Talk to almost any trades or hospitality business in 2026 and a familiar conversation comes up. Someone has tried Jobber, Tradify, ResDiary, OpenTable, an AI receptionist, ChatGPT for quotes, six different scheduling SaaS products. Some stuck. Most didn’t. The owner is now slightly cynical about AI for small business, which is the right place to be — but they also know the work is still being done by hand and that has to stop.
This is a buying guide aimed at that conversation. The lens is practical: where do AI and automation tools genuinely help, where are they mostly marketing, and what do you actually need to spend?
Bookings: the one workflow worth solving first
Across both sectors, the single highest-leverage workflow is taking the booking. A pub fielding seven phone calls during a Friday lunch service. A plumber who can’t pick up at 2pm because he’s under a sink. A removals firm whose admin closes at 5pm — but every Saturday morning brings half the week’s leads.
What good looks like:
- A web form on every device that mirrors what your phone-taker would ask. Date, time, party size or job size, postcode or address, contact details, anything that affects the price. Five fields, no more.
- Real-time availability if your service is genuinely diary-bound. A restaurant taking a table at 19:30 is yes/no based on covers. A plumber juggling three jobs is probably yes Tuesday morning, definitely no Thursday.
- An automated confirmation email within thirty seconds of the booking landing. Not “we’ll get back to you within 24 hours”. The booker has already moved on; you’re competing with their next idea.
- An automated reminder the day before, so no-shows go down. For trades, the reminder doubles as an “is anything different from when you booked?” prompt.
Tools that do this well for the price:
| Sector | Recommended | Why |
|---|---|---|
| Restaurants / pubs | OpenTable or ResDiary for bookings, plus Square for Restaurants if you want POS integrated | Mature, table-management is solid, no surprises |
| Hair / beauty / barber | Fresha | Free for the operator, takes its cut from card processing only — competitive booking flow |
| Mobile trades (plumbers, sparks, gas) | Jobber or Tradify | Quoting + scheduling + invoicing in one. Avoid building a custom equivalent until turnover supports it |
| Removals, hire, multi-day services | A bespoke build is often justified | The off-the-shelf tools assume single-job-per-day workflows; a removals firm has surveys, deposits, scheduling, manpower allocation, all chained |
| Restaurants doing takeaway | Your own ordering site if delivery is >£10k/month | Aggregator commission at 11–30% becomes a six-figure leak fast. See the China Palace worked example below. |
If the volume is small enough that an off-the-shelf tool covers it, use the off-the-shelf tool. Don’t pay £8,000 for something Fresha does for free.
Quotes and estimates: where the AI hype meets reality
Quote generation is the area where ChatGPT-style tools genuinely do help — but in a much narrower way than the marketing suggests.
Where they work:
- Drafting the prose around a quote. The bit where you explain what’s included, what’s out of scope, terms, response times. ChatGPT will write a credible draft in 30 seconds; you then have to actually read it and edit.
- Templating repeat structures. If you do twenty boiler installs a year, the boilerplate (no pun intended) is genuinely the same. A good template + a small AI assist for the variable bits cuts the writing time by 70–80%.
- Pulling product detail from manufacturer datasheets. Useful for trades where the spec lives in PDFs and the customer wants it in their quote.
Where they don’t work, despite confident sales pitches:
- Estimating labour cost. No AI tool understands your team’s actual pace, the access constraints on a job, or whether the customer is going to be a nightmare. The number always needs your hand on it.
- Picking materials at the right grade. AI is genuinely terrible at this — it picks the most plausible answer, not the right one for the property, the regs, or the budget.
- Handling part numbers. Hallucinations are common. Always verify against the supplier’s actual catalogue.
For most trades businesses in 2026, a good quote template plus a 60-second ChatGPT prose pass plus your own labour and material numbers is the right combination. You don’t need a £200/month AI quoting platform; you need a sharper template.
Customer comms: the unsexy quick win
This is where the highest-ROI automation lives, and almost nobody is doing it. The wins:
Pre-arrival texts for trades. “Hi Sarah, this is Mike from Bourne Plumbing — I’ll be with you between 9 and 9:30 this morning. White Transit, reg AB23 ABC.” Sent automatically from the schedule. Reduces no-access rates and gives the customer a face for the visit. Tools: Twilio + a simple webhook, or built into Jobber, Tradify, ServiceM8.
