AI receptionists for Australian small businesses: what they can and can’t do

An AI receptionist is software that answers your business phone, talks with callers in natural speech, and can answer common questions, take messages and book appointments. It works well after hours, when your team is busy, and for routine bookings. It’s a poor fit for complaints, emergencies and anything that needs judgement, so it needs a clear way to hand callers to a person.

This guide covers what it handles well, where it falls short, the privacy questions to ask, and how to test one properly before your customers hear it.

What an AI receptionist actually does on a call

Strip away the marketing and a typical AI receptionist call goes like this:

1. Answers and introduces itself. A good setup tells the caller straight away that they’re speaking with an AI assistant for your business.
2. Works out what the caller wants. A booking, a question, a quote request, an existing job, or something it should pass on.
3. Follows your instructions. It answers from information you’ve approved: services, hours, service areas, price guidance, policies.
4. Collects details. Name, phone number, address or suburb, the job or appointment type, and any notes.
5. Takes an action. Books into your calendar, transfers the call, or takes a message.
6. Reports back. Sends you a summary by text or email and logs the call in your CRM, often with a transcript.

Its limits are real. It only knows what you’ve given it. It can mishear names and addresses, especially with background noise or a poor mobile connection. And like other AI tools, it can sound confident when it’s wrong. That’s why the instructions, the handover rules and the testing matter more than which voice you pick.

Good fits: after hours, overflow, booking, FAQs

AI receptionists earn their keep on calls that are frequent, predictable and would otherwise go to voicemail. Written questions on your website are a different job, covered in our guide to AI chatbots on a website.

• After hours. Evening and weekend callers get an answer, a booking or a clear next step instead of a beep.
• Overflow. The tradie on a roof, the physio with a patient, the salon owner mid-colour. The AI picks up when nobody else can.
• Routine bookings. Standard appointment types with set durations, where the rules are clear.
• Common questions. Opening hours, service areas, parking, what to bring, whether you do a particular type of job.
• First details on new enquiries. Capturing what the job is, where it is and how urgent, so you can call back prepared.

Poor fits: complaints, complex quotes, emergencies

Some calls should go to a person every time, or be handled with a very short script that moves them on quickly.

• Complaints and upset customers. People who are already frustrated rarely want to explain it to software. Transfer them or promise a call back from a named person.
• Complex quotes. Anything that needs judgement about scope, site conditions or pricing. Let the AI collect the basics and book a proper conversation.
• Emergencies. If a caller describes a risk to life or safety, the AI should tell them to call 000 and stop there. For urgent but non-life-threatening jobs, like a burst pipe, route to your on-call number.
• Sensitive matters. Health details, legal problems, financial hardship. Keep the AI out of these conversations beyond taking a contact number.
• Existing jobs with history. Unless the AI can see the customer’s records, it can’t answer “where’s my order” or “when’s the plumber coming back” reliably.

AI vs a human answering service vs voicemail

An AI receptionist isn’t the only option, and for some businesses it isn’t the best one.

A human answering service is often the better choice if a large share of your calls are emotional, unusual or high stakes, such as a funeral home, a legal practice or an emergency repair business taking distressed callers late at night. Voicemail is fine if calls are rare and nobody is lost by waiting until morning. An AI receptionist makes the most sense when you get a steady flow of routine calls that currently go unanswered.

Whichever you choose, look at your own numbers first. Check your phone system or call log for how many calls go unanswered, at what times, and how many of those callers ring back. That tells you what problem you’re actually solving.

Many businesses end up with a mix. Staff answer during the day, the AI covers overflow and after hours, and a missed-call text-back catches anything that still slips through.

Privacy, call recording and telling callers it’s AI

Tell callers it’s AI. Callers should know up front that they’re talking with an AI assistant, and they should have a way to reach a person or leave a message for one. Pretending the AI is a staff member damages trust when people work it out, and it can create problems under consumer law if callers are misled.

Call recording. Most AI receptionists record or transcribe calls. In Australia, recording phone calls is governed by a mix of federal telecommunications law and state and territory surveillance or listening device laws, and the rules differ between states. The safest practice is to tell every caller at the start that the call is recorded and why. Get advice for your situation, especially if you take calls from several states.

Privacy obligations. Recordings and transcripts contain personal information. The OAIC explains when the Privacy Act applies to a business, and has published guidance on privacy and the use of commercially available AI products, including checklists for choosing and using them. It recommends doing due diligence before adopting a product, not just because it’s available. For staff using AI tools more broadly, see using ChatGPT safely.

Questions to put to any provider:

• Where are recordings and transcripts stored, in Australia or overseas?
• How long are they kept, and can we set or shorten that?
• Is our call data used to train the provider’s models?
• Who at the provider can access our calls?
• Can we delete a caller’s data if they ask?

This is general information, not legal advice.

How to test one before going live

Don’t trust a demo. Demos use easy calls. Test with the calls your business actually gets.

1. Build a test list from real calls. Look back over the last month and write down 20 or so typical call types, including the awkward ones.
2. Call it yourself, then have others call. Different voices and accents, a mobile in the car, a noisy job site, someone who talks fast, someone who rambles.
3. Test the edge cases. An emergency, an angry customer, a wrong number, a supplier, a caller asking for a firm price it shouldn’t give, someone outside your service area.
4. Check every booking it makes. Right job type, right duration, right person, inside your service area, with enough travel time.
5. Start narrow. Go live after hours only, or for overflow only, before letting it answer during business hours.
6. Read the transcripts. For the first few weeks, read every one. Fix the instructions each time it gets something wrong.

Keep a simple log of mistakes and fixes. After a month you’ll know whether it’s helping or creating cleanup work.

Connecting it to your calendar and CRM

An AI receptionist that only takes messages is a slightly better voicemail. The value comes from connecting it to the systems you already run.

Calendar rules. If it books appointments, it needs the same rules your best receptionist follows:

• which job or appointment types it can book, and how long each takes
• which suburbs or postcodes you service
• buffers for travel, setup or clean-down between bookings
• which staff can be booked for what, and daily limits
• what needs a person to approve before it’s confirmed

CRM. Each call should create or update the contact, attach the summary and transcript, record where the caller found you, and trigger the next step. That might be a confirmation text, a task for someone to call back, or a quote follow-up sequence.

Notifications. Send urgent calls to the on-call person straight away and routine ones to a daily summary. If everything pings everyone, people stop reading the alerts.

A sensible way to start

Begin with after-hours and overflow calls, keep people answering during core hours, and review transcripts weekly for the first month. Write your always-transfer list before anything else. If the AI handles routine calls well and hands the hard ones over cleanly, extend its hours. If it doesn’t, you’ve learnt that cheaply.

CentriWeb sets up and looks after AI phone receptionists connected to your calendar and CRM, with the handover rules written around your business. We work with trades, clinics and other local services businesses.