AI for home services and field service companies: quotes on WhatsApp and job sheets filled in by voice

If you run a renovation business, a plumbing, electrical or HVAC company, a cleaning, maintenance or gardening firm, your day probably looks like this: the phone rings while you’re up a ladder, customers send photos at 11 pm asking for a quote, and at the end of the month someone in the office is chasing technicians for the job sheets so they can invoice. The good work happens on site. The money leaks out through paperwork, missed calls and job sheets that never come back.

The short answer

• What you can automate: quote requests (WhatsApp, phone and web, at any hour), draft quotes from your own price list, scheduling with reminders and confirmations, job sheets filled in by voice, materials tracking, invoicing from the signed job sheet, follow-ups and reviews after the job, and warranty claims.
• What it saves: office hours spent answering messages, typing up job sheets and chasing them; jobs lost because you replied too late; and materials and hours that never make it onto the invoice.
• What doesn’t change: technicians still decide how the work is done, and a person still approves any quote with prices before it goes out.
• What it costs: an assistant that answers and qualifies requests on WhatsApp or your website from around €1,000; an agent connected to your calendar or job management software €2,000–5,000; a multi-system or voice agent that answers the phone from €5,000. Usage is cents per conversation. A full platform with voice job sheets is custom software and is quoted separately.
• How to start: pick the process that eats the most hours (usually quote requests or job sheets), run it with one crew, and expand once it works.

What you can automate

Process	What AI or automation does	Who benefits
Quote requests	Replies instantly on WhatsApp, web chat or phone, asks for photos, measurements and address, and logs a complete request	Office and sales
Draft quotes	Builds the quote from your line items and your prices; a person reviews and sends it	Whoever prices jobs
Scheduling and routes	Suggests slots by area and technician skills, prevents double bookings and orders the day’s visits	Office and technicians
Reminders and confirmations	Reminds the customer the day before, confirms the visit and tells them when the technician is on the way	Customers and office
Job sheets (work orders)	The technician speaks into the phone and AI fills in the job sheet: work done, materials, hours, photos and customer signature	Technicians and office
Materials and stock	Deducts materials logged on job sheets and flags when to reorder	Stores and purchasing
Invoicing	Prepares the invoice from the signed job sheet, ready to review in your accounting software	Admin
Follow-ups and reviews	Asks how the job went, requests a review if the customer is happy, alerts a person if not	Owner and marketing
Incidents and warranty claims	Collects the complaint with photos, links it to the original job and assigns it	Office and technicians
Recurring maintenance	Reminds customers about annual services (boilers, air conditioning, cleaning or gardening contracts) and proposes a date	Customers and sales

Quote requests answered around the clock

Before: a customer messages on Saturday afternoon: “Hi, I’d like a quote to swap my bath for a walk-in shower.” You read it on Monday, reply asking for photos, they send them on Tuesday, a measurement is missing, you ask again… and by then they’ve booked whoever replied first. Phone calls are worse: if you’re on a job, the call is simply lost.

After: an AI agent replies straight away on WhatsApp, your website or the phone. It asks what your best salesperson would ask: what needs doing, rough measurements, photos of the bathroom, address or postcode, how urgent it is and when a visit would suit. If you don’t cover the area or don’t do that kind of work, it says so politely. Once the request is complete, it lands in your CRM or job management software, neatly organised, and if it fits, the agent offers a slot for the site visit.

For a plumber or an HVAC company, the same agent separates emergencies (a leak, a boiler that won’t fire up in January) from jobs that can wait, and only wakes the on-call technician when it has to. That’s what we mean by an agent; we explain it in what is an AI agent.

Draft quotes with your prices, not the AI’s

With a complete request, AI drafts the quote using your price list and your line items: strip-out, plumbing, tiling, shower tray, screen, labour. Whoever prices the job checks it, adjusts what the photos can’t show and sends it. On a large renovation or installation, the site visit is still needed; what goes away is the time spent typing notes up and building the document.

The rule here is simple: AI never makes up a price. It only uses items from your price list, and anything that doesn’t fit is flagged for a person. To decide what can be sent automatically and what needs a review, we can use decision models such as Jev, from TypeSafe, which returns a probability rather than text: a reply to “what time are you coming tomorrow?” can go out on its own; a quote with prices or a complaint always goes to a person.

Scheduling, routes and confirmations

Before: the schedule lives in one person’s head, a shared calendar and each technician’s WhatsApp. There’s dead time driving from one side of town to the other, now and then two customers get the same slot, and people ring to ask whether anyone is actually coming.

After: once the customer accepts, the system suggests slots based on area, estimated job length and who has the right skills (not every technician does gas or refrigeration). It groups visits by area so the day’s route makes sense. The day before, the customer gets a reminder with a button to confirm or reschedule, and on the day, a message when the technician sets off. If someone cancels, the slot is offered to the next request.

