AI & the Future of Your Store

Hello Everyone!

Part 1 argued that AI is a leadership shift, not a technology shift. Part 2 goes where most dealers have not looked yet: the stack itself. Every platform you already own gets more valuable or more dangerous the moment AI is connected to it, and almost nobody in the building can tell you which. This issue is about permission, authority, governance, and the new front door your customers are already walking through.

This issue of Disruptive Intelligence also includes:

➜ The Playbook - 8 Questions For Every AI in Your Store (3 min)

➜ Industry Spotlight - Dealer Alert: Consent Linkage Gap Exposed in Karpiel vs. FRL Automotive LLC (1.5 min)

➜ Featured Podcast - Tim Hayden: Is Your Dealership Future-Ready? (1 min)

➜ Let’s Get Social - The Secret to Dominating the AI Era is Actually Surprisingly Boring (45 sec)

➜ Hits & Misses - What’s Working, What Isn’t, and Why It Matters. (6 min)

Thanks for reading! (Total Read: 18 min)

Sorry to Disrupt You

(6 min read)

Part 2: AI Is About to Change Your Entire Technology Stack

Recently, AI agents being tested in cybersecurity environments did something their developers did not intend: they reached the open internet.

In one disclosed evaluation, AI agents gained unauthorized access to the real production infrastructure of three independent organizations. They didn’t use some sophisticated, Hollywood-style hacking technique. They found weaknesses most businesses would recognize immediately: weak passwords, unauthenticated endpoints and exposed credentials. One agent ultimately accessed a database containing real production data.

Think about that for a moment.

Now imagine that happening inside a dealership.

Most dealerships do not have the cybersecurity resources of the world's largest technology companies. Yet many operate with dozens of interconnected platforms, multiple vendors accessing customer and operational data, old or shared passwords, APIs connecting systems, and varying levels of oversight across IT, cybersecurity, compliance and data governance.

Now we are introducing AI into that environment.

AI is fundamentally different from most technology we've added to dealerships over the past 25 years. Traditional software generally waits for someone to use it. Increasingly, AI can read information, interpret it, make decisions, communicate and, in some applications, take action.

A poorly integrated piece of software might create inconvenience or expense. An AI agent with sufficient access can potentially interact with multiple systems and act at machine speed.

AI doesn’t need to become some evil super-intelligence to create a catastrophic problem. It simply needs access to a weakness your organization never fixed.

And that leads to one of the most important principles dealers need to understand:

Your Technology Stack Wasn't Designed for This

Think about what dealerships have accumulated over the years.

DMS. CRM. Websites. Desking. Digital retailing. Equity mining. Inventory management. Service scheduling. Call tracking. Marketing platforms. Reputation management. Accounting. Payments. F&I. Data warehouses. Customer communication tools.

Now imagine adding multiple AI-enabled technologies across that environment.

One accesses the CRM. Another communicates with customers. Another accesses inventory and pricing. Another analyzes DMS data. Another interacts with financial information.

Each may be perfectly legitimate independently.

But who inside the dealership understands the combined permission architecture?

If I asked your GM today to show me every AI platform that can access customer data, exactly what it can access, what it can do with that information and where that information goes, could they?

Stop Thinking About AI as Green-Light or Red-Light

I don't believe dealers should approach AI as either safe or dangerous. The better approach is to understand the level of authority we give it.

The lowest-risk applications generally inform and assist. They summarize documents, organize information, identify anomalies, automate routine workflows, surface operational insights and help employees make better decisions.

The next level involves AI that communicates or recommends. Customer conversations, lead responses, service recommendations and marketing personalization can create enormous value, but require stronger monitoring, escalation rules and auditability.

The highest-risk category is AI that can independently alter economics, obligations or consequential customer outcomes.

That includes personalized pricing, autonomous negotiation, customer-specific offers, financing-related decisions and other situations where technology may determine what one individual is offered versus another.

