Dealers: Your Next Customer Will Have an AI Agent Working for Them
If you think AI in the car business is mostly about chatbots, automated follow-up and writing better emails, I think you’re missing what may be the much bigger disruption.
I’ve spent my entire career in the car business. I’ve watched our industry go through the internet, third-party lead providers, digital retailing, online used-car companies, smartphones, social media and now AI.
Some changes were overhyped. Others fundamentally changed the way customers bought cars and the way dealers had to operate.
I believe personal agentic AI has the potential to be one of those fundamental changes.
And I’m not sure most dealers are paying attention to it yet.
I’m not talking about another AI tool for the dealership.
I’m talking about AI that works for the customer.
That’s a very different conversation.
First, understand what just changed
Most people currently think about AI through the lens of ChatGPT, Gemini or another chatbot.
You ask it a question. It gives you an answer.
Personal AI agents are different.
One of the newest examples is Meta’s Muse. It was designed for everyday consumers, not programmers or technology companies.
A personal AI agent can learn about you, remember your preferences, connect to services you use, browse the internet, fill out forms, monitor things over time and perform multi-step tasks on your behalf.
It can negotiate.
It can shop.
It can continue working after you’ve put your phone down.
And when it reaches an important decision or transaction, it can come back and ask for your approval.
Meta actually uses selling a car for more money as one of the examples of what Muse can do.
Think about that for a minute.
This isn’t theoretical technology sitting in a lab.
It’s here.
And consumers are already getting comfortable using AI to shop
The transition isn’t going to happen overnight, and I don’t believe customers are suddenly going to hand an AI agent $70,000 and tell it to buy whatever vehicle it wants.
That’s not the point.
The more important change happens well before the purchase.
Research published in 2026 by Invoca found that 63% of automotive consumers had used generative AI to research a high-stakes purchase, compared with 41% in 2025.
Even among Boomers, usage increased substantially.
Consumers are already learning to use AI as an advisor.
The next step is letting that advisor do some of the work.
And that’s where I think dealers need to start paying attention.
Imagine how someone may shop for a vehicle
Today a customer looking for a Tahoe might:
Google it.
Go to Autotrader.
Check Cars.com.
Look at CarGurus.
Visit several dealer websites.
Check KBB.
Research their trade.
Look for incentives.
Read reviews.
Calculate payments.
Submit a lead.
Get six phone calls.
Get 12 emails.
Negotiate with three dealerships.
Now imagine this instead.
The customer tells their personal AI agent:
“I want a new Tahoe High Country. Black or white. I don’t want to travel more than 150 miles unless the deal is significantly better. I want to put $5,000 down. I’m trading my 2023 Tahoe with 41,000 miles. I don’t want any dealer add-ons. Find me the best three transactions and tell me which ones I should consider.”
And then the customer goes to dinner.
The agent does the work.
It searches inventory.
It compares equipment.
It researches incentives.
It looks at advertised prices.
It checks dealer fees.
It evaluates the customer’s trade.
It calculates payments.
It looks at reviews.
It identifies questionable add-ons.
It potentially communicates with dealerships.
It narrows 75 vehicles down to three.
The next morning the customer isn’t asking:
“Where should I start?”
They’re asking:
“Which of these three deals do I want?”
That is a completely different sales funnel.
Your dealership may eventually be selling to the customer’s AI before you ever talk to the customer
This is the part I think our industry needs to understand.
For the last 25 years, dealers have spent billions of dollars figuring out how to influence the customer digitally.
SEO.
SEM.
Third-party listings.
Retargeting.
Dealer websites.
Chat.
Lead forms.
Digital retailing.
CRM follow-up.
BDC processes.
All of it assumes that a human being is navigating the shopping process.
What happens when the human delegates a meaningful portion of that work to an AI agent?
Your website may still matter.
Your brand absolutely still matters.
Your reputation matters.
Your people matter.
But another question suddenly becomes extremely important:
Can the customer’s AI understand your dealership?
Can it determine:
Dealers have spent years making their websites Google-friendly.
We may be entering a period where they also have to become AI-agent-friendly.
Pricing becomes much more interesting
Dealers hear this and immediately think:
“Great. AI is going to turn everything into a race to the lowest price.”
Maybe.
But I don’t think it’s necessarily that simple.
A good personal agent shouldn’t just compare advertised prices.
It can compare the entire transaction.
Dealer A is $1,500 cheaper on the vehicle.
Dealer B is giving $2,500 more for the trade.
Dealer C has a $995 addendum.
Dealer A charges a $799 documentation fee.
Dealer B has the right vehicle but it’s been in inventory for 83 days.
Dealer C is 180 miles away.
Dealer B has a better interest rate.
