AI for plumbers, as I mean it here, doesn't diagnose a boiler or sweat a joint. It runs the office: drafting quotes for water heater and furnace replacements or a bathroom remodel, sorting the inbox when no-heat calls pile up in the first cold week, and screening supplier invoices and fuel receipts for bookkeeping. In my business these jobs go to AI employees, AI roles with a fixed job, their own rules and your approval before anything leaves the shop. The diagnosis on site, your pricing and every promise to a customer stay with you. The roles on this page run in my own business. Here's how you use them in a plumbing or HVAC shop.
A plumbing or HVAC shop runs on two clocks: scheduled jobs and everything that breaks in between. At the desk:
Peak season: When the weather turns, outages, callbacks and reschedules arrive at once.
Quotes with many line items: equipment, parts, removal, install, startup. Written at night.
Maintenance: service agreements, reminders, tune-ups to spread out before the season.
Supplier invoices: Lots of deliveries mean lots of receipts, plus fuel cards, tools and trucks.
Phone coverage: The phone rings while everyone is at a customer's house.
Website: services, service area, emergency number, often untouched for years.
None of this needs your license, so part of it can go to an AI, as long as you approve it.
AI for plumbers: which AI employee takes which office task
My business runs on an AI workforce of fifteen employees. Three fit a trade shop best, plus Ralf and Sebastian for the website and one block without its own AI employee:
Every row follows my rule: the AI drafts, a human approves.
Quotes, for example for a replacement: Anton
Anton runs in my own business. He turns an initial conversation into a quote PDF in my design and, once a client says yes, creates the contact, quote and invoice as drafts in my accounting tool, Lexware Office. He's trained on five to ten old quotes instead of prompt tinkering. His rules are fixed: without a price source, he sets no price, and if he finds two different prices for the same service, he lists both with their source and asks instead of averaging. No quote and no invoice leaves the system without my click. According to his page, this also works for maintenance contracts.
Lexware Office is a German tool; for other systems his page names an adapter route, with everything as a draft and one test draft before going live.
Here's how it looks in a plumbing shop: after the site visit you record a short voice note with equipment, work, parts and anything unusual. Anton finds comparable replacement quotes in your history, takes over the line items with their units, pulls prices from your current price list and flags what's missing or has changed noticeably. You check quantities and pricing and approve. If a rebate or financing program needs specific wording or line items on the contract, that goes into the template once, so it isn't forgotten in the middle of the season. How the price book behind this has to be built is in Automating Quotes in the Trades.
Frieda also runs in my own business. She sorts the inbox into four piles, Now, Today, Can wait and Trash, plus a fifth for cases that need my judgment. She reads sender, subject and preview first, drafts replies, delivers meeting briefings and pre-sorts receipts. She has never sent anything on her own. The inbox connects through official connectors for Gmail or Microsoft 365, Outlook included.
In a plumbing or HVAC shop: no heat is Now. A request for a new bathroom is Today, with a draft reply asking for photos, measurements and a preferred date. The supply house newsletter can wait. The email where a customer threatens a complaint over a leak lands in the fifth pile, with you. The pile system without AI is in Sorting Your Inbox: The Pile System.
Receipts and supplier invoices: Peter
Peter has been running in my business since August 19, 2026. He works in my bookkeeping interface in the browser, screens newly arrived receipts, assigns each one a receipt type and matches the credit card statement line by line. In his first run he screened 116 receipts, matched 38 of 40 correctly and needed just over an hour. The two left open he flagged instead of forcing them to match. He doesn't book entries; final approval stays with me and my accountant.
In a trade shop: supply house invoices, which Peter also checks against the delivery ticket, plus fuel cards, tools, trucks. Peter is a template to download in the community that you adapt by chat, without code; Frieda pre-sorts receipts. Filing is covered in Digital Receipt Filing in Germany: GoBD Rules, AI in bookkeeping in AI Bookkeeping: What Actually Works Today.
Phone and emergency calls
I don't have an AI employee for the phone, though in a plumbing shop it's often the biggest time sink: a burst pipe doesn't send an email. Which AI receptionists exist, what they cost and what the law says is in AI Receptionist: Providers, Pricing, Law. Who goes out on an emergency call isn't the assistant's decision.
Search for AI for plumbers and most of what you'll find is written by software vendors: field service platforms, estimating tools, answering services. They answer which software is out there. They don't answer who in your shop does what with it, who approves the result and where the AI's work stops. That's where adoption usually fails.
An AI employee answers the second question: a written job with a task, an approval step and a limit, running on a model. Frieda, for instance, is built from Claude, fixed rules and an inbox connection.
What doesn't go to the AI
Diagnosis and adjustment on site. What's wrong with the system gets seen and measured by someone in the basement or the attic.
Pricing. Prices come from your numbers, not from a model guessing market rates.
Rebate promises. Whether a customer qualifies for a rebate is decided by the program. No draft quote promises one.
Warranty and complaints. The AI can put the case together; the answer to the customer comes from you.
Customer data without rules. Addresses, photos of bathrooms and basements, phone numbers: which data an AI gets to see, under which agreement, is in AI and Privacy: What the AI Gets to See.
How you put that into practice in your business, we can look at together.
All articles for plumbing and HVAC
The deeper dives on the individual building blocks:
Field service and estimating software with AI features, plus AI receptionists for the phone. More important than the tool is which task gets handed off for good and who approves it.
Can AI write a quote for a water heater or furnace replacement?
As a draft, yes, if it works from your old quotes and a current price list. Anton sets no price without a price source and asks when prices conflict. Measurements, pricing and approval stay with you.
Is there a real-world example?
Yes. The roles run in my own business, and there's a real case from a tax firm, where the road to approval works the same way: AI Agents for Tax Firms: A Real Case.
What does an AI employee cost a plumbing business?
With the task that costs you the most hours and is quick to check, often the quote or the peak-season inbox. The steps from picking the task to approval are in Hiring an AI Employee: The Process.
How to keep going
For one week, note the hours that go into quotes, the inbox, receipts and the phone. The biggest number you can check fastest is your first AI employee. If you'd like to look at it together, book a strategy session: in 20 minutes we look at the one workflow in your business that depends on you the most and work out your first step.
Kevin Welter
Developer, IT architect, author of technical books (Kubernetes, cloud infrastructures) and speaker. Runs his business with an AI workforce of fifteen AI employees and shows solo business owners in his community how to hire their first AI employee.
