Tradespeople mostly use AI for the paperwork and messages around the job, not the job itself: replying to enquiries from the van, turning voice notes into job sheets, writing up quotes, catching missed calls, chasing invoices, asking for reviews and explaining work in plain English. Pricing, measurements and safety decisions stay with the tradesperson.
The pattern behind all twelve jobs below is the same. Each one is writing or sorting that normally happens at 9pm at the kitchen table. AI does a first draft in seconds, usually from something you said out loud, and you spend a minute checking it. Several job-management apps now build this in (ServiceM8 has a Smart Writing Helper, Auto-Quote and Auto-Invoice; Jobber sells an AI receptionist), and a phone chat assistant covers the rest.
1. Replying to enquiries while you're on a job
What it does: turns a rushed voice note into a polite, complete reply, so the customer hears back in minutes rather than that evening. Speed matters because people who contact three tradespeople tend to book whoever answers first.
Example, from a joiner (illustrative). Dictated between cuts: "reply to the woman about the loft hatch, yes we do them, need to see it, can do Thursday afternoon or next Monday morning, usually a half day job." The reply the assistant drafts:
Thanks for getting in touch. Yes, we fit loft hatches. I'd need to see the space first to check the joists and sizing. I can come round Thursday afternoon or Monday morning; which suits you? Most hatches are a half-day job once we've agreed the details.
How to start: use the voice button in the ChatGPT, Gemini or Claude app and keep one saved instruction: "Reply as [business], friendly, short, no exclamation marks, always offer two time slots."
Watch out: never let it promise a price or a date you haven't said. The full method is in replying to trade enquiries in 60 seconds.
2. Turning voice notes into job sheets
What it does: converts a two-minute ramble at the end of a job into a tidy job sheet with work done, parts used, time, and anything the customer needs to know.
Example, from a heating engineer (illustrative). Voice note: "Service done at number 14, boiler pressure was low at 0.6, topped up to 1.2, told them to watch it, replaced the condensate trap seal, took about an hour and a quarter, they want a quote for two more radiators upstairs." The assistant produces a job sheet with headings for work carried out, parts, labour (1.25 hours), advice given to the customer, and a follow-up line: "Quote requested: two additional radiators, first floor."
How to start: give the assistant your job sheet headings once and ask it to always use them. Some job apps transcribe voice notes attached to the job for you.
Watch out: readings and part numbers. Speech-to-text hears "0.6" as "6" surprisingly often. Read numbers back before saving. Turning on-site voice notes into job sheets goes further.
3. Writing up quotes (the words, not the prices)
What it does: turns your site notes and your own price list into a written quote with a clear scope, exclusions, assumptions and terms. Customers compare quotes, and the clearer one often wins even at a slightly higher price.
Example, from a roofer (illustrative). Notes: "rear slope, approx 12 slipped/cracked tiles, re-bed 3 ridge tiles, check valley, scaffold tower needed, 1 day." The draft quote sets out each item as a line from the roofer's price list, adds "Excludes: replacement of rotten battens if found once tiles are lifted (priced separately after inspection)", and states a 30-day validity.
How to start: keep your price list in a document and paste it into every quote prompt, telling the assistant to use only those prices. ServiceM8's Auto-Quote works from the job card, notes and items you've added.
Watch out: AI arithmetic. Add the totals up on a calculator every time. A trade-specific walk-through is in can a plumber use AI to write quotes?
4. Catching the calls you can't answer
What it does: when you're up a ladder or under a sink, an AI receptionist answers, takes the details, and either books a visit or texts you a summary. A lighter version, missed-call text-back, simply sends an automatic text to anyone whose call you miss.
Example, with real pricing. Jobber's Receptionist is a $29 a month add-on on several plans, including 30 conversations, then $0.79 per extra conversation; it is included on Jobber's Plus plan. You can set words such as "emergency" or "urgent" that make it transfer the call or text you a transcript snippet. ServiceM8 has its own Phone Agent. Answering services run by people usually cost more per call but handle awkward callers better.
How to start: try missed-call text-back first. If more than a handful of callers a week never ring back, a receptionist is worth pricing; what an AI receptionist costs a tradesperson sets out the numbers.
Watch out: callers who don't use your trigger words. Test with a friend pretending to have a leak but never saying "emergency".
5. Chasing unpaid invoices
What it does: drafts a firm, polite sequence of reminders so you don't put off an awkward message, or send an angry one.
