AI for plumbers: quotes, on-site jobs and customers

The AI for Plumbers makes where the trade wastes time away from the plant: answering to those who write while you have your hands in a siphon, moving from the inspection to a readable estimate, keeping in order the cards that each job leaves behind. Diagnosis, choice of materials and signature remain yours. The assistant sorts, transcribes, prepares drafts and remembers deadlines.

A plumber cannot copy the AI of a shop or a hairdresser, for two reasons. The first: your work produces documents with legal value, such as the declaration of conformity for water, thermal and gas systems and the Plant Book the boilers. The second: a part of the requests you receive is a real emergency, and there a wrong automatic response does not just cause a customer to lose.

Bath flooded at seven o'clock in the evening: what the automatic message must answer

Requests that come to a plumber are not all in the same hurry. "I lose the siphon under the sink" can wait in the morning, "the water comes down from the neighbor's ceiling" no. The first job of an AI assistant on WhatsApp or on the site module is this sorting: read the message, ask one or two targeted questions (where does the water come from? Can you close the general tap?) and put the request in the right lane. Urgency to be passed to you immediately, intervention to be fixed in the next few days, estimate to be programmed.

You write the rules once. What situations are ongoing damage to you, what to say to the customer while waiting ("close the general tap and send me a picture of the point where the water comes out"), in which areas and until what time you accept evening calls. The assistant does not improvise technical advice: he repeats the instructions you have prepared and sends you the request already summarized, with address, photo and number to call back. How to connect an assistant to the company number we explain in the guide to WhatsApp Business Automation. If instead you lose mainly phone calls, the starting point is a switchboard with automatic response.

Then there is a contractual detail that the confirmation message helps manage. For contracts concluded outside business premises, the Consumer Code excludes the right of withdrawal when the customer asks for a visit for urgent repair or maintenance work, but leaves it valid for the extra services provided during the visit and for goods other than the necessary spare parts (D.Lgs. 206/2005, art. 59, letter h). An assistant who records in writing what the client asked for, and who keeps the proposal to also change the toilet box separate, leaves you a clean trail. The confirmation texts have your counselor reread them.

From inspection to estimate: a voice note, three photos and a draft to approve

The estimate of a bathroom to be redone arises standing, in a corridor, with the meter in hand. Then it waits for the evening to become a document. With the AI the passage is shortened: in the van you record a voice note of a couple of minutes ("second floor without elevator, shower tray to be replaced with one flush with the floor, cast iron drain column to be verified, thermostatic mixer the customer buys") and take photos of the attacks, the column and the boiler plate.

The assistant transcribes the note, compares it with the items on your list and prepares a draft with workings, materials and exclusions ("supply of the mixer to be paid by the customer"). Where he doesn't find a price he doesn't invent it: leave the line empty and report it to you, along with the open questions, for example who takes away the old plate and if they serve wall works. The mechanism, from free text to structured draft, is the same as we describe in automatic quotes from a message in free text.

Two rules keep the system reliable. The figures come only from your price list and the prices you have already applied, never from a model estimate. And the quote only starts after you've opened it and approved it, even from the phone. On small and repetitive jobs, such as replacing a hose or disposing of a drain, the draft is almost ready; on a new plant or a fault to be understood remains an orderly reminder for the inspection.

Declaration of conformity, booklet, control report: the papers that AI can prepare

Here a plumber takes advantage: he already knows what documents he has to produce, and the norm says who signs them and when. D.M. 37/2008 covers water and sanitation, heating and air conditioning and gas (Art. 1); DPR 74/2013 regulates control and maintenance of thermal systems. The table lines up the most frequent cards and the boundary between what the assistant can prepare and what is left to you.

One distinction is worth more than the others: AI compiles, does not attest. The declaration of conformity is an act of the enterprise and its technical manager, and no assistant can know if the leak proof has really been made. He can, however, avoid you the night he spent copying item codes from the bubbles, and he may notice that the pattern is missing before the file starts. How to build such a flow on forms and PDFs you can find it in the article on document automation with AI.

The opposite request also happens. The customer sells house and asks you for "certification" of a plant made twenty years ago, of which no one has the declaration anymore. The assistant can ask the right questions (year of work, documents remaining, type of plant) and explain that for plants performed before the entry into force of the decree there is the declaration of compliance: a professional registered on the register or, for installations that do not require the design of a professional, the technical manager of a qualified company, in both cases with at least 5 years of experience in the field (Art. 7, paragraph 6). If you can sign it you will evaluate it, not the chatbot.