Job sheets filled in by voice: the technician talks, AI does the paperwork

This is the change technicians notice most, so it deserves its own walkthrough.

Before: the job sheet is a paper pad in the van, a photo of a paper pad sent on WhatsApp, or an app with fifteen fields nobody fills in properly with dirty hands. It arrives late, incomplete or not at all. No job sheet, no invoice, or an invoice without the materials used and the extra hours.

After: when the job is done, the technician presses one button and talks, just like sending a voice message to the office: “Mrs Lopez, 12 High Street. Replaced the kitchen mixer tap and the flexi hose, checked the stopcock, it’s weeping a bit, I’d recommend replacing it. Used our usual mixer and two 40 cm flexi hoses. Hour and a half.” AI transcribes the note and fills in the digital job sheet:

• Work done, written up clearly for the customer
• Materials used, matched to your catalogue and taken off stock
• Hours, with arrival and departure times if the phone records them
• Recommendations (the stopcock) turned into a new quote opportunity
• Photos, before and after
• Customer signature on the phone screen

The technician checks the summary and sends it. The customer gets a copy by email or WhatsApp, and the office has everything it needs to invoice, without ringing the technician to ask whether the job is finished. At the end of the day it gets a summary: jobs closed, job sheets pending, materials used and open incidents. It works the same way for cleaning companies (areas done, issues, products used) or maintenance contracts (inspection done, readings, parts replaced).

This isn’t theory. In the platform we built for a gardening company with 200+ gardeners across 40+ hotels, gardeners record a voice note and AI fills in the incident (description, zone and urgency), and every task is closed with a before and after photo.

Materials and stock

When materials come from job sheets, stock updates itself. AI recognises “two 40 cm flexi hoses” or “a bag of flexible tile adhesive” and matches them to the right item in your catalogue; when it isn’t sure, it asks the technician there and then. You know what each job used and when to reorder.

Invoicing from the job sheet

With the signed job sheet, the invoice is prepared in your accounting software: hours, call-out, materials at your prices and a description of the work. Admin reviews and issues it. The automation passes the data to the software you already use (QuickBooks, Xero, Sage or, in Spain, a Verifactu-ready invoicing program), and that software issues the invoice. We don’t recommend producing invoices outside your accounting system.

Follow-ups, reviews and recurring maintenance

A few days after the job, every customer gets a message thanking them, asking how it went and inviting them to leave a Google review with a direct link. Everyone gets the same invitation, because Google doesn’t allow asking only happy customers. If someone says something went wrong, a person is also alerted to call them and fix it. See our review and reputation management service.

For HVAC, boilers or cleaning and gardening contracts, the system reminds customers about their annual service and proposes a date. It’s recurring revenue many companies lose simply because nobody remembers to call.

Incidents and warranty claims

When a customer writes “the shower tray you fitted is leaking”, the agent asks for photos, finds the original job and its job sheet (date, technician, materials) and opens the claim with all of that attached. Whoever handles it can see straight away whether it’s under warranty, and the customer gets an answer in minutes instead of waiting for someone to find the paperwork.

What a typical rollout looks like

You don’t need to change everything at once. A typical rollout goes like this:

1. Discovery. We look at how jobs come in, which software you use (calendar, job management, accounting) and where hours or money are lost. Usually the answer is obvious: quote requests or job sheets.
2. First process. If requests are the bottleneck, we start with the WhatsApp and web agent that answers and qualifies them, with a person reviewing what goes out. If job sheets are the problem, we start with voice job sheets, tested with one or two technicians who are keen to try it.
3. Tuning on real conversations. We save and evaluate every conversation and every job sheet, fix what goes wrong and test each change against real cases before switching it on for everyone.
4. Next process. Once requests are handled, we connect scheduling; once job sheets work, invoicing and stock. Each step builds on the last.

What still needs a person: site visits for complex jobs, approving any quote with prices, complaints, warranty decisions and any sensitive conversation with a customer. AI takes away the work around those tasks so people have time to do them properly.