Example, from a painter and decorator (illustrative). The three-step sequence the assistant drafts for an invoice 14 days overdue: a friendly "just checking this reached you" on day one; a clear reminder with the amount, invoice number and bank details on day eight; and a final notice on day fifteen stating the date after which you'll take the next step you've chosen. The decorator's only edit is removing "we understand times are hard", which he felt invited delay.
How to start: write the three messages once and save them as templates in your invoicing app, which can usually send reminders automatically. ServiceM8's Auto-Invoice also drafts the work description on the invoice itself.
Watch out: don't paste your whole customer list into a free chatbot to personalise chasers. More in polite invoice reminders for trades.
6. Asking for reviews, and replying to them
What it does: drafts the review request you send when the job's done, and replies to reviews that sound like you rather than a template.
Example, from a landscaper (illustrative). For a three-star review saying "good work but left the side gate open and the dog got out", a generic reply would say "we're sorry to hear about your experience". The AI draft, given the facts, says: "You're right, and I'm sorry. The gate was our mistake. We've added a gate check to our end-of-day list. Glad the patio came out well." Specific, owns it, and shows other readers the problem was fixed.
How to start: send the request the same day the job finishes, with a direct link. Getting more reviews after every job has the wording.
Watch out: never post an AI reply to a negative review without reading it; a defensive tone does more damage than the review.
7. Social posts from before-and-after photos
What it does: writes a short caption from a photo and two lines of notes, so your work actually gets posted.
Example, from a tiler (illustrative). Photos of a bathroom floor before and after, plus notes: "large format porcelain, levelled uneven floor first, 3 days." The caption: "Before and after: this floor dipped 8mm across the room, so we levelled it before laying large-format porcelain. Three days start to finish. Big tiles show every bump, which is why the prep matters." The last sentence came from the tiler's follow-up prompt asking for "one tip a homeowner would find useful".
How to start: keep a folder of job photos and batch five captions on a Sunday evening.
Watch out: house numbers, car registrations and family photos in the background, and posting a customer's home without asking.
8. Explaining the work in plain English
What it does: turns trade language into something a customer understands, which reduces "why does it cost that much?" conversations.
Example, from a roofer (illustrative). Prompt: "Explain to a homeowner in four sentences why a leaking valley needs the tiles lifting, not just sealant." The answer explains that the valley is the channel where two roof slopes meet, that water runs under the tiles there, and that sealant on top hides the leak without fixing the lining underneath. The roofer pastes it into his quote email.
How to start: save the explanations that work; you'll reuse the same five or six for years.
Watch out: check the explanation is right for your method. AI explains the common approach, which may not be yours.
9. Reading manuals and specification sheets
What it does: answers questions from a product manual you've uploaded, so you're not scrolling a 90-page PDF on a phone in a cupboard.
Example, from a heating engineer (illustrative). The installation manual for a new boiler model goes into Gemini Notebook (formerly NotebookLM), which answers from the documents you give it and shows which part of the source each answer comes from. Asked "what clearance does it need below the casing for servicing?", it gives the figure and the page reference. The engineer checks that page before relying on it.
How to start: build a notebook per manufacturer with the manuals you use most.
Watch out: never ask a general chatbot for installation figures without a manual loaded. It will give a plausible number from a different model.
10. First drafts of risk assessments and method statements
What it does: produces a structured first draft of the paperwork a main contractor or commercial client asks for, which you then make specific to the site.
Example, from a small building firm (illustrative). Prompt: "Draft a method statement for removing a non-load-bearing stud wall in an occupied house: two operatives, dust control, waste removal through the front door." The draft covers sequence, protection of floors, dust sheets and extraction, isolation of any cables in the wall, and waste handling. The builder adds what only he knows: the actual access route, the client's working-from-home hours and where the skip will sit.
How to start: use your best existing method statement as the example in the prompt, so the draft follows your format.
Watch out: a generic draft is not a risk assessment. Someone competent has to walk the site and sign it.
11. Materials lists and comparing supplier quotes
What it does: turns measurements into a first materials list, and lays two or three supplier quotes side by side so you can see the real difference.
Example, from a landscaper (illustrative). Three merchants' quotes for a 24 square metre patio are pasted in. The assistant tables them by item and flags that one quote excludes delivery and another prices sub-base by the tonne rather than the bag. Its first "cheapest overall" answer was wrong because it ignored the missing delivery charge; asking it to "list every item that appears in one quote but not the others" found the gap.
