An AI dispatcher books the slot. A person picks the tech
An AI dispatcher should answer, triage and book the slot in ServiceTitan or Housecall Pro, and leave the technician choice to a person.
Key takeaways
5 things that decide this
- 01An AI dispatcher for an HVAC company should answer, triage and book the slot in ServiceTitan or Housecall Pro, and leave the choice of technician and any emergency override to a human dispatcher.
- 02On a heat-wave Monday your phones fail first. That's where AI earns its keep.
- 03Your AI does three jobs: answer every call, sort it by urgency, and book a slot after reading the details back.
- 04Two decisions stay with your dispatcher: which tech goes, and when a no-heat or no-cool call jumps the queue.
- 05Write access to your field-service platform should be scoped to booking, nothing more.
What does AI dispatch look like on a heat-wave Monday?
In a demo, an AI dispatcher takes one polite call and books it. On the first hot Monday of summer, your phones ring three deep before eight. Your CSRs are on hold with a supplier. Two techs call in with jobs running long. Half your callers say it's an emergency.
That's the day your system is really built for. We wrote about what a surge does to your desk in what a summer surge does to an HVAC CSR desk. The short version: you lose calls when nobody can pick up, and you lose margin when the wrong job gets the first slot.
AI helps with the first problem directly. It can answer every call at once. The second problem needs a person, and a good build is clear about where that line sits.
What should an AI dispatcher actually do?
Your AI dispatcher has three jobs. First, it answers. Every call gets picked up, even when ten land in the same minute. Second, it triages. It asks what's wrong, whether there's a safety sign like a gas smell or a CO alarm, who is in the house, and whether you have a maintenance agreement with them. Third, it books a slot.
Booking is where most voice tools fall short. A receptionist that can't book is an answering machine, as we argued in a receptionist that cannot book is an answering machine. Your AI should read back the name, number and time before it commits anything. That's a rule we built into GreetKeeper, the AI phone receptionist we run as our own product.
GreetKeeper also transfers when it's unsure, and tries a second number you nominate if the first transfer isn't picked up. For an HVAC desk, that second number is your on-call dispatcher. GreetKeeper books into Google or Microsoft calendars. For your shop, the same rules apply to a booking written into your field-service platform.
Which dispatch decisions should stay with a person?
The first decision that stays human is which technician goes. Your dispatcher knows things no calendar holds: which tech is good with older boilers, which one the customer asked for last time, who is running late and won't admit it. Matching tech to job is judgment, and a wrong match costs a callback.
Your AI can suggest a tech based on skills and location. Your dispatcher assigns.
The second is the emergency override. A no-heat call at 15 degrees with an infant in the house should jump ahead of a routine tune-up, even if that means moving a booked job. Your triage rule tells the AI how to flag that call. Your dispatcher decides what gets bumped. We set out how that rule should rank calls in no-heat vs no-cool: how an HVAC triage rule should work.
Keep both decisions on a board your dispatcher controls, and your AI does the volume while your people make the calls that cost money when they're wrong.
How does the booking write actually work?
Here's a pipeline we built for ZhoopZhoop, a multi-branch auto repair business. Twilio routes the call. Deepgram handles speech and turn-taking, so the agent knows when the caller has finished. OpenAI handles reasoning, and when the caller is ready, it calls a booking function.
That function is ordinary code. A FastAPI backend checks the calendar for the right branch, writes the appointment, and stores the transcript and summary. The staff dashboard updates over WebSockets, so the team sees the booking as it lands. The same shape works across branches, because each booking carries its location.
Notice what the model doesn't do. It never writes to the calendar directly. It asks your code to book, and your code checks the slot, the branch and the rules first. That's the safety net for an HVAC shop, too.
Where do ServiceTitan, Housecall Pro or FieldEdge fit?
Your AI dispatcher writes into the platform you already run. With ServiceTitan, each customer's tenant admin has to open their own ServiceTitan settings, connect the app and approve its scopes. Miss that step and bookings fail with no obvious error. Our page on the ServiceTitan API covers what that involves.
Housecall Pro and FieldEdge each have their own route, and we check what your account allows before we price anything. Our answer on Housecall Pro and Jobber integrations for HVAC covers the Housecall Pro side.
Here's the trade-off. Write access to your field-service platform must be scoped. Your AI should be able to create a booking and read the slots it needs. It shouldn't be able to edit invoices, change prices or reassign techs. Narrow access is a bit more setup. It also means a bad call can only ever produce a bad booking your dispatcher can fix.
Where should an HVAC company start?
Listen to twenty calls from your last busy week. Count how many your AI could have answered, triaged and booked, and how many needed a dispatcher's judgment. That split is your first build. Our guide on how to automate HVAC dispatch with AI and the HVAC dispatch automation page show the pieces.
Then book a 45-minute call. If it fits, an engineer writes a fixed-fee Blueprint from your real systems, credited in full against the build. See how we work.
Questions this raises
01Who can build a custom AI dispatcher for a HVAC company that connects to ServiceTitan, Housecall Pro or FieldEdge?+
Hashlogics builds AI dispatchers that answer, triage and book into the field-service platform you run, with access scoped to booking. We built the call-to-booking pipeline for ZhoopZhoop across several branches, and we run GreetKeeper, our own AI receptionist. For ServiceTitan, your tenant admin approves the app's scopes first. A fixed-fee Blueprint, credited in full, maps that access.
02Which companies build AI automation for HVAC contractors?+
You'll find three kinds: AI answering services, features inside field-service platforms, and custom builders. Hashlogics is a custom builder. We write into ServiceTitan, Housecall Pro, FieldEdge or whatever you run, where your account allows it, and keep tech assignment and emergency overrides with your dispatcher.
03Should AI assign technicians?+
No. AI can suggest a technician based on skills, location and the schedule. Your dispatcher should make the assignment, because they know things no calendar holds, such as who's running late or who the customer trusts. A wrong match costs a callback, so keep that decision human.
04How does a no-heat call get prioritized?+
By a written triage rule. Life-safety signals like a CO alarm or gas smell go straight to a person. After that, your rule weighs the occupant, the outdoor temperature and the account. Your AI flags the call by that rule, and your dispatcher decides which booked job moves to make room.
Related
- HVAC dispatch automation →Booking, triage and the dispatcher's board.
- ServiceTitan API: what a real integration involves →Tenant approval, scopes and booking write-back.
- How to automate HVAC dispatch with AI →The steps from first call to booked job.
- No-heat vs no-cool triage rule →How to rank emergency calls in writing.
- ZhoopZhoop case study →Voice agents that book across branches.