Your customers' boilers are already an agenda

A thermo-hydraulic firm that follows a few hundred boilers has a calendar of work in hand throughout the year, only that it is scattered among booklets, reports and memory of the holder. DPR 74/2013 asks installers and maintainers to declare in writing to the customer which control operations are needed for his plant and how often (Art. 7, paragraph 4). That statement is the data from which a reminder system should start, instead of a generic "every year" equal for everyone.

The flow is described in a few lines. For each system the system knows the date of the last check and the frequency declared; a few weeks before the deadline it sends the customer a message with two or three free dates taken from your schedule; if the customer does not respond try again once, then it puts the name on the list to call. On heating systems above 10 kW, along with maintenance must be done energy efficiency control, and the report ends at the Region (DPR 74/2013, art. 8). In Lombardy the land register is the CURIT, born in 2008 for the installation, maintenance and inspection of thermal systems.

The season does the rest. In the climatic zone And the heating can be turned on from 15 October to 15 April, in the other areas the dates change (DPR 74/2013, art. 4). Filling September with scheduled checks, before calls for the boiler that doesn't start, is the most concrete gain of an agenda that you compile yourself. On how to write reminder messages that do not look like advertisements you will find directions in the guide to AI for customer callbacks.

Two examples: the evening emergency and the company with 400 boilers

The emergency response of a plumber working alone. Think of any night when six messages arrive while you're still on a construction site. Two are slow drains, one asks for a quote to replace the boiler, two want to know how much a destruction costs, one says the water comes down from the neighbor's ceiling. The assistant answers the top five with tomorrow's free dates and questions for the quote; the sixth comes to you as a notification, with address, photos and instructions that the customer has already received. Just call that back.

Those who work on sewers know this type of customer well. For Simo Spurghi, a sewage management company, we have built a site designed for those who have a fault and look from smartphones: number always in sight, called with a touch, a quote form that asks only what is needed, where and how to be contacted. It's a web project, not an AI system. That short form, however, shows where to start: an already ordered request is easier to sort, for a person as for an assistant.

The thermo-hydraulic company with three technicians. Take, as an example, a firm with 400 boilers in maintenance. At the end of August the system lists the plants with the control due in the autumn, groups them by neighborhood, sends messages and fills the agendas. After each check it transcribes the values from the analyzer, updates the history of the booklet and prepares the copy of the report for the cadastre. Human hands are used to measure, check transcription and sign.

Where an assistant errs with a plumber's customers

• Give prices by voice for jobs no one has seen. A boiler replacement changes a lot depending on flue, condensate discharge and location. The assistant can only indicate a band if you have written it, otherwise he proposes the inspection.
• Making remote diagnoses. "It's the three-way valve" said by a chatbot creates expectations that you then pay for. You can ask for the error code on the display and a photo of the tag, to get you prepared.
• Treat the smell of gas as any request. The right answer is the referral to the prompt intervention of the distributor, always, even at three in the night.
• Mix maintenance reminders and promotions. Remembering the control of the boiler to a customer you follow is a service message; offering him a discount on the air conditioner is marketing and requires different bases. The picture is in the guide on AI and GDPR, to be verified with your privacy consultant.
• Leave photos and addresses on any tools. The photos taken inside the houses, addresses and notes of the type "keys leave them to the doorman" are personal data. Choose tools with a clear contract on where they end and who uses them, as we explain in what happens to business data in AI tools.

Where does a three-person thermohydraulic company start?

Choose one of three areas (sorting requests, quotes, planting cards) and prepare it well, with the data you already have.

• Reread a month of WhatsApp chat and mark which requests were urgent, which quotes and which simple information.
• Write down the situations that are urgent for you and the text to send for the smell of gas.
• Put in a sheet the plants you follow: address, generator, date of the last check, frequency declared to the customer.
• Prepare the list of repetitive processes with the prices you really apply.
• Decide who approves what: quotes and documents always you, service messages automatically.
• After a month it counts how many requests were sorted well and how many you had to correct.

If you also work with masons, tilers and other trades on the construction site, the widest picture is in the guide to AI for craftsmen and construction, while the page AI for SMEs gathers the projects best suited to small businesses. If you want to understand which of the three areas to start from in your company, we can look together at a month of chat, the price list and a couple of files and draw the first flow: it is the work we do in the projects of automation of business processes.