AI phone answering for trades and garages: what to hand over and what to keep
An AI receptionist can catch the calls you miss under a car or up a ladder. Here is what it can sensibly handle, what should stay with a person, and the consent basics for recording calls in Australia.
Tav Palapathwala, Founder
7 min read

On this page
- What an AI receptionist can sensibly handle
- What should stay with a person
- How it fits with your booking calendar and job system
- Privacy and call-recording consent: the basics
- How to start small
- FAQ
- Can an AI receptionist book jobs straight into my calendar?
- Do I have to tell callers they are talking to an AI?
- What happens if a caller has an emergency?
- Will it sound robotic to my regular customers?
Most talk about AI phone answering starts with the technology. For a trade or a garage, it should start with a simpler question: which calls do you lose today, and which of those could a system take without anyone noticing the difference?
The phone rings while you are under a car, on a roof or driving between jobs. Some callers leave a message. Others ring the next business on the list. An AI receptionist is not there to replace the person who knows the work. It is there to make sure the call is answered, the details are written down properly, and the right person gets a clean summary to act on.
Here is what it can sensibly do, where a person should stay in charge, and how to start without betting the business on it.
What an AI receptionist can sensibly handle
The good jobs for an AI on the phone are the ones that follow a script you could write down on one page.
Answering the call you would otherwise miss. The most useful thing is often the plainest. It picks up when you cannot, says who it is, and takes the caller's name, number, suburb and what they need. A garage might ask for the rego and make. A plumber might ask whether water is leaking right now.
Booking into real gaps. If it is connected to your calendar or job system, it can offer times that actually exist, book the job, and send the customer a confirmation text. It works from your rules: which jobs go in which slots, how long a service takes, which suburbs you cover, and how much travel time to leave.
Answering the questions you answer over and over. Opening hours, whether you service a particular make, whether you do callouts on weekends, roughly what a standard service includes. These come from a list you approve, not from the AI making something up.
Following up quotes. A quote sent on Tuesday and never answered is a common quiet cost. An automation can send a polite follow-up a few days later, or ring and ask whether the customer has questions, then log the reply against the job.
Taking after-hours calls properly. At 9pm it can take the details, tell the caller when someone will ring back, and decide from your rules whether the call is urgent enough to send to whoever is on call.
What should stay with a person
The line is not complicated. If the call needs judgement, money or care, a person should own it.
Diagnosis and safety. An AI should not tell a caller that the grinding noise is probably the brake pads, or that a gas smell can wait until morning. For anything that could be a safety issue, the script should be short: take the details, give the emergency advice you have approved (such as calling 000 or the gas emergency line), and alert a person now.
Prices beyond your fixed list. It can quote a standard service if you have a set price. Anything that depends on seeing the job is a quote for a person to write.
Complaints and disputes. An unhappy customer wants to be heard by someone who can fix it. The AI's job is to take the details, say who will call back and when, and put it at the top of your list.
Anything you would not put in writing. If you would not want a sentence read back to you later, it does not belong in the script.
This is the same rule we apply to every automation: the system drafts and organises, and a person signs off on anything that matters.
How it fits with your booking calendar and job system
An AI receptionist that is not connected to anything is an answering machine with better manners. The value comes from where the call ends up.
In practice, that usually means three connections. First, your calendar or job management system, so bookings land in real slots and do not double up. Second, your customer records, so a returning customer is recognised and the job history is there. Third, a message to you: a short summary by text or email with the caller's details, what they asked and what was booked.
Two things are worth deciding upfront. What happens when the system is not sure? The safe answer is that it says so, takes a message, and flags it for a person rather than guessing. And who checks the summaries? Even a good system makes mistakes, so for the first few weeks someone should read every one and correct the rules where they are wrong.
If you want to see what the full process looks like, from mapping your calls to testing on past jobs, it is laid out on how it works.
Privacy and call-recording consent: the basics
Most AI phone systems record or transcribe calls, because that is how they write the summary. That brings in two sets of rules. This is general information, not legal advice, and the rules differ by state, so check your own situation.
Federal interception law. The Telecommunications (Interception and Access) Act 1979 defines interception as listening to or recording a call as it passes over the network "without the knowledge of the person making the communication". That word, knowledge, is why businesses play a message at the start: "This call may be recorded and transcribed."
State and territory listening and surveillance device laws. These vary. In Victoria, the Surveillance Devices Act 1999 is generally read as allowing a party to a private conversation to record it. In New South Wales, the Surveillance Devices Act 2007 is stricter, and a party generally needs the consent of the other parties unless an exception applies. If you take calls from more than one state, plan for the stricter rule.
The Privacy Act. Many small businesses with turnover of $3 million or less are not covered by the Privacy Act, but some are, and the OAIC's small business guidance explains who. Even if you are exempt, the OAIC's approach is a sensible standard. Its APP 5 guidelines note that when personal information is collected by phone, the notice can be given at the start of the call, for example with a script or automated message.
The practical version is short. Tell callers at the start that they are speaking with an automated assistant and that the call is recorded. Say why. Offer a way to reach a person. Keep recordings only as long as you need them, and know where your provider stores them.
How to start small
The mistake is trying to automate the whole phone on day one. A better path is to pick one narrow job and prove it.
Start with overflow and after-hours. Let the AI take only the calls you would have missed anyway: when the line is busy, after a set number of rings, or outside hours. Your existing customers still reach you as normal, and you can compare the summaries against what you would have done.
Write the script before you build anything. List the ten questions callers ask most, your answers, the jobs you will book, the suburbs you cover and the things it must never say. That page is most of the work, and it is useful even if you never automate.
Read every call summary for the first month. Look for wrong bookings, missed urgency and answers that sound off. Fix the rules, not just the individual call.
Add booking and quote follow-up once the basics hold. When the summaries are reliably right, connect the calendar. When bookings are reliably right, add the follow-up on open quotes.
The goal is not an AI that runs the phone. It is a phone that is always answered, with a person still making the calls that matter.
If you want help working out which calls are worth handing over, and what it would cost to build, start with an audit. Our other launch guides cover getting receipts and supplier invoices into Xero and what to check before using AI on client data.
FAQ
Can an AI receptionist book jobs straight into my calendar?
It can, if it is connected to your calendar or job management system and works from your rules for job length, travel time and service areas. Starting with message-taking first, then adding booking once the summaries are reliably right, is a lower-risk way to begin.
Do I have to tell callers they are talking to an AI?
Telling callers upfront is good practice, and if the call is recorded or transcribed, federal and state laws make notice important. A short message at the start of the call that names the assistant, says the call is recorded and offers a way to reach a person is a sensible default. This is general information, not legal advice.
What happens if a caller has an emergency?
The script should keep it simple: take the details, give the emergency advice you have approved, and alert a person straight away. An AI should not try to diagnose or reassure on anything that could be a safety issue.
Will it sound robotic to my regular customers?
Modern voice systems can sound natural, but the bigger factor is the script. Short, plain answers in your own words, and a quick handover to a person when needed, usually matter more than the voice itself.
Ignis AI builds and runs AI automations for Australian businesses. People approve every output that matters.

