What Is an AI Receptionist? A Plain-English 2026 Guide
An AI receptionist answers your business phone and turns each call into a message, a quote, a booking or a dispatched job. What it is and isn't, what it costs at published prices, the rules that apply, and how to decide.

An AI receptionist is software that answers your business phone, talks with callers in plain language and turns each call into something you can act on: a message, a quote, a booked appointment or a dispatched job, depending on the product and how you set it up. It is not voicemail, a press-1 phone menu or a human answering service, though it can work alongside any of them. This guide explains what an AI receptionist is and isn't, what it costs at published 2026 prices, the rules that apply, and how to decide whether your shop needs one.
Key takeaways
- An AI receptionist is an AI system in the sense NIST uses: a machine-based system that generates outputs such as predictions, recommendations or decisions. Here the output is what happens to your call.
- Products differ most in what the call produces: a message for you to return, a price, a time on your calendar, or a technician on the way.
- Published 2026 prices start at $0 for 25 calls a month at Smith.ai, $29 a month for Jobber's add-on and $49 a month at RingCentral, against $18.27 an hour median pay for a human receptionist.
- The usual rules still apply: announce call recording, and know that outbound AI calls fall under the TCPA, as the FCC confirmed in 2024.
- It fits shops that miss calls on jobs or after hours and have repeatable intake; it doesn't replace judgment on hard calls.
What is an AI receptionist, in plain English?
Think of the part of a receptionist's job that happens on the phone. The Bureau of Labor Statistics describes receptionists as doing tasks such as answering phones, receiving visitors and providing information about the business. An AI receptionist does the phone part with software: it answers, asks what the caller needs, answers common questions and records the details, at any hour you route calls to it.
It counts as artificial intelligence in the formal sense. NIST's AI Risk Management Framework describes an AI system as "an engineered or machine-based system that can, for a given set of objectives, generate outputs such as predictions, recommendations, or decisions influencing real or virtual environments" (NIST AI 100-1, January 2023). For a receptionist, the objective is to handle the call the way you would, and the outputs are the words it says and the actions it takes: a message, a booking, a transfer.
Under the hood, speech recognition turns the caller's words into text, a language model decides what to say, connected tools check things like your calendar, and text-to-speech speaks the reply. Our guide to how an AI receptionist works follows one call through each of those steps.
How is it different from voicemail, phone menus and answering services?
| Option | What the caller gets | What you get | Best for | Limits |
|---|---|---|---|---|
| Voicemail | A beep | A message, if they leave one | Very low call volume, fast callbacks | No answer, no price, no time |
| Phone menu (IVR) | "Press 1 for..." | Calls routed by button | Sending calls to the right person | Doesn't take details or answer questions |
| Missed-call text-back | A text after the call rings out | A reply, if they send one | Catching some missed callers | The caller already hung up once |
| Human answering service | A person | A message, sometimes a booking | Calls that need a human touch | Billed by the minute or call; may not know your prices |
| AI receptionist | A spoken conversation at any hour | A message, quote, booking or dispatch, depending on setup | Coverage and consistent intake | Judgment and unusual calls |
Voicemail's weak spot is the callback. In a 2020 Pew Research Center survey, 67% of U.S. adults said their general practice is not to answer calls from unknown numbers but to check voicemail if one is left, so a message-and-callback system depends on them answering your return call. For deeper comparisons, see our guides to phone menus, missed-call text-back, voicemail and human receptionists.
What can an AI receptionist do?
What a given product does depends on its features and your setup, but the menu looks like this:
- Answer every call you route to it, nights and weekends included.
- Take a complete message: name, callback number, address, what's wrong and how urgent it is.
- Answer common questions about hours, service area, payment methods and services.
- Quote from a price list or rules you load. Goodcall, for example, says its agent can give a price range from your price list.
- Book into a calendar. Jobber says its Receptionist can help with booking a job in your calendar, and RingCentral says its AI can schedule or reschedule appointments by connecting to your calendar.
- Dispatch, meaning assign a technician and give an arrival window, on systems built for it.
- Text confirmations or payment links, and transfer calls to a person.
- Speak more than one language. Quo says its Sona agent speaks English, Spanish and French; others publish English and Spanish, and some English only.
Our guide to what booking and dispatch require explains the difference between those levels and what each needs from you.
