Repair shops are short on technician hours, not on demand. That changes the pitch from more leads to protecting the hours they already sold. Here is the vertical playbook: the three calls to build the demo around, what an AI front desk cannot fix, and how to price it.
Lumina9 min read
Sell the shop on the phone, not on AI. An independent repair shop sells bay hours, and every call that lands in voicemail or gets a guessed price is bay hours given away for free. Lead with the three calls the service writer cannot get to, show an agent handling those three, and leave the phrase "artificial intelligence" out of the first five minutes.
That is the whole pitch. It works because auto repair is not HVAC with different tools, and the playbook you use for roofing will lose you this deal. Repair shops have a demand problem in reverse. More work is coming at them than they have hands for, so an offer built around "more leads" sounds like an offer to make Monday worse. The angle that closes is protecting the hours they already sold.
Why demand is not the thing a repair shop is short of
The fleet is old and getting older. S&P Global Mobility put the average age of the 289 million light vehicles on US roads at 12.8 years in 2025, with passenger cars alone at 14.5 years, and it expects a wave of service demand as 2015 to 2019 model years move through the six to fourteen year window where things start breaking.
The hands are not keeping up. The Bureau of Labor Statistics counts 825,800 automotive service technicians and mechanics, with about 66,200 openings projected each year over the decade and median pay of $50,620 in 2025. Only 25 percent of those techs work at mechanical and electrical repair shops. Dealers hold 32 percent, and they pay. The Auto Care Association keeps a standing page on the technician shortage, which tells you it is structural rather than a bad quarter.
Put those two together and you get the shop you are walking into. Work is available. Labor is capped. The front desk is the part of that system under the most pressure, because it is one or two people absorbing every interruption that would otherwise hit a tech. That is where your system goes.
What the owner will tell you the problem is
Do not walk in assuming he agrees with you about the phone. IMR Automotive Research asked 500 shops in December 2025 what their biggest challenges were, and the answers were parts pricing at 46 percent, keeping overhead down at 35.8 percent, staying current on diagnostic tools and software at 30 percent, bringing customers back at 28.6 percent, and finding qualified techs at 23.8 percent.
Missed calls are not on that list. Neither is the website. If you open with "you are losing money on the phone," you are telling someone they are wrong about their own business before you have earned the right to.
So attach to what is on the list. Customer retention and the tech shortage are both front desk problems wearing other names. The customer who did not come back often did not get called back. The tech shortage bites hardest when the service writer is on hold with a parts supplier and cannot build the next repair order. Frame the system as giving the front desk its day back, and you are answering a question he already asked himself.
The three calls to build the demo around
Before you build anything, get the owner to walk you through yesterday's calls. Ask for them sorted, roughly, into buckets. You will almost always end up with these three, and they are what your demo should handle in order.
The status call. My car has been there since 8am, is it done. These are the highest volume and lowest value calls in the building, pure interruption. They are also the hardest to automate well, for reasons in the limits section below. Handle this one honestly and you sound like someone who has been inside a shop.
The price shopper. How much for front brakes on a 2016 Civic. The wrong answer is a number. The right answer is a band, the reason the band exists, and an inspection booked. An agent that holds a range without flinching and then asks for the appointment outperforms a service writer who is tired and discounts to end the call.
The after-hours call. The alternator went at 9pm and the car is in a parking lot. That caller will reach somebody tonight. If it is the chain two blocks over, that is a repair order and a tow fee gone. This section lands easiest, because the owner knows exactly who gets those calls now.
Setting the demo up
Auto repair shop is one of the 100+ vertical packs, and so are auto glass, auto body, tire shop, towing, car dealership, and detailing, which matters because an owner with a body shop next door is a second close. The pack carries the services, the price bands, and the objections in the language a shop uses, so the agent asks for the year, make, and model before it talks about money, the way a service advisor would.
Build it the way you would any other: pick the pack, type in the shop's name, point it at their site. Arrange the nodes so the live demo runs in the order of the three calls above, with the missed-call text-back node last, because that is the one that makes the after-hours story concrete instead of hypothetical. Run the website audit against their real domain while they watch. Shops tend to have old sites with the hours wrong, and they know it.
