In a dealership AI arrives in three places before anyone decides anything: the business development desk that answers the internet lead and books the appointment, the finance office that works the credit application and presents the products, and the service lane that writes the repair order and sends the estimate. The owner's job is not to pick the tool. It is to decide three things before a tool touches a deal, a customer's credit application or a repair order: who in the store owns the question, what may never go into an outside tool, and what the dealership will never let a machine do alone. This article is for the dealer principal or general manager of a single-point store or a small group with a few dozen to a few hundred people.
I was COO and then President of a public company that closed more than forty acquisitions, and I ran the integration on most of them. I am a lawyer by training. Now I write fixed-fee AI plans for CEOs of private-equity-backed and owner-led mid-market companies through 013 Labs, and dealerships are the group where a tool's mistake becomes a price quoted to a customer, a credit decision nobody can explain or a safety repair nobody approved.
Where AI already is in your store
Ask the business development desk what the free tools on the phones are doing this week and the honest answer is: answering the lead's first message, quoting a payment from a number it made up, writing the follow-up text. Ask the finance office and it is the summary of the customer's application, the draft of the product presentation, the answer to the lender's conditions. Ask the service lane and it is the repair-order story, the estimate text, the answer to the customer's "is it safe to drive". Ask the office and it is the job posting, the collections note and the answer to the manufacturer's survey. And the dealer-management vendor and the customer-relationship vendor each switched on a drafting feature with their last release. None of it was approved. Some of it has already promised a payment or a price with nobody's name on it, and some of it has carried a customer's credit application, driver's license image or social security number onto a vendor's server.
The three decisions, in the order a dealer makes them
Who owns the question. Not the IT provider, not the dealer-management vendor and not a new hire. In most stores it is the general manager, the general sales manager or the controller, because they are the people the desk and the lane already ask when something new shows up. Give that person a written scope of half a page: what they may approve alone, what comes to the owner, when the page is reviewed. The article on AI for a small business with no IT department draws the line between their column and yours.
The data rule. One sentence. For a dealership it usually reads: no customer's credit application, identification, bank or income document, deal jacket or vehicle history, no manufacturer's program terms or incentive data, and nothing about an employee go into a tool the store has not approved. The owner writes it, because the credit application is protected by federal rules the store already answers to, and the manufacturer's terms are under the franchise agreement.
The never-build list. The three things the dealership will never let a machine do alone. For a dealer the list writes itself: no price, payment, trade value or rate is quoted to a customer by a tool without a named manager behind it; no credit decision, lender submission or product presentation is made or sent by a tool; no repair order is written up, approved or declined, and no safety-related repair is described to a customer, by a tool alone. Everything else is a question of fit. These three are not.
What fits at fifty people, and what does not yet
The first project is the work your people already do by hand, the same way, every week: the first draft of the answer to the internet lead for the desk to send, the summary of the application for the finance manager to check against the documents, the repair-order story drafted from the technician's notes for the advisor to finish, the appointment reminder, the job posting, the collections note. The article on picking your first AI project has the five tests. What does not fit yet, in most stores this size, is anything that decides a number or a customer's standing: the price, the payment, the trade, the rate, the lender, the repair approval. A tool can prepare any of those for a person. It should not finish them.
The manufacturer, the lender, the regulator and the buyer will ask
The AI question reaches a dealership from four directions that have nothing to do with technology. The manufacturer's dealer audit and its customer-experience surveys now carry questions about which systems answer customers and who stands behind the answers. The floor-plan lender and the finance sources ask which systems touch the credit application and whether a person made the submission, because a submission a tool sent is a submission the dealer sent. The state regulator and the federal consumer rules ask it about advertising, pricing and the safeguarding of customer information. And a buyer, a dealer group or private equity, asks it in diligence, because stores are bought in groups now; the article on whether AI makes a business worth more when you sell has that version, and the one on what to tell your bank about AI covers the bank behind the floor plan. One dated page answers all of them.
Picture a two-store family dealership of one hundred sixty people. The business development desk at the larger store has been answering internet leads from a free tool on the staff's own phones for a season, and the appointment count has been up for it. The owner finds out when a customer arrives holding a payment the tool quoted on a vehicle with a different price, and then finds that the tool has also been quoting trade values, that a customer's uploaded driver's license and pay stub went into it, and that nobody can say which messages a person wrote. The fix takes a week and one call to the store's compliance counsel. The general manager is named as the person who owns the question, with a half-page scope. The owner writes the data rule: no customer document, no manufacturer program data, no employee file into an outside tool. The never-build list goes on the page, starting with no number quoted to a customer without a named manager. The free tool is replaced by the store's plan the same week, the desk's reply templates are rewritten by a person with no numbers in them, the customer documents are deleted from the vendor's account with the deletion confirmed in writing, and the customer who arrived with the wrong payment leaves with a deal a manager put a name to. This is a hypothesis, a composite and not a client.
The one page
What is in use, by desk and task, including the free tools and the features the dealer-management and customer-relationship vendors switched on. Who owns the question, with the written scope. The data rule in one sentence. The never-build list, the three items above. The review date and who signs. The owner signs it, the controller keeps it with the safeguarding program, and the general manager has it ready for the next manufacturer audit and the next lender review. The board-briefing template has the same page in a form a dealership can fill in. If you want to know which of the three is missing before you write it, the AI readiness scorecard asks the questions a dealer principal can answer from memory.
Mistakes that repeat
Letting the vendors switch it on. Two vendors, two releases, two drafting features defaulting to on, and nobody decided. The owner decides, feature by feature.
The credit application in a free tool. The one mistake that puts the store in front of a regulator and the lender. The data rule exists for this.
A payment a tool made up. The desk's tool quotes a payment from a number nobody checked; the customer drives in holding it. A manager puts a name to every number before it goes to a customer.
A repair the tool declined. A tool writes "customer declined" on a safety item nobody explained. The advisor explains every safety item in person and writes the record.
No review date. Manufacturers change their audits, lenders change their conditions, people change stores. A page without a date is a page nobody opens.
Frequently asked questions
Should a car dealership use AI?
It already does, through the business development desk, the finance office and the service lane. The decision worth making is who owns the question, what may never go into an outside tool, and the three things the store never lets a machine do alone: a number quoted to a customer without a named manager, a credit decision or lender submission, and a repair order written up or declined.
Who should own AI in a dealership?
The person the desk and the lane already ask when a new tool shows up, usually the general manager, the general sales manager or the controller, with a written scope of half a page and a review date. Not the IT provider and not the dealer-management vendor.
Can AI answer our internet leads?
It can draft the reply, with no numbers in it, for a person to send. It should not quote a price, a payment or a trade value, because a number a tool quoted is a number the dealer quoted, and the regulator and the customer's lawyer will hold the store to it.
What about customer information in these tools?
It is first in the data rule. No customer's credit application, identification, income document, deal jacket or vehicle history goes into a tool the store has not approved, and the approved tools are on business plans whose terms keep the data out of the vendor's training. If customer documents have already gone into a free tool, delete them from the vendor's account, get the deletion confirmed in writing, and tell the store's compliance counsel the same day.
Why does my manufacturer or my lender ask about AI?
Because each carries the risk when an answer to a customer, a submission or a repair record that a tool touched turns out to be wrong, and each wants to know whether a person stands behind it. One dated page with names on it is the answer, and the same page answers the regulator and a buyer.
Not sure where you stand? Take the 3-minute AI readiness scorecard.