AI receptionist use cases for trades and home services

Trades are the obvious market for AI receptionists, and for once the obvious market is the right one. Not because tradespeople are behind the times, but because the structure of the work makes missing calls unavoidable. You cannot answer the phone with both hands inside a consumer unit, and you should not answer it on a roof.

This piece goes trade by trade. The details differ more than you would expect, and knowing them is what makes you sound like you have done this before rather than like someone reading a script.

Plumbing and heating

The strongest fit in the whole sector, for one reason: urgency is bimodal. A plumber's calls are either "there is water coming through the ceiling" or "we are thinking about a new bathroom in the spring", and those two need completely different handling.

What a well-configured setup does here:

• Triages emergencies first. Is water actively escaping? Have you turned the stopcock off? Do you know where it is? These questions are useful to the caller and they tell the plumber whether to abandon the current job.
• Captures the boiler make and model for heating calls. Vaillant, Worcester, Ideal - half the callers will not know, and the ones who do save an hour of guessing about parts.
• Confirms the postcode against the coverage area before anything else, because plumbers get calls from twenty miles out constantly.
• Separates landlord and letting agent calls from homeowners. Different billing, different access arrangements, different urgency.

Where it struggles: a caller in a genuine flood is panicking, and a calm synthesised voice asking for their postcode can be maddening. Configure a short path - establish it is an emergency, get a number, and either transfer if the provider supports it or promise an immediate callback. Do not make a distressed person answer eight questions.

Electricians

Similar shape, with a distinctive wrinkle: a large slice of an electrician's inbound is compliance work with a name the caller does not know. They want "one of those certificates for the landlord" - an EICR - and they will not use the word.

Configure the knowledge base with the plain-English versions alongside the proper terms. EICR, landlord safety certificate, periodic inspection, fuse box, consumer unit, RCD, "the trip switch keeps going". The model handles the mapping well when it has been told what the business calls these things and what it charges to look at them.

Useful fields to extract: property type, number of bedrooms or circuits if the caller knows, whether it is a rental, whether there is currently power. That last one determines whether this is a job for tomorrow or a job for now.

Roofing

Roofers have the most extreme seasonality of any trade in this list. It is quiet, then there is a storm, and then the phone rings for four days without stopping.

That surge is exactly the case a human receptionist handles worst and an AI handles best. There is no queue, no engaged tone, and no fifteenth caller giving up. What you want configured for storm weeks:

• Is anything actively coming into the property?
• Is it a house, a bungalow, or something with access difficulties?
• Is it going through insurance? This changes the whole job.
• Can they take a photo from the ground and send it to a number?

That last one is worth setting up properly. A photo turns a site visit into a phone quote for a decent share of jobs, and the AI can give the caller the number to send it to.

Locksmiths

The only trade in this list where the AI receptionist is answering calls from people standing in the street in the dark, and the constraints follow from that.

Calls are short, urgent and almost entirely out of hours. Ninety percent of the value is: what is the address, is it a UPVC door or a wooden one, are you locked out or is the lock broken, and how quickly can someone be there. Everything else can wait.

Two cautions. Background noise on a street at midnight is brutal for speech recognition, so keep questions short and confirm the address back. And there is a real safety dimension - a caller who says they are locked out with a child inside needs a person, not a form. Configure that path deliberately.

Gas engineers

Everything from the heating section applies, plus one absolute rule: any mention of a gas smell must stop the conversation.

The AI should tell the caller to leave the property and ring the national gas emergency number, and it should not try to book anything. This is not a judgement call to leave to a model deciding what seems helpful - it should be an explicit instruction in the configuration, tested before the client goes live. Test it yourself by ringing in and saying you can smell gas. If the response is anything other than the right one, fix it before it is live.

Cleaning, gardening and grounds maintenance

A different shape entirely. Almost nothing is urgent, and the calls are about recurring arrangements rather than one-off jobs.

The useful configuration here is about qualification: how many bedrooms, how often, weekly or fortnightly, domestic or commercial, do you have parking, is there a dog. Gardeners want to know the size of the plot and whether there is rear access wide enough for a mower. This is a service where a well-briefed AI can genuinely quote a range on the call, provided the client has given you a pricing structure that survives contact with reality.

Removals, skip hire and waste

Heavily price-shopped. Callers are ringing four firms in a row and comparing. Speed of response matters more than depth, and a business that answers when the other three do not has already won a share of those calls.

Fields: from and to postcodes, date, property size, whether there is a lift, parking restrictions, permit needed. For skips, the size and whether it is going on the road - because that means a permit, and a permit means a longer lead time the caller needs to hear about now rather than later.

Where an AI receptionist is a poor fit

Say this to prospects and they will trust the rest of what you say.

• Businesses with a full-time office person. If someone sits there and answers the phone all day and does it well, you are selling a solution to a problem they do not have. There may still be an out-of-hours case, but be clear that is what you are selling.
• Very low volume. A sole trader taking three calls a week does not need this.
• Highly technical B2B specification work. If every call is a detailed engineering discussion, the AI cannot follow it and will annoy people who know their subject.
• Anything with a strong regulatory or safeguarding dimension beyond a simple stop-and-redirect rule. Care services, for example, need a person.

Practicalities that apply to every trade

Coverage area first, always. It is the single most valuable filter and the easiest to get right.

Ring the client's own number as a mystery shopper before you configure anything. You will learn more in one call than in an hour of questions, and it usually gives you the strongest thing you can say in the follow-up meeting.

And be careful with what you promise on the platform side. Transferring a live caller to a mobile, dropping a job straight into a calendar, recording calls, or pushing details into whatever software the client already uses - each of those depends on the voice provider and the specific configuration, not on the AI. Confirm before you commit. Where call recording is switched on, the business is responsible for telling callers and complying with UK data protection law.

Finally, disclosure. If a caller asks whether they are speaking to a person, the answer is no, plainly given. Trades depend on local reputation more than almost any other sector, and a caller who feels tricked will say so on a community page by lunchtime.

For how the service is packaged and priced, see pricing and what's included. For the wider set of services you can sell into the same businesses, local business services is the place to start, and if you want to hear one handle a trade call, the live demo is there.