If you have not run one of these calls before, the demo call playbook covers the sequence. The short version for this trade is to hand over the phone and let them try to break it. Shop owners test things.
What an AI front desk will not fix at a repair shop
This is the section that gets you the deal, so do not skip it on the call.
It cannot diagnose, and it must not try. A caller describing a grinding noise will get a guess out of any model allowed to guess, and a guess becomes an expectation, and an expectation becomes an argument at pickup. The agent gathers symptoms and books the inspection. That is the ceiling and it should be.
It cannot quote firmly on anything that needs a lift. Ranges only, with the diagnostic fee stated. An over-specific quote on the phone costs the shop more than the call was worth.
Live status answers are a scoping question. This is the piece people oversell. Tekmetric, for instance, documents customer-facing online booking and a CRM integration path where a third party can push appointments in and pull appointment status out. Appointment status is not repair order status. Knowing a customer confirmed for 9am is a different object from knowing whether the car is waiting on a part. Find out what the shop runs, find out what it exposes, and price the status feature as work. If the shop is on something old with no API, the agent takes the status call, logs it, and texts the service writer, and a human still answers it. That is a win. It is not the one you get by promising a live lookup you cannot deliver.
It does not add a technician. If the shop is booked three weeks out, booking more appointments is not revenue, it is a longer wait and a worse review. Those shops should buy this for the front desk's hours and for the callbacks that never happen, not for lead capture. If the owner only hears "more leads" and likes it, slow them down, because that is a cancellation in month four.
Who it is not for. The one-bay owner-operator who answers his own cell between jobs and likes it that way. They are not short of front desk capacity, they are the front desk, and they would rather keep it. Sell them a website and a text-back line if they want one, or sell them nothing and keep the relationship.
Pricing it
The platform is $49 a month and client builds are quoted wholesale on a 30-minute call. The top tier, the one with voice included, runs $1,150 to $1,400 setup and $315 to $460 a month plus $0.35 a minute on voice, against $1,730 to $2,280 if the same pieces were bought line by line. You set retail and keep the gap.
For a shop, anchor the monthly against one repair order. Do not use a number from another client, and do not use one from a blog post. Ask him for his average ticket and his gross profit on it, then do the arithmetic in front of him with his figures. An owner who watches you divide their own numbers is a different buyer from one who sat through an industry average. The fuller version of that conversation is in how much to charge for an AI buildout.
FAQ
Do I need to integrate with their shop management system to sell this?
No, and the first deal should not depend on it. Booking, after-hours capture, price-band handling, and missed-call text-back all work without touching their system. Live repair order status needs a connection, and it belongs in a second conversation once you know what they run.
Can the agent give prices over the phone?
Bands, yes. Firm quotes on anything requiring diagnosis, no. The pack ships with price bands rather than fixed numbers for exactly this reason, and you fill them in from the shop's own menu during onboarding.
Will their customers know it is an AI?
On your published demo the sandbox disclosure is always visible, since the page states the agent is a demo and bookings in it are simulated. On a live build, disclosure is a decision you and the shop make together, and in some states it is not optional. We went through the rules in do you have to tell callers it's an AI.
Is auto repair a good first vertical to pick?
It is a good second. The calls are high volume and the value per saved job is clear, but the status-call problem puts integration questions on the table early, and you want a close or two behind you before you scope a software connection on a first call. Start with a trade where booking is the whole job, then come back to shops.
What about tire shops, body shops, and towing?
Separate packs, and worth treating as separate pitches. A body shop's inbound is insurance-shaped and the conversation is about estimates and rental cars, not oil changes. Towing is almost entirely after-hours urgency, which makes it the easiest voice demo in the catalog.
Where to start tonight
Open Google Maps, search auto repair in one zip code, and sort the results by reviews. Read the one-star and three-star reviews for the shops with 4.3 to 4.7 stars, which are good shops having bad days. Find the ones where customers complain that nobody answered, nobody called back, or nobody told them the car was ready. Those complaints are a transcript of the three calls above. Build the demo for that shop, with that shop's name on it, and send the link with one line quoting their own review.
Then do the next zip code. Start your workspace if you have not, and the limits tell you how many demo calls the month holds.