An AI agent can potentially put all of that together.
So instead of sorting by:
Lowest advertised price
it can calculate:
Best overall transaction for this particular customer.
That’s actually good news for well-run dealerships.
But it could be very bad news for dealers whose digital pricing strategy depends on getting the customer in the door before revealing the rest of the transaction.
Think about what happens to trade-ins
This may be one of the first areas where dealers feel it.
A consumer tells their agent:
“I’m thinking about selling my 2024 F-150. Figure out what it’s worth.”
The agent gathers the VIN, mileage, equipment and vehicle history.
It looks at market information.
Then it starts soliciting offers.
CarMax.
Carvana.
Local Ford dealers.
Independent dealers.
Online buyers.
Maybe marketplaces.
The consumer doesn’t have to fill out eight forms.
The agent does it.
Now take it one step further.
The customer’s agent contacts your dealership:
“My customer has a 2024 F-150 Lariat with 22,400 miles, clean history and these options. What will you pay for it?”
Your dealership responds:
$48,500 subject to physical verification.
The customer’s agent responds:
“Another buyer has offered $49,200. Can you improve your offer?”
That entire conversation could eventually occur in seconds.
For dealers who depend heavily on acquiring inventory from consumers, this matters.
The consumer’s AI agent could become one of the most powerful vehicle-acquisition intermediaries we’ve ever seen.
Your BDC could start communicating with machines
Think about today’s typical internet lead.
Customer submits information.
CRM creates lead.
Autoresponder goes out.
BDC calls.
BDC texts.
Salesperson calls.
Manager gets involved.
Customer doesn’t answer.
Follow-up cadence starts.
Now imagine the lead looks like this:
“I represent Todd. He is interested in stock #A12345. Please confirm current availability, selling price, mandatory fees, applicable incentives, estimated trade value for the attached vehicle information and available delivery appointments.”
That’s not a traditional lead.
That’s an AI agent representing a customer.
Your dealership’s ability to answer that request accurately and immediately becomes important.
The old BDC question:
“How quickly did we respond to the lead?”
may eventually become:
“Can our systems actually transact intelligently with the customer’s agent?”
F&I could change substantially
Now put the customer’s AI inside the F&I office.
The business manager presents a service contract for $3,495.
Customer asks:
“Is this worth buying?”
The agent knows:
The customer’s expected ownership period.
How many miles they drive.
Factory warranty coverage.
The vehicle’s reliability history.
The deductible.
Exactly what the service contract covers.
Average repair costs.
Competing coverage.
Potentially even what similar coverage costs elsewhere.
The agent responds:
“The coverage could make sense based on how long you plan to own the vehicle, but $3,495 appears expensive relative to comparable coverage. Ask whether they can sell it for $2,500.”
Same thing with:
GAP.
Tire and wheel.
Maintenance plans.
Financing.
Appearance protection.
Extended coverage.
This doesn’t mean F&I disappears.
It means information asymmetry gets smaller.
Good products presented transparently can still sell.
Products that depend heavily on the customer not understanding what they’re buying may become much more difficult to sell.
Now think about your service drive
This may be just as important as vehicle sales.
Customer brings their vehicle in.
Advisor says:
“You need brake fluid, a transmission service, cabin filter, alignment and fuel-system service. Total is $1,187.”
Today the customer might Google a couple of those services.
Tomorrow they may simply ask:
“Do I actually need all this?”
Their personal agent may know:
The VIN.
Mileage.
OEM maintenance schedule.
Previous repair orders.
What services were already performed.
Driving habits.
Vehicle age.
Open recalls.
Warranty status.
Average local pricing.
The agent may answer:
“The brake fluid and transmission service appear appropriate. The cabin filter was replaced eight months ago according to your service history. I would ask why it needs replacement again. The alignment isn’t mileage-based, so ask whether the dealership measured something that indicates you need one.”
Imagine that conversation happening while your advisor is standing in front of the customer.
That changes the service sales process.
Again, the answer isn’t to fight it.
The answer is to make sure your recommendations are accurate, defensible and easy to explain.
The customer’s AI may know they’re in the market before you do
This is the one that really gets my attention.
Our industry spends enormous amounts of money trying to predict purchase intent.
Equity mining.
Lease maturity.
CRM campaigns.
Service-to-sales.
Website behavior.
Third-party intent data.
But think about what a persistent personal AI agent could eventually know.
It knows the customer’s current vehicle.
Mileage.
Payment.
Payoff.
Insurance cost.
Repair history.
How long they typically keep vehicles.
What they’ve been researching.
Family changes.
Driving habits.
Current market value.
Current incentives.
Maybe even their financial goals.