How to start: ask for a comparison table, not a recommendation.
Watch out: quantities and waste allowances. Check areas and tonnages yourself; AI miscalculates more often than you'd expect.
12. Planning the week
What it does: takes a list of jobs, rough durations and areas and suggests an order that cuts driving and fits the awkward ones (school-hours-only customers, deliveries) in the right places.
Example, from a two-van plumbing firm (illustrative). Eleven jobs for the week go in with durations, areas and constraints. The draft plan groups jobs by area per day and puts the two customers who can only do mornings on Tuesday and Thursday. It also scheduled a boiler install on the day the parts weren't arriving, because it didn't know; adding "parts arrive Wednesday" fixed it.
How to start: use it for a draft, then check drive times in your maps app. For more than one or two vans, scheduling software does this properly. Once the order is fixed, ask the assistant for a one-line text to each customer confirming their arrival window; it takes seconds and cuts the "are you still coming?" calls on the day.
Watch out: AI doesn't know traffic, parking or which customer always runs late.
One saved instruction that makes all twelve better
The difference between AI drafts you can send and drafts you rewrite is usually one block of text the assistant sees every time. Most chat apps let you save it as custom instructions or inside a project. A filled-in example for a small roofing firm (illustrative):
About us: two-person roofing firm, domestic work, repairs and
re-roofs. We're direct and friendly, never salesy.
Writing rules: short sentences, no exclamation marks, no "we pride
ourselves", sign off as "[first name], [business name]".
Never: state a price, a date or a guarantee I haven't given you;
give safety advice about roofs or ladders; mention other firms.
Always: offer two appointment times when replying to enquiries;
end quotes with "Valid for 30 days"; ask me if anything is unclear.
Our area: [the towns you cover]. Hours: 8-5 weekdays.With that in place, job 1's enquiry replies, job 3's quotes and job 6's review replies come out in the same voice, and the "never" line blocks three of the mistakes that cause most trouble. Update it whenever you catch yourself making the same correction twice.
Three first-month mistakes, and how they showed up
These are illustrative, but typical of what the first few weeks turn up.
The promised Tuesday. A decorator dictated "tell her we could maybe look at next week" and the assistant wrote "We can start on Tuesday." The customer took the day off work. The fix was the "never state a date I haven't given you" line above, plus reading every reply before sending, which in the first month nobody should skip.
The review reply that argued. A landscaper asked for a reply to a two-star review "explaining it wasn't our fault the turf died". The draft was accurate and defensive, and it read worse than the review. Asking instead for "a reply that acknowledges the problem, explains the watering advice we gave in one sentence, and offers to visit" produced something a future customer would find reassuring.
The total that was out by a line. A joiner's AI-drafted quote listed seven items and a total that included only six. The customer spotted it, accepted the lower figure, and the joiner lost $140. Every total now goes through the quoting app's own calculation, never the chat assistant's.
None of these was a failure of the tool so much as of the setup: missing rules and skipped checks. All three were fixed in a minute once spotted.
What tradespeople don't hand to AI
The same people who use AI for all of the above generally keep these firmly in their own hands:
- Pricing. AI can apply your price list; it can't know your costs, your market or how much you want the job.
- Measurements and calculations that decide what gets ordered or how something is built.
- Certificates and sign-offs. Anything you sign as competent stays yours.
- Safety judgements on site. A checklist draft helps; the decision is made by someone looking at the job.
- Difficult conversations. A draft is fine; a complaint or dispute needs your voice, often on the phone.
