Construction and hospitality are usually the sectors told they're being left behind on AI. Recent UK government research backs that up: roughly nine in ten construction and hospitality businesses neither use AI nor have any plan to. Retail sits at a similar level.
That framing misses the more useful point. Trades businesses are not behind because the work doesn't suit AI. They're behind because most AI advice is written for people who sit at a desk all day, and a plumber, electrician, or builder does not. The pains in a trades business are actually some of the clearest AI wins going, precisely because they're physical and specific: a phone that rings while your hands are covered in something, a quote that has to wait until you're home and tired, an invoice that sits unchased because chasing it feels like nagging a customer you'll see again next month.
Here are the three that come up most, what the AI version actually looks like, and where it stops.
AI can capture a missed call
You're up a ladder, under a sink, or halfway through a first fix. The phone rings. You can't answer it, and by the time you're free, the caller has rung the next name on the list.
This is the most common way trades businesses lose work, and it isn't really about being disorganised. It's a physical constraint: you cannot do the job and answer the phone at the same time.
What an AI phone assistant actually does here is narrow but useful. It picks up, gives a normal-sounding greeting, takes the caller's name, number, and a rough description of the job, and books a callback slot. It doesn't try to sound like you, and it doesn't try to diagnose the job or quote a price on the spot. The moment a caller says something the assistant can't handle cleanly (an emergency, a complaint, someone asking for a mate's rates), it hands straight to a person, or clearly offers a callback rather than pretending it's already sorted. A caller who feels stuck in a loop, repeating themselves to a system that isn't helping, is worse than a call going to voicemail. The whole point is that nobody feels fobbed off.
The limit is real: this is a booking and note-taking layer, not a salesperson. It gets you the details and the slot. You still ring back and close the job.
AI can draft the first quote
You get home, you're tired, and now you've got three site visits to turn into three priced quotes. Most tradespeople either do this late at night or let it slide a day, and a day's delay is often the difference between winning the job and the customer going with whoever replied first.
The useful AI version here is a quote builder: you describe the job (materials, the size of it, site conditions, anything unusual) and it drafts a structured, priced quote for you to check and send. We built this for our own site, for exactly this reason: a visitor describes what they want and gets back a realistic, priced quote instead of a "we'll get back to you" form. The AI isn't inventing your prices out of thin air; it works from the pricing logic and rates you give it. The job it does is turn a rough description into something structured and ready, quickly enough that you can send it the same evening instead of after a late night at the laptop.
You still check it before it goes out. Materials prices shift, sites have quirks a description doesn't capture, and a customer relationship sometimes needs a number nudged for reasons that have nothing to do with the job. That review step doesn't disappear; it's just quick, and it happens from the sofa rather than at a desk at midnight.