The short answer: hand over your receptionist work first. That means answering calls, following up quotes and chasing invoices. Keep the judgment calls, the pricing exceptions and the hard conversations with people.
This guide is for owners of US home service companies with crews and trucks. It covers what AI can do today, what it takes, how to pick who builds it and why these projects die. It also covers how all of this fits with ServiceTitan, Housecall Pro and Jobber.
What can AI actually do for a home service company today?
AI can answer the phone, book jobs into your calendar, reply to texts and follow up on work that went quiet. It is good at repeat tasks with clear rules. It is not good at reading a tense room or deciding what a customer is worth to you.
This is no longer a lab trick. The big job software companies now ship it. ServiceTitan says its AI Virtual Agent handles after-hours, overflow and emergency calls and schedules the job. Housecall Pro says its CSR AI books jobs as calls and chats come in. Jobber says its Receptionist answers calls and texts around the clock and can book the visit during the call.
Those are the makers' own descriptions, so test them on your own calls. Still, the table below shows the work AI can take on today.
| Job | What AI can do today | Where you still need a person |
|---|---|---|
| Inbound calls | Answer, ask the right questions, book into open slots | Upset callers, odd requests, real emergencies |
| New web and ad leads | Reply by text within seconds, offer open times. ServiceTitan's Speed to Lead is one example | Large or unusual jobs |
| Open quotes | Send follow-ups until the customer books or opts out. | Discounts, scope changes, the final sales talk |
| Unpaid invoices | Send reminders and payment links. ServiceTitan's FAQ says its agent can text a secure link for outstanding invoices | Final notices, disputes, write-offs |
| Reviews | Check in after the job and ask for a review. | Replying to an angry one-star review |
| Spanish-speaking callers | ServiceTitan's FAQ says its agent switches to Spanish when the caller speaks it | Anything sensitive |
Other jobs, such as ads, SEO, bookkeeping and recruiting, are also getting AI tools. We would not start there, for the reasons below.
What should I hand over to AI first?
Start with the receptionist work: calls, quotes and invoices. The phone is where new work enters your company, so a gap there costs you the job before you ever see it.
Picture the usual day. The person who could answer the phone is under a sink. The person who could chase a quote is dispatching the next job. Nobody is being lazy. The work just has no one free to do it. We wrote more about how fast HVAC leads go cold, and the same pressure hits any trade with urgent calls.
There are five reasons we would start here.
- The loss is the easiest to count. Your call log shows every missed call. You can compare before and after within weeks.
- The rules can be written down. Service area, hours, job types and open slots are facts. That makes them safe for AI to follow.
- A mistake is cheap to catch. A wrong booking can be moved. A wrong tax filing or a bad hire is much harder to undo.
- It removes the most interruptions. Every call it handles is one your crew or office does not have to stop for.
- The other jobs follow from the call. The same conversation leads to a quote, a booked visit, an invoice and a review. One AI employee with the full history does this better than four tools that cannot see each other.
Some owners feel their biggest pain is quotes, not calls. That does not change the answer, because quote follow-up belongs to the same job. Our post on which AI to build first walks through how a 10-technician shop can rank its own gaps.
How do I know if this is worth doing for my company?
Do the sum with your own numbers. Do not trust ours or anyone else's.
- Pull 30 days of calls from your phone system. Count the ones nobody answered.
- Take out spam, wrong numbers and existing customers who called back. What is left is likely new work.
- Multiply by the share of answered calls you normally book.
- Multiply by your average ticket.
That gives you a monthly figure for the phone alone. The calculator on our homepage uses the same sum. Then list your open quotes and overdue invoices from the last three months and add them up. If those totals are small, you may not need any of this yet, and that is a fine result.
What should stay with people?
Keep anything that needs judgment, a relationship or real money. AI should work inside rules you set, and anything outside those rules should come to a person.
Even the vendors plan for this. Housecall Pro says its CSR AI can "send the call to a number of your choice" when a caller wants a person. Expect some calls to need one.
These should stay with people:
- The scared caller. Someone with a burst pipe needs help fast. Real emergencies should go straight to your on-call tech.
- Disputes. A fight over last month's bill needs someone who can listen and decide.
- Exceptions to your prices. AI can offer what you allow. It should not invent a discount.
- Your best accounts. A commercial customer of ten years may expect a person who knows them.