What these twelve jobs cost to run
Most of this list works on a free or $20-a-month chat assistant plan. Rough costs, as of September 2026:
| Setup | Covers jobs | Typical cost |
|---|---|---|
| Free chat assistant on your phone | 1, 2, 6, 7, 8, 11, 12 for light use | $0, with usage limits that change |
| Paid chat plan (ChatGPT Plus, Claude Pro, Google AI Pro) | All drafting jobs, plus uploading manuals and quotes | About $20 a month |
| AI built into job-management software | 2, 3, 5 inside the app | Part of the subscription on some plans; check yours |
| AI receptionist add-on | 4 | Jobber: $29 a month for 30 conversations, $0.79 each after |
A week with five of these in a two-person firm
To see how the jobs add up, picture a two-person bathroom fitting business: the owner does surveys and quotes, and a fitter runs most jobs. Before AI, the owner spends roughly eight hours a week on paperwork and messages, mostly in the evening. With five of the jobs above in place, a plausible week looks like this (illustrative):
| Job | Before (hours a week) | After | What changed |
|---|---|---|---|
| Enquiry replies | 2 | 0.75 | Dictated replies between jobs, sent within the hour |
| Quote write-ups | 3 | 1.25 | Drafts from survey notes and the price list; totals checked by calculator |
| Job sheets | 1 | 0.25 | Fitter's voice note at the end of each day, tidied by the assistant |
| Invoice chasers | 1 | 0.25 | Saved three-step templates sent from the invoicing app |
| Review requests and replies | 1 | 0.5 | Same-day request after every finished job |
That is about five hours back in the week, from tools costing around $20 a month. The more telling change is timing: enquiries answered within an hour instead of that night, and quotes sent the day after the survey instead of the following week. Those two usually matter more to winning work than the hours saved. Your own numbers will differ, which is why the timing exercise below is worth the effort.
How to tell whether it's paying off
Hours saved are only half the story for a trade business. Track two speed measures as well, because they are what win work:
- Time to first reply on new enquiries. Check the last ten before you start and the next ten after. Going from "that evening" to "within the hour" is the change customers notice.
- Days from survey to quote. The same comparison. A quote that lands the next morning competes differently from one that arrives a week later.
Add a third, the win rate on quotes, once you have 20 or so sent with the new method; it moves slowly and depends on more than paperwork, so don't judge it on a handful.
If you only try two of these this month
Pick the two that match where your evenings go. For most one- and two-person firms that is enquiry replies (job 1) and quote write-ups (job 3), because both win work directly. Give each two weeks. Time yourself before and after on five jobs, keep a note of every correction you had to make, and only add a third once the first two feel routine. If corrections keep coming back to the same problem (wrong prices, wrong dates), the fix is a better saved instruction, not a different tool.
Further reads
- The First 30 Days of AI in a Trades Business — A month-by-week plan for putting a few of these in place.
- Seven AI Mistakes That Cost Tradespeople Jobs — The AI errors that lose trades businesses work.
- Free vs Paid AI Tools for a Small Trades Business — Which of these jobs the free tools handle well enough.
- Nine Everyday AI Uses for Electricians Beyond Estimating — Nine more uses specific to electrical work.
- Can AI Price a Job Accurately? What Tradespeople Should Check — Why pricing stays with you, and what to check.
- AI Job Scheduling and Dispatch for Trades and Field Service Teams — Scheduling software for when a whiteboard stops coping.
- Why Is AI Bad at Maths? What to Check in Quotes and Invoices — Why totals and quantities need a calculator check.
- When Should a Trades Business Bring In an AI Consultant? — The signals that a trades firm needs outside AI help, a ten-minute scoring sheet, what to gather first and what a good engagement leaves behind.
- Jobber, Housecall Pro, and ServiceTitan AI Features Compared — What each platform's AI actually does for calls, quotes, invoices and reviews, what's included or extra, and a ten-task trial script to test them yourself.
- How a Plumber Can Save Five Hours a Week on Admin With AI — An illustrative time log for a sole plumber, six changes that add up to five hours a week, a four-week rollout and what it costs.
- AI Scheduling for Small Builders: Subcontractors and Deliveries — Use AI to build the trade sequence, turn lead times into order-by dates, run a Friday look-ahead and re-plan when a subcontractor drops out.
- AI Route Planning for Landscaping Crews: Time and Fuel Savings — Measure a baseline week, group customers into area-days, let a routing tool order the stops, and work out what the saved miles and hours are really worth.
- Can AI Produce Accurate Estimates for a Small Building Firm? — Where AI speeds up a builder's estimate, where its prices drift, and a rate-library method that keeps the numbers yours.
- Can AI Give Customers a Ballpark Price for Decorating Work? — How a decorator can let AI give honest ballpark prices from a room-by-room grid, and which jobs it should refuse to price.
- Why Most Landscapers Haven't Adopted AI Yet, and Where to Start — The honest reasons landscaping firms hold off on AI, the three office jobs to start with, and a first-month plan for a four-crew firm.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Jobber help pages, Receptionist powered by Jobber AI (pricing, escalation settings); ServiceM8 AI features page (Smart Writing Helper, Auto-Quote, Auto-Invoice, Phone Agent); Google help page on Gemini Notebook.