The short list
The five, and why these five.
The five AI automations every home service business should adopt are AI call answering for missed and after-hours calls, voice notes that become estimates and invoices, automatic follow-up on open estimates, AI-assisted scheduling, and review requests with maintenance reminders after each job. Each one replaces a task your team does every day, by hand, usually late.
We picked these five with three tests. The task happens daily, not monthly. It follows rules you can write down. And it sits close to revenue: a call answered, an estimate sent, an estimate closed, a job booked, a customer who comes back. Automations that fail those tests tend to cost more to build and babysit than they save.
None of these replace your people. Each takes the repetitive part of a job off their plate and leaves the judgment calls with them. You don’t need all five at once, either. Most shops should start with one, run it for a few weeks and add the next.
Automation 1
AI call answering for missed and after-hours calls.
The problem. Calls come in while your techs are under a sink, on a roof or driving, and again at 9 p.m. when a pipe bursts. A caller who reaches voicemail can dial the next company in seconds, and many do. Every one of those calls cost you money to generate, through ads, SEO or years of reputation.
What the automation does. An AI receptionist answers every call in a natural voice, any hour. It answers common questions from your own knowledge base, qualifies the caller, books the job into your calendar, sends a follow-up text and transfers the call to a real person when it should. Because it handles unlimited concurrent calls, a storm-day spike never becomes a busy signal.
What it costs. Lithium’s AI receptionist is $179 a month, fully managed, with no contract and no setup fee. Call (503) 766-5964 to hear it answer our own line.
What it can do with custom work. One plumbing company we worked with found that 60% of its after-hours emergency calls were going to voicemail. We built a custom AI phone system that triages how urgent each call is, texts the on-call tech for emergencies and books non-emergency jobs straight into the schedule. The result was 3x more leads. That urgency triage and on-call texting were custom automation, built on top of what the receptionist does.
Watch out for. Write the transfer rules before launch: which calls go to a person, which person, and what happens if nobody picks up. The missed calls guide covers the setup in detail.
Automation 2
Voice notes that become estimates and invoices.
The problem. Paperwork happens at night. A tech finishes the last call, then someone types up the estimate or invoice from memory, scribbled notes and photos. Estimates go out a day late, and line items get missed.
What the automation does. The tech records a one-minute voice note in the truck: what they found, what they recommend, the parts and the hours. The automation transcribes it, pulls out the line items, prices them from your own price book and builds a draft estimate or invoice in the software you already use. Someone reviews the draft, fixes anything wrong and sends it.
What you need. A current price book, your estimate and invoice templates, and field-service or accounting software that accepts outside connections. Most of the popular ones do.
Watch out for. Numbers and measurements. Speech-to-text can mishear “fifteen” as “fifty,” so every draft gets a human check before it reaches a customer. Never let it send on its own. The voice notes to estimates guide walks through the full workflow.
Automation 3
Automatic follow-up on open estimates.
The problem. An estimate goes out and the customer goes quiet. Nobody follows up, because the office is busy and calling feels pushy. A good share of those estimates were winnable: the customer had a question, got distracted, or was waiting for a nudge.
What the automation does. When an estimate sits unanswered, the automation sends a short series of texts and emails over the following days, written from the estimate itself: the work, the price, the warranty, the next open slot. AI drafts each message to match the job, so it reads like your office wrote it. The moment the customer replies, the sequence stops and a person takes over.
Proof it works. We built a lead qualification bot for a real estate agency, and it converts at 52%. Different industry, same principle: fast follow-up that fits the inquiry, with a human at the finish line.
Watch out for. Consent and stop rules. Text only people who gave you their number for the job, honor every opt-out, and stop the moment someone answers. The lead follow-up guide has sample sequences.
Automation 4
AI-assisted scheduling and dispatch prep.
The problem. Booking a job takes a phone call, a look at the calendar, a guess at how long it will take and a text to confirm. Multiply that by every call, then add reschedules, and your dispatcher spends the day on the phone instead of running the board.
What the automation does. AI books jobs into the arrival windows you set, by job type, expected length and service zone. It flags emergencies for a person, sends confirmations and reminders, and handles simple reschedules. For the dispatcher, it prepares each job with a clean summary: the problem in the customer’s words, the address, the equipment and any notes from past visits.
Who decides what. The dispatcher keeps the final call on which tech goes where. AI handles the booking and the paperwork around it; a person handles the judgment.
What it costs. The AI receptionist books appointments as part of its $179 monthly service. Custom scheduling logic, like zone rules or skill matching from your field-service software, is scoped as custom work. The AI scheduling guide explains how to set the rules.
Automation 5
Review requests and maintenance reminders.
The problem. Happy customers rarely leave a review unless you ask, and asking by hand after every job doesn’t happen. The same goes for maintenance: the furnace tune-up, the water heater flush and the annual panel check all depend on someone remembering.
What the automation does. When a job is marked complete, the customer gets a text a few hours later with a direct link to your Google review form, plus one reminder if they don’t respond. Separately, customers get maintenance reminders on the schedule you set, like a fall tune-up for every furnace you installed. AI can also draft replies to new reviews for you to approve.
Follow Google’s rules. Ask every customer, not only the ones you think are happy. Never offer a discount or gift for a review, and never steer unhappy customers away from posting. Google prohibits all three, and its contribution policy spells it out. If a customer reports a problem in reply, the automation should alert a person right away; that’s good service, not a filter.
Our guide to getting more Google reviews covers wording and timing.
Next in line
Two more worth a look once the first five run.
Equipment manual search for techs. Load your manuals, install guides and service bulletins into a private AI search tool. A tech asks a question on site and gets the answer with the page it came from. It’s a bigger build than the five above, and it pays off most for shops that service many brands and models. The equipment manuals guide explains how it works.
Parts and truck stock records. Parts used on each job can be pulled from the same voice notes and invoices that automation 2 produces, then checked against what each truck should carry. The automation prepares a restock list for someone to approve. It won’t forecast demand on its own, and nobody should let it order parts without a person signing off.
Where to start
How to pick your first automation and roll it out.
Pick by pain, and measure it first. Count last month’s missed calls in your phone system. Ask the office how many hours a week go to typing estimates and invoices. Count the estimates sent in the last 90 days that never got an answer. Look at how many reviews you got compared with how many jobs you finished. The biggest number tells you where to start.
- 1
Pick one task
The one with the biggest number from your count above.
- 2
Write the rules
What starts it, what it does, what it never does, and who takes over when it’s unsure.
- 3
Run a pilot
Two to four weeks on real work, with someone checking every output.
- 4
Measure, then add
Compare against your starting count. Keep it, fix it or drop it, then pick the next one.
Lithium’s AI receptionist is $179 a month and goes live in days. Every other automation on this list is custom AI automation , scoped after a free discovery call; most custom builds take one to two weeks. Bring your counts to the call. They make it quick to see which automation to build first.
