An AI receptionist answers every call, captures the vehicle, the problem and the address, quotes from your price list and books the job, including at 7:40 PM when everyone has gone home. It won’t diagnose a noise over the phone or talk down an angry customer as well as your best service advisor. Set it up for the calls it handles well, route the rest to a human fast, and it pays for itself on the calls you were missing anyway.
This guide covers what to expect: what it’s good and bad at, the disclosure and recording rules, how setup works, a first-week checklist and the numbers to watch.
Why this matters: the calls you don’t see
Across auto shops, 55% of calls go unanswered (Invoca), and 68% of callers who reach voicemail don’t call back (Marchex). With an average repair order around $450, every missed call that would have booked is real money that went to the next shop on the list.
Mobile mechanics have it worse: the person who answers the phone is often the person under the car. We built our AI desk because our own operation needed it; the system behind it has run 160,000+ repair jobs.
What an AI receptionist handles well
| Call type | How well AI handles it | Notes |
|---|---|---|
| New booking: “my brakes are grinding” | Well | Captures vehicle, concern, address, preferred window; quotes a range from your menu; books |
| Price question: “how much for an oil change?” | Well | Answers from your price list, not a guess |
| Hours, service area, “do you come to me?” | Well | Consistent answers every time |
| Reschedule or confirm an appointment | Well | Updates the booking and texts a confirmation |
| After-hours and overflow calls | Well | This is where most of the gain comes from |
| Status: “is my car ready?” | Depends | Good if job status is in the system it reads from |
| Diagnosis over the phone | Poorly, by design | It should collect symptoms and book an inspection, not diagnose |
| Angry customer or warranty dispute | Hand off | Route to a person with the full transcript |
| Fleet account with special terms | Hand off or scripted | Depends on how well the account rules are set up |
The pattern: an AI receptionist is excellent at structured intake and booking, and should hand off anything that needs judgment, negotiation or empathy under pressure.
Quoting from your price list
The single biggest quality lever is your price list. A good AI receptionist doesn’t invent prices; it quotes from the menu you give it and gives a range where the final price depends on inspection.
Before go-live, make sure you have:
- Your top 15–25 services with a price or range (brakes per axle, battery replacement, diagnostic fee, oil change by type, and so on).
- What changes the price: vehicle size, parts grade, travel zone.
- Your diagnostic fee policy and whether it’s credited toward the repair.
- Services you don’t do, so the AI can say so politely instead of booking them.
- Your service area, by ZIP code or drive-time zone for mobile work.
If you don’t know your own prices precisely yet, fix that first. Our mobile mechanic pricing guide walks through it.
Disclosure: say it’s an AI, every time
Our AI opens every call with: “I’m an AI assistant and this call is recorded.” It answers “yes, I’m an AI” whenever asked, and callers can say “person” to reach your team.
Why be this explicit:
- Several states have laws against bots misleading people about being artificial, and more are being passed.
- Callers who learn later that they were talking to a bot feel tricked; callers told upfront mostly don’t care as long as the call goes well.
- It makes the recording disclosure natural.
Call recording and consent
Recording and transcripts are how you check quality and resolve disputes, but recording laws vary by state. California, Delaware, Florida, Illinois, Maryland, Massachusetts, Montana, Nevada, New Hampshire, Pennsylvania and Washington require consent from all parties to record. A clear disclosure at the start of the call, with the caller choosing to continue, is the standard approach, which is why the disclosure plays on every call and can’t be turned off.
This is general information, not legal advice. If you operate in several states, have your attorney review your notices.
Outbound calls: inbound yes, marketing no
Answering inbound calls is straightforward. Outbound AI calls are different. Under the TCPA, calls using an artificial or prerecorded voice to market to consumers need prior express written consent. Returning a call to someone who just called you in is lower risk, but it still needs the caller to have provided the number for that purpose.
Practical rules:
- Use AI follow-up only with customers who opted in, for purposes they agreed to.
- Don’t use AI to cold-call or reactivate old customers without written consent.
- Honor opt-outs quickly, by any reasonable means the customer uses.
Setup: how calls reach the AI
Most setups take a day or two and keep your existing number:
- Conditional call forwarding. Calls you don’t answer within a set number of rings, or calls when the line is busy, forward to the AI. Most carriers support this with a short code or a setting in the phone system.
- Or full forwarding after hours. Every call after closing goes straight to the AI.
- Business profile. Hours, service area, services, prices, techs and booking windows.
- Handoff rules. Who gets a transfer or an urgent text, and for which call types.
- Test calls. Call it yourself with 10 real scenarios before customers do.
Forwarding fails open: if anything goes wrong, calls keep ringing your line as they do today.
Your first-week checklist
- Price list loaded, with ranges where inspection decides the price
- Service area and drive-time zones set
- Booking windows match real tech capacity
- Handoff rules tested: angry caller, warranty, fleet account, “I want a person”
- Confirmation texts reviewed for tone and accuracy
- Read 20 transcripts on day 2 and fix what the AI got wrong
- Read another 20 on day 5
- Owner and one advisor know how to search calls and recordings
Reading transcripts is the most important habit in the first two weeks. Most early issues are missing prices or unclear service-area rules, not the AI itself.
The metrics to watch
| Metric | What it tells you |
|---|---|
| Answer rate | Share of inbound calls answered by a person or the AI. Target: every call |
| Booking rate on new-job calls | Of calls with a real repair need, how many booked |
| After-hours bookings | Jobs that would have gone to voicemail |
| Handoff rate | Share of calls routed to a human; too high means missing info or rules |
| Show rate | Booked jobs where the customer was there when the tech arrived |
| Revenue from AI-booked jobs | The number that justifies the cost |
Compare the first 30 days to your previous 30. If your phone system can’t show last month’s missed calls, count voicemails as a floor.
Cost framing: AI vs an answering service
A human answering service takes messages; an AI receptionist books jobs. The costs are structured differently too. The comparison below is illustrative; get real quotes for your volume.
| Voicemail | Human answering service | AI receptionist | |
|---|---|---|---|
| Answers after hours | No | Yes | Yes |
| Books the job into your schedule | No | Usually takes a message | Yes |
| Quotes from your price list | No | Rarely | Yes |
| Consistent answers | n/a | Varies by agent | Yes |
| Typical pricing model | Included | Per minute or per call | Monthly per location plus usage |
The simple break-even test: take your average repair order, multiply by the number of extra jobs booked per month, and compare that to the monthly cost. See Front Office AI pricing for our numbers.
Your data
Calls contain your customers’ names, numbers and addresses. Ask any vendor where recordings are stored, how long they’re kept and whether they’re used to train models. We never train on customer calls. Recordings and transcripts are used to run your account, and you control retention. Details are on our security page.
Key takeaways
- An AI receptionist is strongest at structured intake and booking, especially after hours.
- It should hand off diagnosis, disputes and anything that needs judgment.
- Disclose AI and recording on every call; 11 states require all-party consent to record.
- Inbound answering is straightforward; outbound AI marketing needs written consent.
- Load real prices, read transcripts in week one, and measure bookings, not just answers.
Hear a representative sample call or see how the AI Receptionist works.