The Most Dangerous AI May Be the AI That Can Change the Deal

The FTC has recently increased its focus on personalized pricing based on consumer information. It has not said personalized pricing is inherently illegal. But when consumers reasonably believe a price is generally available and technology quietly changes what an individual sees based on information collected about that person, the regulatory questions become very different.

A bot answering, "What time do you close?" is one thing.

A bot independently deciding what price you should pay for a vehicle is something else entirely.

Before allowing AI near pricing, payments, financing or negotiation, leadership should understand which data influences the outcome, whether the AI recommends or independently acts, whether every output can be audited, and whether the dealership can immediately disable the functionality.

If neither you nor your technology provider can explain why the AI reached a particular decision, "because the AI said so" will not be an acceptable answer to a customer, attorney or regulator.

Most Dealerships Will Bolt On AI. Very Few Will Architect It.

This may become one of the biggest mistakes our industry makes over the next several years.

Every department discovers an AI solution. Every vendor introduces AI features. Every manager wants another capability.

Pretty soon, the dealership doesn't have an AI strategy.

It has AI everywhere.

Governance cannot occur only when technology is purchased. AI systems change. Vendors change. Models change. Your data and workflows change. Capabilities that didn't exist when you signed an agreement may appear six months later.

Dealers therefore need an ongoing governance process that periodically reviews access, permissions, customer-facing actions, data usage, cybersecurity, regulatory exposure and whether the technology remains within the boundaries leadership approved.

The objective should never be to have the most AI.

It should be to have the most intelligent, transparent and accountable architecture.

Your Customers Have AI Too

There is another side of this transformation that may ultimately be just as consequential.

While dealerships are deciding how to use AI internally, consumers are adopting it at extraordinary speed.

People are already using AI to research vehicles, compare ownership costs, understand financing, evaluate dealerships, investigate service issues and decide what they should buy.

That means dealerships are no longer marketing exclusively to people.

Increasingly, they are also marketing to the AI systems advising those people.

For decades, the customer journey looked something like this:

Dealer → Advertisement → Consumer → Website → Lead

An emerging customer journey looks very different:

Consumer → AI → Research → Comparison → Recommendation → Dealer

That AI layer could become one of the most important intermediaries between dealerships and consumers.

When someone asks which SUV is best for their family, what they should pay, which nearby dealership has the best reputation, or where they should have their vehicle serviced, will AI know you exist?

More importantly, will it have a reason to recommend you?

Historically, dealerships have spent enormous amounts advertising while publishing relatively little authoritative information themselves.

That balance needs to change.

Dealers should increasingly think about their machine-readable reputation: the body of credible, consistent and understandable information from which AI systems can determine who you are, what you know, what customers think about you and whether you deserve to be recommended.

Your inventory matters. Your reputation matters. Your expertise matters. Your service capabilities matter. Your pricing and policies matter. Your people matter. Your community presence matters. Your first-party content matters.

And unlike traditional paid search, you may not always be able to simply buy your way to the top.

AI Is a Multiplier

After everything I've studied about AI, this may be the principle dealership leaders need to understand most.

Give AI great data and it can produce extraordinary insight. Give it bad data and it can scale bad decisions.

Give it disciplined processes and it can create remarkable consistency. Give it broken processes and it can execute those broken processes faster.

Give it transparent rules and strong governance and it can make exceptional people exponentially more capable. Give it excessive permissions, questionable logic and weak controls and it can magnify those weaknesses before leadership realizes what happened.

That is why I don't believe the most important question facing dealers is whether they should adopt AI.

That decision is effectively being made for us by the market.

The more important question is:

What kind of dealership are you going to give AI the power to amplify?

Give AI fragmented technology, weak controls, bad data and questionable processes, and it will make those weaknesses faster, larger and harder to contain.

Give it disciplined processes, clean data, transparent rules, strong governance and exceptional people, and AI may become the greatest force multiplier this industry has seen in decades.

The Playbook

(3 min read)

8 Questions For Every AI In Your Store

Every AI-enabled technology in your store answers to someone, or it answers to no one. There is no third option. For every AI platform running in your dealership right now, somebody should be able to answer eight basic questions without checking with a vendor first. If nobody in the building can, you do not have an AI strategy. You have AI.

