AI phone answering for trades: how it works

What AI phone answering actually is

Most trade owners have used an answering machine, a call-forwarding service or a human answering bureau. AI phone answering sits somewhere between those. When a call comes in, software converts the caller's speech to text, a language model works out what they mean and what to say next within the rules you have set, and a synthetic voice replies. The conversation continues until the system has done what it was set up to do: book the job, take a detailed message, transfer the call or give the caller an answer you have approved.

The difference from a recorded menu is that callers can speak normally. Someone can say "the boiler's showing F28 and there's no hot water, we're in Kettering" and the system can pick out the fault, the postcode area and the urgency, then ask the follow-up questions you would ask. The difference from a human bureau is that it can answer many calls at once, at any hour, using the same script every time.

What happens on a typical call

1. 01

   Greeting

   The call is answered in your business name with your approved wording, and usually states that the caller is speaking with an automated assistant.

2. 02

   Understanding the request

   The caller explains in their own words. The system identifies the type of job, the location and any urgency signals.

3. 03

   Questions you would ask

   Name, number, address and postcode, who is calling (homeowner, tenant, landlord or agent), the fault in their words, appliance make and fault code, and access details.

4. 04

   Checking your rules

   Is the postcode in your area? Is this work you do? Does it meet your definition of an emergency? Is there a safety script to read first?

5. 05

   Acting

   Book into the diary, transfer to the on-call engineer, take a call-back request, or answer a question from your approved list.

6. 06

   Summary

   A written summary, transcript and often a recording are sent to you, so you can check what was said and follow up.

The quality of each step depends far more on the set-up than on the technology. A system given your real call-out order, postcode list and safety scripts behaves like a well-briefed office assistant. One given a generic script behaves like a stranger.

What it handles well

• Calls you cannot take: up a ladder, under a floor, driving, in the evening or at weekends.
• Collecting complete job details in a consistent format, so engineers stop phoning customers back for the postcode or the boiler make.
• Routine bookings such as annual services, safety checks and cleans, when connected to a diary that has been tested.
• Turning away out-of-area or out-of-scope work politely, without promising a visit.
• Answering the same questions you answer every day: areas covered, hours, the kinds of work you take on and your call-out policy.
• Busy-period overflow, because it can take several calls at once.

Notice what these have in common. Each one is a call where the right answer is already known and written down somewhere in the business, usually in the owner's head. AI answering works when that knowledge is moved into clear rules; it struggles when the right answer depends on seeing the job, knowing the customer's history or weighing up one emergency against another. That is also the line between the calls it should finish on its own and the calls it should pass to you.

Where it still falls short

Being honest about the limits is what makes AI answering safe to use. The common weak points are predictable, and each one has a mitigation.

How widely trades use AI today

Construction, which includes most heating, electrical and roofing firms, is one of the slowest sectors to adopt AI. ONS figures show only 13% of UK construction businesses reported using AI, compared with 58% in information and communication^[1]. A government survey of 3,500 businesses found 88% of construction businesses neither used AI nor planned to, against 80% across all sectors^[2].

Size matters too. 28% of UK businesses with up to nine employees reported using at least one AI technology, compared with 49% of those with 250 or more^[3]. The most common reasons for holding back are a lack of identified need and limited skills, rather than cost^[4]. For small trade firms, the practical question is not whether AI is fashionable but whether missed calls are costing work, and whether a tool can be set up to handle them safely.

How callers feel about it

Your customers are more familiar with AI than many owners assume. In a government tracker survey, six in ten UK adults said they had used an AI chatbot in the past three months, and 44% used one at least monthly^[5]. Attitudes are mixed, though: around four in ten expected AI to have a positive impact on them and three in ten a negative one^[6].

That mix argues for transparency and choice. Say at the start that the caller is speaking with an automated assistant, keep the conversation short and focused, and always give a route to a person for anything the caller is uneasy about.

Questions to ask before you choose a system

1. Can I control exactly what it says, including the greeting, the questions and the answers it may give?
2. What happens when it does not know the answer, and can I see those calls?
3. How does it handle emergency keywords such as gas smell, sparks or water through a ceiling?
4. Can it follow my on-call order: engineer, backup, then call-back request?
5. Which diaries or job systems does it connect to today, and which are only planned?
6. Do I get a recording, transcript and summary of every call, and how long are they kept?
7. Where is call data stored, and is there a data processing agreement under UK GDPR?
8. How quickly can I change a rule, such as switching off routine bookings during a rush?

Ask for a test number and ring it yourself with your five most common calls, plus one awkward one. You will learn more in twenty minutes of testing than from any demo.

Example: a roofer's evening calls

Example (fictional). A three-person roofing firm in Plymouth gets most new enquiries between 5 pm and 8 pm, after customers spot damage on their way home. The owner sets up AI answering for evenings and weekends only. It asks for the address, the type of roof, where the water is coming in and whether it is getting worse, and checks the postcode against the firm's area. Active leaks during storms are passed straight to the owner's mobile; everything else becomes a survey request with a summary in the morning.

In the first month the owner reads every transcript, tightens two questions that confused callers and adds an approved answer about gutter cleaning, which the firm does not offer. By the second month he only reads the summaries. The phones are no different during the day; the evenings simply stop leaking work.

Where an AI employee fits

Some products go beyond answering into what is often called an AI employee: following your call-out order, filling routine slots in a tested diary and handing over a complete job summary. Ava is built this way for trades, with recordings, transcripts and summaries for every call. Whichever tool you consider, judge it on the set-up controls and hand-over rules above rather than on how human the voice sounds.