- Final notices and write-offs. These affect the relationship and your cash, so you decide.
- Hiring, firing and big purchases. These stay with you.
Here is the rule we use for our own AI receptionist. It books from your open slots, inside the rules you set with us. Anything outside those rules that costs money, commits a crew or changes a price comes to you first. Real emergencies go straight to your on-call tech.
Notice what that rule does not say. You do not approve every booking. If you have to check each one, you have not saved any time.
What does it take to get AI working in my company?
It takes your rules written down, access to your systems and a few hours of your time and your office's. The software is the easy part. The hard part is getting what is in your head onto paper.
Here is what any builder, or you on your own, will need:
- Your phone setup. Who answers now, what forwards where, and what happens after hours.
- Your services and service area. What you do, what you turn down and how far you drive.
- Your booking rules. Which job goes to which tech, how long it takes and how many you can run in a day.
- Your prices. Diagnostic fees, trip charges and what may be quoted over the phone.
- Your on-call list. Who gets the emergency at 2 a.m., and who gets it if they do not pick up.
- Your exceptions. The things only you know, like "we never book that street on Fridays."
- Clean records. Duplicate customers and made-up job types get copied at speed.
How long does it take?
It depends how much is being built. Housecall Pro says you can be live "in minutes" using settings you already have. Being live is not the same as being right, though. Expect getting the rules right to take longer than getting it live.
What about the legal side?
Answering a call that came to you is one thing. Having AI call people is another. The FCC ruled in February 2024 that AI-generated voices count as "artificial" under the Telephone Consumer Protection Act, and that telemarketers need prior express written consent before robocalling. Business texting has its own opt-in rules too. Before any AI calls or texts your old customer list, ask who checked the consent side, and ask your lawyer if you are not sure. Also ask whether your AI should say it is AI. Housecall Pro leaves that choice to you, so it is a decision you will have to make.
How do I choose who builds it?
Choose the one who can answer plain questions about your calls, your software and what happens when it breaks. You have three real options: the AI inside your job software, a software product made for the trades, or a team that builds and runs it for you.
| Option | Fits best when | Watch out for |
|---|---|---|
| AI inside your job software | You mostly need calls answered and booked, and your work stays in that software | It may stop at the edge of that software. Check usage bills |
| A software product for the trades | You have an office manager or call center lead who will own it and tune it | Someone on your side still has to give it feedback |
| A team that builds and runs it | You have no one free to own it and want it done for you | You depend on that team. Ask for terms and how you leave |
| Building it yourself | You have a technical person with time to spare | Most owner-run shops do not, and upkeep never stops |
Whoever you talk to, ask these questions. Vague answers tell you plenty.
- Can I call it right now and try to break it?
- What exactly does it write into my software, and what do my people still type?
- What happens on a gas smell call at 2 a.m.? Show me the path.
- What does it do when it does not know the answer?
- Who fixes it when it books the wrong job, and how fast?
- What number will you hold it to, and what happens if it misses?
- Who improves it after launch? Is that me or you?
- What do I pay for calls that do not book?
- How long am I tied in, and what happens to my data if I leave?
- Can I talk to a company my size that uses it? If not, why not?
That last one matters. A builder with no references is a real risk. You should know that before you sign, and it should be reflected in the terms they offer you.
Why do AI projects fail?
Most fail because nobody owns them after launch, they do not connect to the real work, or no one set a clear goal. The technology is rarely the main reason.
The research is about big companies, not shops with ten trucks, so read it as a warning and not a forecast. Gartner predicted in June 2025 that over 40% of agentic AI projects will be canceled by the end of 2027, due to escalating costs, unclear business value or inadequate risk controls. A 2025 MIT NANDA report found 95% of organizations were getting zero return from their generative AI spend. It calls its own findings preliminary, and it says its figures are directional and come from interviews. It blames tools that "do not retain feedback, adapt to context, or improve over time."
The same report says external partnerships reached deployment about 67% of the time, against about 33% for tools built in-house. Those are self-reported numbers, and we benefit from that finding, so weigh it with care.
The report also says back-office automation often pays better than front-office tools at large companies. That is worth knowing, since we are telling you to start with the phone. Those companies were cutting spend on outside firms. In an owner-run shop, the back office and the front office are often the same person. That is why quotes and invoices belong in the same job as the calls.