What does a call with an AI receptionist sound like?
Here is an illustrative after-hours call to a garage door company, written to show the shape of a good call rather than any particular product:
AI: "Thanks for calling [shop name]. This call is recorded, and you're speaking with our automated assistant. How can I help?"
Caller: "My garage door won't close and I can't leave the house open overnight."
AI: "I can help with that. What's the address?" (The caller gives it, and the AI reads it back.)
AI: "Is anyone in danger, or is the door stuck partway with a broken spring or cable?"
Caller: "Nobody's hurt. I heard a loud bang and now it won't move."
AI: "Thanks. I've marked this urgent and sent it to our on-call technician, who will call you back at this number. Is this the best number to reach you?"
On a message-taking plan, the call ends there and the owner gets a summary with the address, the problem and the urgency. On a plan that books, the AI would offer the next open slot from the calendar instead; on one that dispatches, it would name the technician and an arrival window. Our annotated call transcript walks through a longer example line by line.
What kinds of AI receptionist are there?
Products also differ by where the AI lives, which decides how it connects to the rest of your business:
- Built into field-service software, such as Jobber's Receptionist or Housecall Pro's CSR AI, so bookings land in the calendar and customer list you already use.
- Built into a business phone system, such as Quo's Sona or RingCentral's AI Receptionist, so calls, transfers and logs stay in your team's phone app.
- Standalone answering apps, such as Smith.ai, Rosie, Goodcall and Dialzara, which answer whatever calls you forward to them.
- Trade-focused systems that pair answering with scheduling, dispatch and payments, such as TheKeyBot's Core plans.
Our roundup of AI phone answering apps compares all of these on their published prices and features.
What can't it do, and what shouldn't it do?
It can't use judgment the way a person does. An upset customer, a negotiation over a commercial job or a situation that fits no category needs a human. Good setups hand those calls off; our guide to what an AI receptionist can't do lists the triggers.
It shouldn't answer from memory. Language models can produce what NIST's Generative AI Profile calls confabulation: "the production of confidently stated but erroneous or false content." Prices, open times and service areas should come from your own data, and when the data has no answer, the AI should say a technician will confirm.
It must put safety first. A caller reporting a child or pet locked in a car needs to hear "call 911" before any question about the vehicle; the National Weather Service says it is never safe to leave a toddler, disabled person or pet locked in a car. Write your safety rules before anything else.
What does an AI receptionist cost in 2026?
List prices from each vendor's own page, as of October 2026. They bill in different units, so convert each to your own monthly calls and minutes.
| Product | Entry price as published | Billing unit |
|---|---|---|
| Smith.ai AI Receptionist | $0 for 25 calls a month, then $3.00 a call | Calls |
| Jobber Receptionist (add-on) | $29 a month with 30 conversations, then $0.79 each | Conversations |
| Dialzara | $29 a month for 60 minutes | Minutes |
| RingCentral AI Receptionist | $49 a month with 100 minutes, then $0.50 a minute | Minutes |
| Rosie | $49 a month for 250 minutes | Minutes |
| Goodcall | $79 a month for 100 unique callers | Unique callers |
| TheKeyBot Lite | $149 a month with 100 calls, then 50 cents a minute (pricing) | Calls, then minutes |
| TheKeyBot Core | $500 a month with 500 minutes, then 45 cents a minute (pricing) | Minutes |
Disclosure: TheKeyBot, which publishes this guide, makes an AI receptionist. Lite answers 24/7 in English and Spanish, takes the message and sends it to your phone, and doesn't quote or book. Core adds scheduling and GPS dispatch, POS and payments, QuickBooks sync and unlimited users. No plan has a setup fee, and Core, Pro and Elite include a 14-day free trial.
For comparison, Ruby's live receptionists start at $250 a month for 50 minutes, and the BLS puts the median receptionist wage at $18.27 an hour in May 2025. Cheap and expensive options don't do the same work, so compare what each call produces before comparing price. Our roundup of AI phone answering apps prices nine options at the same volume.
What rules apply to an AI receptionist?
Recording. Most AI receptionists record calls. Federal law allows recording when one party consents (18 U.S.C. § 2511), but some states require everyone's consent; Washington, for example, treats a recorded announcement as consent (RCW 9.73.030). Announce the recording at the start of every call. This is not legal advice; check your state.