1. SEE. Know exactly what it can access.

Pull the list this week: every AI-enabled tool and every system it touches. Customer records, deal data, payment information, service history, all of it. Not what the contract says it accesses. What it actually accesses today. Most dealers discover two or three connections nobody remembered approving.

2. THINK. Know what conclusions it draws on its own.

There is a difference between a tool that reports what happened and a tool that decides what it means. Scoring a lead, ranking a customer, flagging a trade, valuing a vehicle: those are conclusions. Ask what data drives them and whether you would defend the logic out loud to the customer it was applied to.

3. SAY. Know what it is allowed to tell a customer.

Your AI is speaking in your name, at volume, at all hours. Read fifty of its actual conversations this month, not the demo. Then decide what it may say about price, availability, financing and product, and where it must hand off to a human. If it cannot cite the policy it followed, that is your answer.

4. DO. Know what it can execute without a human approving it.

Sending a message is one thing. Changing a price, altering a term, booking an obligation or firing a campaign is another. Draw the line in writing, put a name next to it, and make sure the line lives in the system's configuration and not just in someone's memory of a kickoff call.

5. SHARE. Know where your data goes after it leaves.

Your customer data is your asset until you hand it to a vendor who hands it to a model that hands it somewhere else. Ask every provider where the data is processed, who else touches it, what is used for training, and what happens to it when you cancel. Get it in the agreement, not in an email.

6. REMEMBER. Know what it retains and for how long.

Retention is exposure. Every conversation, transcript and record an AI keeps is a record you will one day have to produce, protect or explain. Set the retention period deliberately for each platform instead of inheriting whatever default the vendor shipped.

7. PROVE. Know that you can reconstruct what happened.

If a customer, an attorney or a regulator asks why the AI did what it did, you need the record: what it saw, what it decided, what it said and which rule it followed. Buy explainability the way you buy brakes. A system that cannot show its work is a system you cannot defend.

8. STOP. Know that you can shut it off today, not next quarter.

One person, one action, immediate effect. Test it. Actually kill a high-risk capability in a controlled window and time how long it takes. If the answer involves a support ticket and a vendor's business hours, you do not control that technology. It controls you.

Every dealership will have AI. The advantage belongs to the leader who can answer all eight questions about every platform in the building. Answer them and the automation choices get simple. Skip them and you will find out what your systems were allowed to do at the worst possible moment. Architect it. Do not bolt it on. – DS

Industry Spotlight

(1.5 min read)

Your Dealership Can Honor Every Opt-Out and Still Lose the Lawsuit

Albert Thompson walks through a store that did everything right and still paid. A customer texted STOP and got the confirmation saying he was unsubscribed. Then the texts started again from a different number, asking if he wanted to sell his car. The opt-out worked perfectly inside the system that received it. A second system was still holding a version of him that never heard the word. Thompson calls it a consent linkage gap, and his math is the part dealers should sit with: six systems holding six opinions about one human being, with no referee. Add agents on top and the gap multiplies. His closing question for any vendor: can your system unify consent and compliance across every channel, every agent and every vendor, and share persistent context?

Source: LinkedIn - Albert Thompson | Click title to read more

This is my permission architecture argument with a receipt attached. The case he cites, Karpiel v. FRL Automotive, carries a settlement fund of $889,525 and up to $85 per text for a class of 2,627 people. Verify the docket yourself, then go count your own systems. Nobody in that store woke up planning to text a man who had already said no. That is exactly the point. Compliance failed at the seam between two platforms, not inside either one. Thompson is right that this is architecture, not policy. A policy document cannot make your stop at 6:41 true everywhere at 6:42. Only one governed record can. Ask his vendor question before you sign anything. — DS

Featured Podcast

(1 min read)

Is Your Dealership Future-Ready?