Post-visit review prompts for trades and hospitality alike. Two days after the job, an automated “if we did a good job, here’s the link to leave a Google review; if we didn’t, please reply to this and tell us”. The two-track ask is critical — you’re filtering bad reviews into private feedback and good reviews into public reputation. Tools: Reputation.com, NiceJob, or a 30-line bespoke build that does the same.
Birthday and anniversary emails for hospitality. Sent automatically a week before. Adds an offer (free starter, free dessert). Conversion rates on these are routinely 25–40% compared to <2% for cold marketing emails. Tools: Klaviyo, Mailchimp, or built into your booking system if it captures dates.
Lapsed-customer reactivation. A removals firm hasn’t heard from a previous customer in two years. An automated “we noticed it’s been a while — is there anything we can help with this year?” recovers about 8–12% of dormant relationships. Tools: same email-marketing platforms.
The unifying point: customer comms automation isn’t about replacing the human relationship, it’s about being a touch more attentive than your competitors who haven’t got round to setting any of this up.
The categories that are mostly noise
Some categories of AI for small business tooling are dramatically over-marketed relative to the value delivered. Treat with caution:
“AI receptionists” that answer the phone with a synthetic voice. The technology works — sometimes — but customers in trades and hospitality routinely tell us they’d rather hear “we’ll call you back in twenty minutes” than struggle through a robot. The hidden cost is the calls you lose to people who hang up. Better answer: a real call-answering service like alldayPA at £40–80/month, plus a website form for the people who’d rather not phone.
“AI website builders” that promise a complete site for £19/month. They produce credible drafts, but the SEO, the structured data, the local listings, and the brand consistency aren’t there. You end up with a site that looks fine but ranks nowhere. If your business is local and your customers are searching, you want someone who builds it properly once.
“AI marketing platforms” that automatically generate social media content. Mostly produces forgettable copy that nobody reads. The exception is using ChatGPT to help a human draft posts — that works. Fully autonomous social posting almost always degrades the brand.
“Virtual assistants” that can do everything. No they can’t. They can do narrow, well-defined tasks. Buy them for those tasks; do not buy them on the strength of the demo of the all-singing all-dancing version.
A worked example: a Bourne restaurant
A Chinese restaurant in Bourne (real client, anonymised) runs phone-only ordering with no online channel. They’re losing roughly £11,000 a year in aggregator commission because half their delivery comes through Just Eat at 14% commission, plus they’re missing perhaps fifteen orders per week to their engaged-tone phone line during peak. Total leak: about £17,000 a year.
What they need is not a £19/month off-the-shelf restaurant CMS. They need a bespoke ordering site at chinapalacebourne.co.uk, with collection and delivery, Stripe at 1.4%+20p (a tenth of aggregator commission), a kitchen iPad ticket display, and a customer database that captures email addresses for the 35% of customers who order more than once.
Build cost: £2,500 fixed-price. Annual saving: £17,000+. Payback: Day 60. After that, the saving compounds annually for as long as the restaurant is open.
This example breaks my own rule that off-the-shelf tools are usually fine. The reason: the volume justifies the spend. £17,000 a year of leakage is a different problem to the local barber losing thirty minutes a week to manual diary management.
The lesson is to size the prize before you pick the tool.
A buying checklist
Before you click “buy” on any AI or automation tool for your business:
- State the workflow it’s replacing. Specifically. Not “admin”.
- Quantify the hours per week the workflow currently takes. Stopwatch test, one week.
- Multiply by 52 and by your fully-loaded hourly cost. That’s your ceiling on annual spend.
- Pick a tool whose annual cost is at most one-third of that ceiling. Anything costlier needs to deliver something more than time saving — usually revenue capture.
- Trial for 14 days, with one specific success metric. If it doesn’t meet it, cancel. The trial is a real evaluation, not a free month.
- Re-evaluate every six months. Tools that earned their keep last year may not next year.
The punchline: AI for trades and hospitality is mostly about replacing manual touchpoints with automated ones, not about replacing the human at the centre. The businesses that win do the unsexy customer-comms automation first, the cheap-but-decisive booking system second, and only build bespoke when the volume genuinely justifies it.
If you’re unsure whether a specific workflow in your business justifies a tool, a custom build, or just a sharper template, that’s exactly what a Launchpad consultation answers in 30 minutes.