By Rakibul Hasan, Moonlab
An AI receptionist is a voice assistant that answers your business phone, talks to the caller in normal speech, and handles routine requests such as opening hours, quotes and bookings. It does not replace your team. It covers the calls nobody has time to pick up and passes the tricky ones to a person.
This guide explains how an AI receptionist for a small business works, what happens during a call, how booking and handoff are set up, what costs to look at, and when it is not a good idea. Examples are hypothetical, and tool prices change, so check each vendor's current pricing.
What an AI receptionist does
When a call comes in, the assistant answers with a greeting you have approved, listens to what the caller wants, and responds in a natural voice. It can answer questions from information you give it, take a message, collect contact details, book an appointment in your calendar, and send you a summary afterwards.
It is built from three parts: speech recognition that turns the caller's words into text, a language model that decides what to say, and a voice that speaks the reply. Behind those sits a set of rules and connections to your calendar, CRM and phone system. The quality of the result depends far more on those rules and connections than on the voice.
How calls, booking and handoff work
A well-built assistant follows a clear path for each type of call. Think of it as a flowchart with a conversation on top. The same call types come up again and again, so most of the work is deciding what should happen for each one.
- Questions. Opening hours, service areas, parking, what you charge for a standard job. The assistant answers only from the information you provide, and says it will pass the question on when it does not know.
- Booking. It checks your real calendar for free slots, offers two or three times, confirms the choice and sends a text or email confirmation.
- Messages. It takes a name, number and reason for the call, and sends the details to the right person by text or email.
- Urgent calls. Words like emergency, flooding or no heating can trigger a transfer to your mobile straight away.
- Handoff. The call is transferred to a person, or a callback is booked, whenever the caller asks for a human or the assistant is unsure.
AI phone answering compared with voicemail and a human receptionist
Voicemail is free but many callers hang up without leaving a message, and a message left at 6 pm may not be read until the next afternoon. A human receptionist is excellent but works fixed hours, can only take one call at a time and costs a salary.
AI phone answering sits between the two. It answers every call straight away, at any hour, several at once, and writes everything down. It cannot read a caller's mood the way a good person can, and it can misunderstand accents, bad lines or unusual names. For a deeper look at the human side of this choice, see moonlab.space/guides/ai-answering-service-vs-call-center.
What to set up before you switch it on
Most failed rollouts come from skipping preparation. Spend an afternoon on the following and the assistant will sound far more competent.
- A one-page fact sheet: services, areas covered, hours, prices you are happy to quote, and answers to the ten questions callers ask most.
- A list of things it must never do, such as promising a price for a custom job, giving medical or legal advice, or taking card numbers by voice.
- Your real calendar connected, with buffers and holidays set, so it cannot double-book.
- Clear transfer rules: who gets urgent calls, what happens if they do not answer, and where messages go.
- A greeting that tells callers it is an automated assistant. In some places this is expected and in others required, so check your local rules.
Costs to consider
There are usually four cost lines: the voice platform or phone number subscription, usage fees per minute of conversation, any AI model usage, and the one-off setup of the call flow. Some vendors bundle these into a monthly plan with a minutes allowance, and others charge per minute. Prices vary a lot and change often, so ask each vendor for a current quote based on your expected call volume.
A simple way to judge value is to count your missed calls for two weeks and note what a typical new customer is worth. Say a plumbing firm in Manchester misses six calls a week, and one in three would have turned into a job. Even a modest job value makes the sum worth doing before you compare vendors.
Remember the cost of getting it wrong too. A badly set up assistant that annoys callers costs more than a missed call. Pay for testing, not just for launch. Our general cost overview is in moonlab.space/guides/business-automation-cost.
When an AI receptionist is a good fit, and when it is not
It fits businesses with many routine calls, calls outside office hours, or a small team that is often on the job, such as trades, clinics, salons, cleaners and local professional firms. It also helps when calls arrive in bursts that one person cannot handle.
It fits less well when most calls are emotional or complex, such as sensitive legal, medical or bereavement matters, where a calm human voice is part of the service. It also fits poorly if your business details change daily and nobody will keep the fact sheet current. In those cases, consider a human answering service or a missed-call text back instead, covered in moonlab.space/guides/missed-call-text-back-and-online-booking.
Privacy and consent basics
A call assistant handles personal data: voices, names, phone numbers and sometimes health or financial details. If you record or transcribe calls, tell callers at the start. Recording and consent rules differ by country and by US state, and GDPR and UK GDPR apply to callers in Europe and the UK. This is not legal advice, so check your obligations with a professional.
Ask any vendor where call data is stored, how long it is kept, and whether it is used to train models. Choose a retention period that fits your needs and delete what you do not need. Our GDPR checklist is at moonlab.space/guides/gdpr-ai-automation.
If you want to see whether an AI receptionist suits your call patterns, the free automation audit at moonlab.space/contact is a low-pressure place to start.