The risks of doing it alone

Trying out a chatbot or a job sheet app is easy. Making it work in your company, with your technicians and your customers, for years, is something else. These are the failures we see most often in this sector:

• An app technicians won’t use. If filling in the job sheet takes longer than paper, they’ll go back to paper or WhatsApp, and you’ll be paying for an empty tool. It has to be designed for the phone, with few taps and voice, and tested with technicians before it’s rolled out.
• Quotes with the wrong prices. An AI model without limits can invent a plausible price. If that reaches the customer, you either lose margin or lose the customer. That’s why prices only come from your price list and a person approves them.
• Double bookings. An agent booking against a calendar that isn’t in sync with the technicians’ gives two customers the same slot. The one left waiting at home doesn’t call again.
• Customers’ data and their homes. You handle addresses, phone numbers, photos of the inside of people’s homes, sometimes door or alarm codes, and know when a house will be empty. That’s personal data under GDPR, and information that shouldn’t end up in an AI service that keeps it or trains on it, or on the personal phone of a technician who leaves. We cover this in our guide to EU data residency and GDPR for AI.
• Depending on a generic platform that charges per technician. Many field service tools charge per user per month. With five technicians that’s manageable; with fifty, the bill grows every time you hire, and your way of working bends to the software instead of the other way round. We compare the numbers on our custom software with AI page.
• WhatsApp connected the wrong way. There are two ways to connect an agent to WhatsApp, and the wrong choice can mean paying for every message for no reason, or putting the company number at risk with a use it isn’t suited to. We explain both in Evolution API vs the WhatsApp Business API.
• Failures nobody notices. An agent that stops answering over a weekend, or an integration that silently stops sending job sheets to invoicing, can cost you weeks of unbilled work before anyone spots it.

A team that has done this before knows where it breaks and what to monitor once it’s live.

Want to know what you’d automate first? Book a free call and we’ll look at it with your numbers.

How we do it at Aitomatik

At Aitomatik we’re an AI and automation agency based in Spain, working in English and Spanish with clients in Spain, Europe and beyond. What’s described above is what we’d build for a renovation, installation, cleaning or maintenance company, and these are the projects closest to it:

• AI management platform for a hotel gardening company. 200+ gardeners across 40+ hotels: voice notes turned into incidents by AI, tasks closed with photo proof, attendance from card readers with the hotel alerted within 5 minutes if someone hasn’t arrived, and AI-written monthly reports. It’s exactly the kind of platform a service company with crews in the field needs.
• AI roadmap for a multi-company group. Among other processes, around 100 quote requests a day drafted automatically, and more than 1,600 hours a year freed across eight departments.
• WhatsApp AI agent for a domestic staffing agency. An agent that handles conversations on WhatsApp through Evolution API with no per-message cost: the same foundation as an agent that takes quote requests.

How we work:

• We design for the technician, not the office. Phone first, voice, few taps. We test it with your crew before rolling it out.
• We save and evaluate every conversation of the agents we build, and test every change against real conversations before it goes live.
• Data in Europe. For EU clients, we use models processed in Europe, with zero data retention and no training on your data, plus a backup model from a different provider, also in the EU. International clients aren’t limited to EU models.
• We connect to what you already use (calendar, job management, accounting, WhatsApp) with process automation in n8n (self-hosted on your own server) or custom code, or build the whole platform as custom software, with no per-user fees. The code is usually yours; we agree it per project.

Agent pricing is broken down in how much does an AI agent cost. You’ll find more in our guides and every service in services.

Frequently asked questions

What AI do home service companies use? Mostly three things: an AI receptionist or agent that answers calls and WhatsApp or web messages and books jobs, AI that drafts quotes and follow-up messages, and AI that turns technicians’ voice notes and photos into job sheets and reports. The best results come when these are connected to your scheduling and invoicing, not used as separate tools.

Can AI write quotes or estimates for contractors? It can draft them from photos, measurements and your price list, which saves a lot of time on repeat jobs. On larger renovations a site visit is still needed, and any quote with prices should be checked by a person before it’s sent.

Do my technicians need to learn a complicated new app? They shouldn’t. The best field tools ask for as little as possible: a button to record a voice note, a photo and the customer’s signature. If it takes longer than paper, technicians won’t use it, so we test it with them first.

Is a free job sheet app enough? If all you need is to record jobs, a cheap or free job sheet app may be enough. Something custom makes sense when you want the job sheet connected to quotes, WhatsApp, stock and invoicing, or when you have enough technicians that per-user fees add up.

Do I have to replace my field service management software? Usually not. We connect AI to the calendar, job management and accounting software you already use. We only suggest a custom platform when your current tools don’t fit the way your team works or cost too much per technician.

How much does AI cost for a small field service business? An assistant that answers and qualifies requests on WhatsApp or your website starts from around €1,000; an agent connected to your calendar or job management software costs €2,000–5,000; multi-system or voice agents start from €5,000. Usage costs cents per conversation. A full platform is quoted after the discovery call.

Let’s talk about your company

Tell us how jobs come in, how many technicians you have and which software you use, and we’ll tell you what we’d automate first and what it would cost. See our AI customer service and custom software with AI services, or book a free call.

Indicative prices for SME projects as of October 2026. Last reviewed: 15 October 2026.