The category guide
The category guide: AI booking, explained without the hype
What "AI booking" actually means
A customer decides they need a plumber, an HVAC tech, a roofer. Then one of three things happens: they call, they visit a website, or — the new path — they ask an AI assistant to find and book someone. "AI booking" products automate the moment where that intent becomes a job on your schedule. They differ in which door they stand at:
- AI receptionists stand at the phone door. They answer calls (and sometimes texts and web chats), talk to the customer, and book into your software. Housecall Pro CSR AI, Jobber AI Receptionist, Goodcall and Rosie AI live here.
- Assistant-booking services stand at the AI-assistant door. They make your business appear — with real availability and real booking — when a customer asks ChatGPT or another assistant to book a service. Canopy lives here, and so far it is a category of mostly one.
- Platform AI add-ons are the same receptionist technology sold inside a field-service platform. Same door as the first group; the difference is the bundle and the lock-in.
Who needs which
You lose jobs to the voicemail. Industry rule of thumb every owner already knows from their own call log: after-hours and overflow calls are where the money leaks. An AI receptionist is the direct fix, and the built-in option on your current platform is usually the fastest path.
Your customers are starting to arrive through AI assistants. Ask ChatGPT today for the best HVAC company in your town and watch what it answers — in Copp Oak's own audit of 11,772 AI answers about home-service companies, the assistants named plenty of companies but could almost never complete a booking with one. Being named is marketing; being bookable is revenue. That gap is what assistant-booking closes.
You are a one- or two-truck shop. You are the office. Either product buys you an office; the receptionist also buys you sleep.
How the money works
- Per minute (most AI receptionists and all human services): predictable per call, punishing on long calls and surge days. Human answering commonly runs several dollars a minute fully loaded; AI receptionists typically well under that, bundled or metered.
- Per seat / per month (platform add-ons): simple, but you pay in slow months too.
- Per lead (marketplaces): you pay for contact details, including the ones that never become jobs.
- Per job that shows up (Canopy's model): you pay when a booked customer actually becomes a job. It aligns the bill with the outcome, and it makes slow months cheap. The honest catch: it only pays off if assistant-driven customers exist in your market — a growing flow, not yet the main river.
The five buying mistakes
- Buying a receptionist that can't book. A friendly AI that "takes a message" is a voicemail with better manners. Require a confirmed job in your own system, with a named test: book a job yourself through it before you sign.
- Switching platforms for an add-on. Migrating your whole operation to Housecall Pro or Jobber for their AI is a six-month decision wearing a six-minute purchase. If you were switching anyway, fine.
- Ignoring the assistant channel entirely. Your competitors are being named by ChatGPT today; the ones who become bookable there first collect the early flow.
- Judging on the demo call. Every AI receptionist sounds great on a quiet Tuesday demo. Test it at your worst hour: after-hours, mid-surge, with a difficult customer.
- No kill criteria. Decide in advance what "working" means — answered-call rate, booked jobs, cost per booked job — and review at 30 days.
What to do Monday morning
Call your own business after hours and see what a customer gets. Ask ChatGPT to book a company like yours in your town and see whether anyone — including you — can actually be booked. Those two free tests tell you which door is leaking, and the comparison tells you what plugs it.