The agent could proactively tell the customer:
“Your vehicle is currently worth approximately $31,500 and you owe $23,800. There is a manufacturer incentive on the model you’ve been considering, and your current vehicle is approaching a major maintenance interval. This may be a good time to evaluate replacing it. Would you like me to see what’s available?”
The customer didn’t submit a lead.
They didn’t search Google.
They didn’t visit your website.
They didn’t click a Facebook ad.
Their AI created the shopping event.
That should get every dealer’s attention.
This could change who controls the customer journey
For years, everyone in automotive has fought to own the customer.
OEMs want the customer.
Dealers want the customer.
Third-party marketplaces want the customer.
Google wants the search traffic.
Social platforms want the customer’s attention.
Lead providers want the lead.
Dealer websites want the conversion.
But personal AI introduces another possibility:
The customer has their own representative.
And that representative works for them.
Its job isn’t to maximize dealership gross.
Its job isn’t to maximize advertising revenue.
Its job isn’t to sell a lead.
Its job is to get the best outcome for its user.
That’s an important change.
I don’t think AI is going to autonomously buy everyone’s next car next year
I want to be clear about that because our industry has a bad habit of taking new technology and immediately running to one of two extremes.
Either:
“This changes everything tomorrow.”
Or:
“Customers will never do that.”
Both are usually wrong.
Research still shows that consumers are hesitant to turn complete purchasing authority over to AI, especially for important purchases.
That makes sense.
Buying a vehicle is emotional.
People want to drive it.
Sit in it.
Touch it.
Ask questions.
See the color.
Talk to someone.
Negotiate.
And ultimately approve a large financial decision.
I don’t expect that to disappear.
But here’s the mistake dealers could make:
The AI doesn’t have to buy the car to completely change how the car gets bought.
If the customer’s agent does the:
Research.
Inventory search.
Comparison.
Trade valuation.
Incentive research.
Payment analysis.
Dealer comparison.
Negotiation.
Appointment scheduling.
And paperwork preparation…
and the human makes the final decision…
the retail process has already changed dramatically.
We’ve seen this movie before
I was a dealer when customers started shopping online.
There was a time when dealers debated whether customers would really shop for cars on the internet.
Then whether customers would submit leads.
Then whether pricing should be online.
Then whether customers would buy used cars sight unseen.
Then whether customers would complete portions of the transaction digitally.
Every time, consumer behavior moved faster than some dealers expected.
I think we’re at another one of those moments.
Except this time the change isn’t simply another website or another lead source.
The customer may be bringing their own intelligence into the transaction.
An intelligence that doesn’t get tired.
Doesn’t forget.
Can compare hundreds of options.
Can read thousands of pages.
Can remember every number you’ve given it.
Can calculate complicated transactions instantly.
And can work continuously on behalf of the customer.
That’s different.
What should dealers do right now?
I’m not suggesting dealers run out tomorrow and buy another piece of software.
Actually, I’d suggest the opposite.
Start by understanding what’s happening.
Download and experiment with personal agents.
Have your leadership team use them.
Ask one to research a vehicle.
Ask one to compare vehicles.
Ask it to value a trade.
Give it a hypothetical deal and ask it to analyze it.
Ask it to compare financing.
Ask it to evaluate a service recommendation.
Then look at your dealership from the agent’s perspective.
Can it find accurate information about you?
Can it understand your inventory?
Can it determine your real pricing?
Can it understand your fees?
Can it tell whether a vehicle is actually available?
Can your dealership respond intelligently when an AI agent contacts you?
Can your service recommendations stand up when an AI independently checks them?
Those are the questions I’d start asking.
Not because every customer is doing this today.
They’re not.
Because enough of them will eventually do it that dealers should understand the technology before it changes consumer behavior around them.
The next digital customer may not look like a lead
That’s probably the biggest point I want dealers to take away from this.
We have built an enormous automotive technology ecosystem around getting a human being to:
Search.
Click.
Browse.
Submit.
Call.
Text.
Visit.
Buy.
Personal agentic AI has the potential to insert an intelligent representative between Searchand almost everything that follows.
And once consumers become comfortable saying:
“Handle this for me.”
a lot of assumptions we’ve built the modern automotive retail funnel around are going to have to be reconsidered.
I don’t know exactly how quickly that happens.
Nobody does.
There will be privacy issues, trust issues, mistakes, regulation and plenty of technology that doesn’t work nearly as well as advertised.
But after watching technology change this business for three decades, I wouldn’t dismiss this because it’s early.
I’d learn about it precisely because it’s early.
Dealers don’t need to panic.
But they do need to wake up.
Because the next major disruption in automotive retail may not be another company trying to sell cars differently.
It may be your customer showing up with an AI agent that’s better prepared for the transaction than they have ever been before.
— Todd Caputo