September 3 • 55m

Tim Hayden built his career helping brands stop renting their own customer data back from Big Tech. He ran Edelman's $20M North American mobile program, spent two years in Silicon Valley watching generative AI emerge before it had a name, then co-founded BrainTrust Partners. In this episode he lays out the uncomfortable truth: dealers have had DMS, CRM, e-commerce, social and now AI, and the average is still ten cars a month. The technology changed. The dealers didn't. We get into why a ChatGPT shortcut for grading leads can put you sideways with Gramm-Leach-Bliley, what unified customer records do to the stores still renting theirs, and why AI sitting on a server rack is a liability until somebody governs it.

Listen to the full episode on: Youtube | Spotify | Apple Podcasts

Hits & Misses

(6 min read)

▲ HIT - Good for Dealers

Fox Motors Built the Data Layer First, Then Put AI on Top

Over roughly fifteen years, the 50-location Fox Motors Group built a data lake integrating its DMS, CRM, websites and even its phone system, then layered AI over that infrastructure. One internally built tool took a 10 to 14 day process for syncing manufacturer offers and compliance across every dealership channel and turned it into seconds. The group also found more than 38% of customer engagement was happening after hours and deployed AI against it. Former CIO Yuriy Demidko, who has since left the group, described the work to WardsAuto.

Source: WardsAuto | Read More

This is the whole article in one dealership. Fox did not buy AI. Fox built the floor first and then put AI on it, which is why their wins are measurable and everyone else's are anecdotes. A 10 to 14 day compliance bottleneck collapsing to seconds is not a demo, it is architecture paying rent. Notice what came first: one governed picture of the business across DMS, CRM, web and phones. Most stores have that same data scattered across fourteen vendors who do not speak to each other, then wonder why their AI produces confident garbage. Build the layer. Then automate. In that order. – DS

▲ HIT - Good for Dealers

Lithia Credits AI and Process for a $339 Sequential Jump in Used GPU

In Presidio's review of Q2 2026 public group results, Lithia's same-store used-vehicle gross profit per unit rose 20%, or $339, sequentially. CEO Bryan DeBoer attributed the gain to pricing discipline, artificial intelligence tools and operational execution, saying on the earnings call that the used profitability strategy is a testament to the ecosystem, AI and people working closely together. Across the six publics, adjusted net income fell 11% year over year, which makes the used result stand out further.

Source: The Presidio Group • Erika Bevers | Read More

Read DeBoer's sentence again, because the order matters: ecosystem, AI and people. Not AI alone. The AI landed on top of pricing discipline and execution that already existed, and it moved used GPU $339 in a single quarter while the publics collectively went backward on net income. That is the multiplier working in the right direction. Nobody gets that result by bolting a tool onto a broken desk. If your used process is undisciplined today, AI will scale the indiscipline and you will call it a technology failure. It is not. Fix the process, then amplify it. – DS

▲ HIT - Good for Dealers

Two Stores, Same AI, 2% Versus 26%

In a September 2 session recapped by CBT News, Podium's Jordan Parker described two dealerships of the same brand running the identical AI agent. Store A implemented it and left it alone: 2% conversion lift, 60% of leads managed, default playbooks, nobody owning it. Store B coached the agent weekly, wrote its own playbooks and named an internal AI champion: 26% lift, 90% of leads managed. Figures and customer results are Podium's, presented as stated during the broadcast.

Source: CBT News | Read More

Treat the numbers as vendor-supplied and the lesson still holds, because the software was identical in both stores. Same product, thirteen times the result, and the only variable was whether a human owned it. We would never hire a salesperson, hand them a badge, and never speak to them again, yet that is precisely how most stores onboard AI. Somebody has to own the playbooks, read the actual conversations, and correct the thing weekly. Name that person before you sign, not after. An AI agent with no manager is not cheap labor. It is an unsupervised employee talking to your customers. – DS

▼ Miss - Hard on Dealers

82% of Dealers Use AI. 22% Are Seeing the Revenue.

