number, address, problem, urgency and preferred time. It logs each lead in a sheet and pushes the contact into HubSpot. Emergencies trigger an instant WhatsApp alert, and the plumber gets the full job details either way. In a typical month it answers about 70 calls, and about 10 of them turn into booked jobs. It runs on Retell and ElevenLabs for the voice, n8n behind the scenes and Twilio for the alerts. You can read how the plumbing voice receptionist was built, with screenshots, or browse our other case studies.

What owners and the industry are saying

I only list pages I could open. Trade forum threads on Reddit and ContractorTalk would not open for me, so they are not here. If you have one that changed your mind, send it and I will add it.

WhereWhoWhat they found or saidWhat it tells you
Invoca blog, 2024Call analytics vendor27% of home services calls unanswered; under 3% of voicemail callers leave a messageVoicemail is not a safety net
Invoca benchmarks, 2026Call analytics vendor, 70M+ calls52% of callers speak with a person; 55% of businesses do not ask the caller to bookAnswering is only half of it; asking for the booking matters
ServiceTitan, 2026Field service software14.1% of June HVAC calls came after hours; peak 5pm to 9pmSeasonal trades need evening cover most
Patient Prism, 2026Dental call analytics, 11.5M calls31 of 100 callers hang up before reaching anyone; 21 of 100 bookClinics lose patients at the phone, not the chair
Weave TrueLark pageDental software vendorClaims 1 in 3 calls are missed and 80% of missed calls are booking requests (no source given)Vendor numbers point the same way, but check the source
Dentaltown, April 2025Mike Derlatka, AI vendorWorks today for after-hours, overflow and basic intake; needs real practice system integration and proper HIPAA setupStart narrow, connect it properly
RevUp Dental, April 2025Nick Fotache, dental marketer and criticCalls AI receptionists a grift; says patients got frustrated repeating names and birthdatesA badly built agent makes things worse, not better

Vendors and critics agree that calls are being lost. They disagree on whether AI fixes it. My reading: a narrow, well-connected agent helps. One that tries to replace the whole front desk frustrates people.

AI receptionist vs answering service vs voicemail

Prices come from the providers' own pages in October 2026. They change, so check before you buy.

OptionTypical costGood atWeak at
VoicemailFreeNothing costs lessUnder 3% of callers leave a message
Missed call text backLow monthly feeCatches callers who will reply to a textCallers who want an answer now
AI receptionist planSmith.ai AI: $150 a month for 75 calls, about $2 a callEvery call answered, 24/7, books into a calendarEmergencies, angry callers, complex quotes
Human answering serviceSmith.ai human: $300 a month for 30 calls; Ruby: $395 a month for 100 minutesEmpathy, judgement, unusual callsCost at volume, limited access to your systems
In-house receptionistMedian $18.27 an hour, about $38,010 a year (BLS, May 2025), before taxes and benefitsKnows your customers, handles everythingOne person, one call at a time, goes home at 5

If you are not sure a text back is enough, our guide to missed call text back for small businesses covers when it does the job on its own, what it costs, and the texting rules.

A cost sum you can redo with your own numbers

Here is the arithmetic we use on audit calls. The example is a small plumbing business. Swap in your own numbers.

LineHow to get itExample
Missed calls per weekCount them from your phone log for one real week15
Share that are new jobsListen to a sample, or ask callers who ring back. Some are suppliers or spam40%, so 6 calls
Booking rate when answeredJobs booked divided by new-job calls you did answer45%, so about 2.7 jobs
Average job valueRevenue divided by jobs, last three months$350
Revenue missed per weekJobs x job value2.7 x $350 = $945
Revenue missed per monthWeekly figure x 4.3About $4,060

The 45% booking rate comes from Invoca's 2026 benchmark for home services leads. Use your own if you know it.

Now compare that with the cost. An AI plan at about $2 a call, on 60 calls a month, is about $120. A built agent connected to your calendar costs more. Even if it catches only half of those missed jobs, the example business gets back about $2,000 a month. That number decides whether this is worth doing.

When it is not worth it

We use one rule from our homepage. If your phone rings fewer than about 20 times a day and a person already answers it, keep the person. At that volume a good receptionist costs little per call, knows your regulars, and handles the odd ones better than any software. An AI receptionist only beats that person on cover, meaning nights, weekends and the moments they are already on another line.

Skip it too if your callers are mostly regulars happy with a text back, or if you miss two or three calls a week. And if nobody will read the call summaries, the agent will book jobs nobody turns up to. That is worse than a missed call. For a clinic, if most lost calls are about insurance, fix the front desk workflow first.

What we build, and when we say no

We build voice agents for trades businesses and small clinics. The agent answers in your business name, books into the calendar you already use, flags emergencies to the on-call person, and sends you a summary after every call. The details are on our AI receptionist for trades page, and there is a version set up for plumbing businesses and one for healthcare practices and clinics.

For bookings, the agent uses the same setup we describe in our post on automatic appointment booking for service businesses. If you are new to all of this, our 24 questions owners ask about AI automation puts phone answering next to the other jobs AI can do for a service business.

We read every call summary for the first month before anyone trusts the agent. And we say no often. Under about 20 calls a day with a person answering, we tell you to keep them. If a text back covers it, we tell you to buy that. If the sum is smaller than the cost, we stop there.

The fastest way to judge it is to hear it. Call our live AI receptionist demo and try to trip it up.

Questions people ask

Is an AI receptionist worth it for a small business? It is worth it when you miss calls that would have become paid work, and nobody is free to answer them. Count your missed calls for one week, multiply by the share that are new jobs, your booking rate and your average job value. If that number is bigger than the monthly cost, it pays. If you already have a person answering fewer than about 20 calls a day, keep the person.

Can an AI phone answering service handle emergency calls for plumbers and HVAC? It can recognise an emergency and route it, but it should not handle it alone. A good setup listens for words like burst, flooding, no heat or gas smell, then rings the on-call tech straight away or sends an urgent text. For a gas smell it should tell the caller to leave the building and call the gas company or 911 first.

Will callers know they are talking to an AI? Some will and some will not. Modern voices sound natural, but we still set the agent to say it is an automated assistant if asked, and to offer a person. Being honest costs very little. Getting caught pretending costs trust.

How much does an AI receptionist cost compared with an answering service? Published prices in October 2026 put AI receptionist plans at roughly $1.70 to $3 per call, for example Smith.ai's AI plan at $150 a month for 75 calls. Human answering services run about $7 to $11.50 per call on Smith.ai's human plans, or $395 a month for 100 minutes on Ruby. A built AI receptionist connected to your own calendar and job system costs a setup fee plus a monthly fee.

Can an AI receptionist book appointments into my calendar? Yes, if it is connected to the calendar or job system you already use. It offers the slots you mark as open, books the one the caller picks, and sends a confirmation. For dental offices the hard part is insurance and treatment questions, which usually still need the front desk.

Is a missed call text back enough instead of an AI receptionist? For many small trades businesses, yes. If most of your missed calls come from people who will happily reply to a text, an instant text back catches them for far less money. An AI receptionist earns its extra cost when callers want an answer or a booking on the call, or when you get a lot of calls after hours.

Find out what your missed calls cost.

Bring one week of your call log to the free audit. We run the sum with you, and tell you whether an AI receptionist, a text back, or your current setup is the right call. If it does not pay, we say so.

Book the free audit →

Sources