In a home service company, we would expect failure to look like this:
- No one owns it. It goes live, it gets a call wrong, nobody fixes it, and your team starts working around it.
- It sits next to your software. Someone retypes every booking, so you have traded one job for another.
- The rules stayed in your head. It books the wrong job type, and your dispatcher loses trust in a week.
- There is no number. Six months later, no one can say whether it paid for itself.
- It was too wide on day one. You tried calls, ads, social media and bookkeeping at once, and none got proper attention.
- No one tuned it. Your prices, crews and seasons change. An AI that does not change with them gets worse every month.
The fix is dull and it works: one job, one owner, one number, reviewed every week.
How does this fit with ServiceTitan, Housecall Pro and Jobber?
All three now sell their own AI for calls, so check that first. If it does the job, you may not need anything else. Look further when your work crosses into other tools or you need someone to run it for you.
| Software | What it offers for calls | Cost, as of November 2026 | Worth checking |
|---|---|---|---|
| ServiceTitan | AI Virtual Agent, listed as available in the core subscription in its Fall 2025 release | Usage-based. "All calls, regardless of their outcome or the caller's intent, are included in your bill." $2.75 per call on its Voice Agent page | Outbound agents are for ServiceTitan Max customers |
| Housecall Pro | CSR AI for calls and chats. Its page lists text answering as coming soon and "in Alpha" | An optional add-on. We did not find a public price for it | If a caller asks for a human, it sends the call to a number you pick |
| Jobber | Receptionist for calls and texts, with urgent keywords you choose | Included on the Plus plan. On select plans, a $29 monthly add-on with 30 conversations, then $0.79 each | Online booking must be on for it to reschedule |
Prices and features change often, so confirm them with the vendor. We go deeper on one of these in what ServiceTitan's own AI covers.
Built-in AI is often the simple choice. It already sits on your customer records and calendar. You should still ask these questions about it:
- Does it call back on quotes that went quiet, or only answer when someone calls in?
- Does it chase overdue invoices, and who decides when to stop?
- Does it work with what lives outside the job software, like your books, your Google profile or your email?
- Can it follow your odd rules, like sending a 15-year-old unit straight to replacement quoting?
- Who sets it up and keeps it right? If the answer is you or your office, plan time for that.
Also think about whether you need AI or a live service at all. We compared them in AI or a live answering service, and for some calls a person is still the better answer.
Does this change by trade?
The order stays the same, but the rules change. A no-heat call, a wasp nest and a storm-damaged roof need different questions, different urgency and different crews.
- AI employees for HVAC companies: seasonal spikes, no-heat and no-cool calls, maintenance plans. See also our AI answering service for HVAC.
- Plumbing: leaks that cannot wait and clear rules on what counts as an emergency. See our plumbing answering service.
- Electrical: safety questions come first, then sorting service calls from larger quotes.
- Roofing: storm surges, inspections and long gaps between the quote and the yes.
- Garage door: stuck doors and same-day calls, often at the worst time of day.
- Pest control: recurring plans and renewals matter as much as new calls.
- Cleaning: repeat bookings, reschedules and quotes for one-off jobs.
Not ready for a full setup? Missed call text back is the smallest useful step. It reopens the conversation, though someone still has to finish it. You can see every trade on our industries page.
What should I do this week?
Pull your numbers, write your rules and call the AI you are thinking of buying. None of that costs anything, and you will know far more by Friday.
- Monday: Count last month's missed calls, open quotes and overdue invoices.
- Tuesday: Write one page: what you book, what you refuse, who is on call.
- Wednesday: Ask your job software rep what its AI does and what it costs at your call volume.
- Thursday: Call two AI options as a customer. Try a hard call on each.
- Friday: Pick one number you want to move. Do not buy anything that will not be held to it.
Where do we fit in?
We build and run AI employees for home service companies, starting with the receptionist. It works in the software you already run where it allows it. Here is how it works, step by step, and here is every role we have planned. Only the Receptionist is live today.
We have no client case studies to show you yet, so weigh that. What we offer instead is a promise you can hold us to: each AI employee is hired to hit one goal, agreed with you first. If it misses in a month, you do not pay for it that month. It goes live in about 10 days, guaranteed within 30, or your setup fee is refunded. There is a 3-month minimum.
On the call you will talk to a founder, and we will tell you whether we are the right people for this. Book a free assessment call.