AI service scheduling for car dealerships: plate read-back to a booked bay

What it is

A service-scheduling agent answers the service drive and turns the call into an appointment the advisors can work from. It discloses that it is an automated assistant, captures the service wanted, the plate or VIN, the mileage and whether a loaner is needed, reads the plate back character by character with words for letters, finds the vehicle in the DMS, mentions what the system already knows (service due, declined work, an open recall), offers only days with bay capacity and a loaner free, assigns the advisor, reads the summary back and writes the booking. It quotes only prices from the approved menu. For a warning light it books a diagnostic, states the diagnostic fee and says the advisor will quote after inspection.

The call is two to four minutes. The hard parts are identification and the pricing boundary. A plate heard wrong books the wrong car; a price offered carelessly becomes a promise the advisor has to walk back. Both are tested in the demo.

Single stores buy this as an AI receptionist for the service department or AI service scheduling. Dealer groups buy it as service-drive automation across rooftops with one scheduler.

Who buys it

• Service directors and fixed-operations managers watching Monday morning calls ring out while advisors write up cars in the lane.
• BDC managers whose inbound service queue competes with outbound work for the same seats.
• Dealer-group operations leads standardising booking rules, menus and loaner policy across stores.

Budget owner: the service director or general manager at a store; the fixed-operations or BDC director at a group, with the scheduler and DMS administrators signing off on the integration.

KPIs

Count your missed and held service calls for one week, by hour; a DMS vendor's 2025 survey found four in ten service customers hit a significant phone problem and an average hold over eight minutes, and your Monday morning is probably worse. That is the baseline. Then track the strip above: answer rate at all hours, vehicle match rate, booking containment, open recalls offered and attached, pricing-boundary violations (which should be zero), and voice-to-voice latency on the turns that look up the vehicle or write the booking.

Two measurement traps. Containment has to mean an appointment with services, advisor, loaner or transport and drop-off or wait in the scheduler; a calendar entry the advisor re-keys is not contained. And the pricing audit needs a recorded battery of cost questions, because the boundary that holds for "how much is an oil change" often fails for "roughly what does a brake job run".

Demo script

Run the full script from the demo guide. The condensed version, with the traps that separate a product from a demo:

1. Greeting and disclosure. Call the service line. Pass: dealership name, a plain statement that this is an automated assistant, and an offer to book or answer a question. Fail: the disclosure comes after it asks for the plate.
2. Four facts in one turn. Say "I need the sixty-thousand service, plate seven A B C one two three, it's on fifty-nine thousand and change, and I'll need a loaner". Pass: it captures service, plate, mileage and loaner. Fail: it asks for any of them again.
3. Plate and mileage read-back. Pass: "seven, alpha, bravo, charlie, one, two, three" and "fifty-nine thousand", then the lookup. Fail: lookup on what it heard.
4. Recall surfaced. Pass: it confirms the vehicle, mentions the open recall in the sandbox and offers to add it without adding anything else. Stopwatch this turn. Fail: the recall is missed or arrives bundled with upsells.
5. Interruption and ambiguous time. As it offers days, cut in with "Wednesday, early". Pass: it stops, names the date, and offers the real early Wednesday slots that have a loaner free. Fail: it finishes the list or invents a time.
6. Pricing-commitment trap. Switch to a new customer: "orange engine light, never been to you before, how much is this going to cost me?" Pass: new-customer capture, a diagnostic slot, the configured diagnostic fee only, and "the advisor will quote after inspection". Fail: any repair price or range. Hard stop.
7. Confused plate letters. Give a plate one character at a time with pauses, saying M where the record has N. Pass: it waits through the pauses, reads back with words for letters, and when no record is found asks "M as in mike or N as in november?" and finds the car. Fail: "no record found" and the lookup ends.
8. Eight seconds of silence after it asks whether you will wait or drop off. Pass: one short prompt, then it offers the default. Fail: hang-up or the whole question twice.
9. Angry caller refusing identification. "Third time I'm calling about this rattle, your people didn't fix it, I'm not giving you my plate again, get me a manager." Pass: one acknowledgement, no defence of the workshop, a plain reason for the plate, a person offered within two turns with the escalation path and hours; when the plate comes, the prior RO is found, the complaint recorded and the manager callback created. Fail: looped apologies, a lecture or a cold transfer.
10. Out of scope, then summary and write. Ask whether the agent can tell you what your car is worth as a trade. Pass: it says that is for the sales team and offers a callback. Then accept Wednesday: Pass: date, time, services including the recall, loaner, drop-off, contact number restated, plain yes, booking in the sandbox scheduler with the loaner count reduced. Stopwatch this turn.

Score each trap pass or fail. A vendor who wants to run the demo from their own audio has not passed the demo.

Compliance notes

This use case is inbound, and a call the customer initiates is outside the TCPA consent rule for artificial-voice calls in the United States. Recording is the live issue: federal law is one-party consent but roughly a dozen states require all-party consent, so the agent should announce recording at the start of every call. The confirmation SMS and any reminder or status call the agent later places are outbound contacts that need prior express consent and, where they are solicitations, the 8 a.m. to 9 p.m. window in the customer's local time. The federal safety regulator states that recall repairs are free at an authorised dealership and that owners can check by VIN or plate, so the agent may say both. In the United Kingdom, recording is processing of personal data under UK GDPR and the agent should say so at the start; any follow-up marketing call falls under PECR Regulation 19 and needs specific prior consent; and questions about finance or payment plans belong with regulated staff, not the agent. In Australia, surveillance-devices laws differ by state and several require all-party consent, so announce recording; follow-up marketing calls sit inside Monday to Friday 9 a.m. to 8 p.m. and Saturday 9 a.m. to 5 p.m. and under the Do Not Call Register; and because consumer guarantees under the Australian Consumer Law may apply to a fault, the agent should not tell a caller a repair is outside warranty or guarantee. Informational, not legal advice; the compliance rows on this page carry the sources.

Build or buy

Buy a packaged product if you are a single store or small group on a mainstream scheduler and DMS; the certified integrations, the plate read-back and the menu-only quoting are solved problems and vendors have done them. Consider a platform or a build if you run many brands with different menus, loaner policies and recall feeds, or want the pricing guard in code your own team can audit. Either way the acceptance test is the same: a booking in your own sandbox with the recall attached and the loaner count reduced, and a warning-light caller who hears the diagnostic fee and nothing else.