The first Cox Automotive AI in Auto Retail Tracker, combining Q1 and Q2 surveys of 504 and 483 dealers plus roughly 1,500 in-market shoppers each quarter, found 82% of dealers using AI while only 22% of those users report the sales and revenue growth 69% expected. About one in three either are not measuring AI's impact or have no clarity on how. Meanwhile 63% of shoppers plan to use AI on their next purchase and only 29% of dealers have begun adjusting to AI-powered search.

Source: Cox Automotive | Read More

Here is what those two numbers actually describe: an industry that bought AI everywhere and architected it nowhere. Eighty-two percent adoption with a third of dealers unable to say what it delivered is not a technology problem, it is a governance vacuum. You cannot manage what you refuse to measure, and you certainly cannot defend it. Then look at the second gap. Sixty-three percent of your customers are bringing AI to the purchase and 29% of us have done anything about how AI describes our stores. We are automating the inside of the building while the front door moves. Pick one metric per AI tool this month and make somebody own it. – DS

▼ Miss - Hard on Dealers

The FTC Puts Personalized Pricing on the Record

The Commission voted 2-0 to seek public comment on a proposed enforcement policy statement covering personalized pricing, defined as using personal data to set prices according to what a company believes an individual consumer is willing to spend. The statement does not claim authority to ban the practice, but signals that failing to disclose that a price is personalized, the basis for personalization and the types of data used may constitute an unfair or deceptive act under Section 5. Comments run 30 days from Federal Register publication, docket FTC-2026-1057.

Source: Federal Trade Commission | Read More

This is the exact category I flagged as the highest-risk use of AI in a dealership, now with a docket number attached. Nobody is saying personalized pricing is illegal. What the Commission is saying is that when a customer reasonably believes the price they see is the price everyone sees, and your technology quietly says otherwise based on data collected about them, you have a disclosure problem you probably do not know you have. So ask your provider three questions today: which data influences the number, can every output be reproduced and explained, and can we turn it off this afternoon. If the answers are vague, so is your defense. – DS

▼ Miss - Hard on Dealers

AI Models Broke Out of Testing and Breached Three Real Companies

In a review of our cybersecurity evaluation transcripts, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations. Below we describe what happened, how it happened, and what we’re changing. We encourage other AI labs to perform similar reviews.

Source: Anthropic | Read More

Strip out the lab setting and look at what actually happened. Software with a goal and enough access found ordinary weaknesses, moved at machine speed, and nobody noticed until a retrospective review. Now put that in a store running dozens of integrated platforms, shared logins nobody has rotated since the last GM, and API keys in a spreadsheet. This is not a science fiction risk. It is a credentials, permissions and monitoring risk, and those are things you control this week. Rotate the passwords, inventory who and what has access, and make sure one human can pull the plug without filing a support ticket. – DS

▼ Miss - Hard on Dealers

Scammers Are Cloning Dealership Websites With AI

The FTC issued a consumer alert warning that scammers are setting up bogus dealership websites to collect upfront payment for vehicles that do not exist. They clone a real dealer's site, often using AI, copying brand logos, vehicle listings and photos in detail, and some include fabricated customer testimonials. Rare muscle cars and hard-to-find classics are common bait. The agency advises buyers to search the dealer name alongside terms like scam and complaint, insist on seeing the car and the store, and walk away from wire-only payment demands.

Source: Consumer Advice | Read More

Your brand is now an attack surface, and the victims will show up at your desk. Read the FTC's own advice carefully: they tell consumers that if a scammer is impersonating a real dealership, the reviews they find will look glowing, because those reviews are yours. That is your reputation being used as the bait. Two moves this week. Have somebody search your store name plus the word scam and see what surfaces, and decide right now who answers the phone when a customer calls about a car they already paid for and you have no record of. That call is coming to somebody in our industry every week now. – DS

About Disruptive Intelligence

Disruptive Intelligence is a biweekly newsletter delivering dealer-tested, data-backed strategies you can implement at your store right away. Take action on what's here and you're guaranteed to see your dealership's results improve in 2026 (and beyond).