Outbound calls. Answering inbound calls and placing outbound ones sit under different rules. In February 2024 the FCC confirmed that the TCPA's restrictions on "artificial or prerecorded voice" encompass current AI technologies that generate human voices (FCC 24-17), so AI follow-up calls generally need the called party's prior express consent.
Health information. Most trade businesses are not HIPAA covered entities; HHS says an entity that is neither a covered entity nor a business associate does not have to comply with the HIPAA Rules. Our privacy and HIPAA guide covers the details.
Claims. If a vendor promises a percentage, ask how it was measured. Then-FTC Chair Lina M. Khan said in 2024 that "there is no AI exemption from the laws on the books" (FTC).
Does your shop need an AI receptionist?
Answer these with last month's call log in front of you.
| Question | If yes | It points toward |
|---|---|---|
| Do calls ring out while you're on jobs, driving or asleep? | Coverage is the gap | An AI receptionist or an answering service for those hours |
| Do two or more calls arrive at once? | Capacity is the gap | An AI receptionist, which can be set up to take several at once |
| Do most callers want a price before they book? | Quoting matters | A product that quotes from your price book, or keep quoting yourself |
| Do you want to approve every job before anything is promised? | Control matters | A message-taking AI and fast callbacks |
| Do many callers prefer Spanish? | Language matters | A product that publishes Spanish, tested after hours |
| Do you take a few calls a week and return them within minutes? | The gap is small | Voicemail may still be fine |
On language: in the Census Bureau's 2018-2022 survey, 78.3% of people age 5 and older spoke only English at home, and Spanish was 61.1% of the rest. Your local share may be far higher or lower, so check your own calls.
A worked example, with made-up numbers, shows how to think about the gap. Suppose a solo HVAC contractor misses about 10 calls a week while on jobs, and from experience 3 of every 10 answered calls turn into paid work averaging $250. If even half the missed callers would have booked at that rate, that's roughly $375 a week in work going elsewhere. Your own call log, booking rate and ticket size are the only numbers that matter; plug them in.
What do you need to set one up?
Setup is mostly writing down what you already know. Expect to provide:
- Business basics: name and how to pronounce it, hours, service area, payment methods, licensing if callers ask.
- Services and common questions, in the words your callers use.
- Your rules: safety calls, what counts as urgent, who gets woken up, and which calls always go to a person.
- A price book, if the product will quote, including after-hours and trip charges.
- Calendar and roster access, if it will book or dispatch.
- Where messages should go: text, app, email or your scheduling software.
How long it takes varies by product. TheKeyBot's Lite page describes a setup wizard and about 10 minutes to go live, Dialzara says every paid plan includes a one-to-one onboarding call, and products that quote and book take longer because the price book and calendar have to be right. Our first-week setup guide goes day by day.
What goes wrong in the first month?
Most early problems are setup gaps rather than technology failures, and they show up in the first few days of recordings:
- Calls that never reach the AI because forwarding was set only for busy, or only for some hours.
- Questions it can't answer because nobody wrote down the answer.
- Wrong service area or hours, usually from an outdated website or setup form.
- Mispronounced names and streets, which most products let you correct.
- Summaries missing a detail you need, such as the vehicle or the gate code, which means the intake questions need one more line.
Listen to a handful of calls every day in the first week, fix what you hear, then switch to a weekly review. Our 90-day review guide lists the numbers worth tracking after that.
How do you choose one?
- Decide what the call should produce for each call type: a message, a quote, a booking or a dispatch.
- Price the options at your volume, in the unit each vendor bills.
- Check languages and hours in writing.
- Ask the data questions: what's recorded, who can hear it, how long it's kept and whether it trains AI models. Our buyer's checklist has 25 questions.
- Test on your own line with your real details loaded; our trial test-call script lists the calls.
What should you do this week?
- Pull 30 days of calls and count the ones that rang out, by hour.
- Tag 20 recent calls by type and note what each caller needed by the end of the call.
- Write your safety rules and your "always send to a person" list.
- Shortlist two or three products that produce what you need, and book trials.
- Place the same test calls on each and read what arrives on your phone.
Frequently Asked Questions
What is an AI receptionist?
It's software that answers your business phone, talks with callers in plain language and turns each call into a message, a quote, a booking or a dispatched job, depending on the product and your setup. It can answer at any hour you route calls to it.
How is an AI receptionist different from an answering service?
An answering service puts a person on the line, usually billed by the minute or the call, while an AI receptionist uses software. People handle judgment calls better; AI covers every hour at a lower published price and asks the same questions every time.
How much does an AI receptionist cost?
At October 2026 list prices, entry plans include $0 for 25 calls a month at Smith.ai, a $29-a-month add-on at Jobber, $49 a month at RingCentral and Rosie, and $149 a month for TheKeyBot Lite. Vendors bill by calls, minutes, conversations or unique callers, so price each at your own volume.
Can an AI receptionist book appointments?
Many can, if they're connected to your calendar and you allow it. Others take messages only, so check what the caller has when the call ends, and test a real booking on a trial.
Is it legal to use an AI receptionist?
Yes, but the usual rules apply. Announce call recording, since some states, such as Washington, require every party's consent, and get prior express consent before AI-voiced outbound calls under the TCPA. This is not legal advice; check your state.
Will callers know they're talking to an AI?
Some will, and you should tell them anyway, for example in the greeting that announces recording. We know of no reliable public study of how trade customers react, so listen to your own recordings and offer a person when asked.
Sources
- NIST AI 100-1, "Artificial Intelligence Risk Management Framework (AI RMF 1.0)" (January 2023): https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-1.pdf
- NIST AI 600-1, "Generative Artificial Intelligence Profile" (July 2024): https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.600-1.pdf
- U.S. Bureau of Labor Statistics, Occupational Outlook Handbook, "Receptionists": https://www.bls.gov/ooh/office-and-administrative-support/receptionists.htm
- Pew Research Center, "Most Americans don't answer cellphone calls from unknown numbers" (December 14, 2020): https://www.pewresearch.org/short-reads/2020/12/14/most-americans-dont-answer-cellphone-calls-from-unknown-numbers/
- Smith.ai, AI Receptionist pricing: https://smith.ai/pricing/ai-receptionist
- Jobber Help Center, "Receptionist–powered by Jobber AI": https://help.getjobber.com/en/articles/receptionistpowered-by-jobber-ai/
- RingCentral, AI Receptionist plans and pricing: https://www.ringcentral.com/pricing/ai-receptionist.html
- Quo, AI answering service (Sona): https://www.quo.com/answering-service
- Goodcall, how it works: https://www.goodcall.com/how-it-works
- Goodcall, pricing: https://www.goodcall.com/pricing
- Dialzara, pricing: https://dialzara.com/pricing
- Rosie, pricing: https://heyrosie.com/pricing
- Ruby, plans and pricing: https://www.ruby.com/plans-and-pricing/
- Housecall Pro Help Center, "CSR AI Overview": https://help.housecallpro.com/en/articles/9740104-csr-ai-overview
- National Weather Service, "Children, Pets and Vehicles": https://www.weather.gov/safety/heat-children-pets
- 18 U.S. Code § 2511, Cornell Legal Information Institute: https://www.law.cornell.edu/uscode/text/18/2511
- Revised Code of Washington 9.73.030: https://app.leg.wa.gov/RCW/default.aspx?cite=9.73.030
- Federal Communications Commission, Declaratory Ruling FCC 24-17 (released February 8, 2024): https://docs.fcc.gov/public/attachments/FCC-24-17A1.pdf
- U.S. Department of Health and Human Services, "Covered Entities and Business Associates": https://www.hhs.gov/hipaa/for-professionals/covered-entities/index.html
- Federal Trade Commission, crackdown on deceptive AI claims (September 25, 2024): https://www.ftc.gov/news-events/news/press-releases/2024/09/ftc-announces-crackdown-deceptive-ai-claims-schemes
- U.S. Census Bureau, language spoken at home, 2018-2022 ACS (December 7, 2023): https://www.census.gov/newsroom/press-releases/2023/language-at-home-acs-5-year.html
- TheKeyBot pricing: https://www.thekeybot.com/pricing
About the Author
TheKeyBot Team is dedicated to helping locksmiths grow their businesses through AI automation and smart technology. With years of experience in the locksmith industry, our team provides actionable insights and proven strategies.
