Got Agent?

As AI agents take on more of the vehicle-buying process, who controls the information they give customers? Jim Lawrence examines what dealerships need to know about data ownership, vendor contracts and preparing their systems for agentic commerce.

What if “Agentic Ads” Could Sell Vehicles for Auto Retailers?

AI’s expanding agent-to-consumer (A2C) and agent-to-agent (A2A) capabilities could trade “the click” for “the close”

For 25+ years, digital advertising in retail automotive has run the same basic course: an ad earns a click, the click lands on a vehicle detail page (VDP), and a lead is sent to the customer relationship management system (CRM). A trained business development center (BDC) rep, Internet salesperson, or finance and insurance (F&I) manager then takes over to work the deal.

That process is being redesigned, from inventory visibility and consumer contact to trade-in, the F&I menu, the four-square deal worksheet and payment. The direction is toward agentic ads: AI agents that can carry a conversation forward as each step resolves into the next, potentially all the way to the close. Dealers may be closer to that shift than they realize.

The Tech Stack Evolving in the Digital Showroom

Between September 2025 and mid-2026, the technology and payments industries shipped a stack of competing standards designed to let AI agents complete purchases on a shopper's behalf rather than simply pointing to a website. OpenAI and Stripe launched the Agentic Commerce Protocol (ACP) in September 2025, powering ChatGPT's Instant Checkout for a small group of Shopify and Etsy merchants. Google and Shopify followed in January 2026 with the Universal Commerce Protocol (UCP), covering discovery through post-purchase support; its technical council now includes Amazon, Meta, Microsoft, Salesforce and Stripe, with Visa, Mastercard, American Express, Best Buy and Macy's among more than 20 endorsing partners.

Google's Agent Payments Protocol (AP2) launched in September 2025 with more than 60 partners. Visa's Trusted Agent Protocol and Mastercard's Agent Pay — including a version built for machine-speed microtransactions — moved from pilot to commercial rollout by mid-2026. Visa also struck a direct integration with OpenAI in June 2026 to let ChatGPT agents pay with tokenized Visa credentials: digital substitutes for the underlying card details.

In short, although this looks like the inside of a Tauntaun — the Luke Skywalker’s unfortunate overcoat from The Empire Strikes Back — it is happening in some form across the entire economy. These A2A retail sales protocols are designed to help justify the stock and equity valuations that support the hundreds of billions in fraternal debt taken on by 10 tech players for their quarterly revenue reports.

The AI build-out avalanche across a short list of tech companies with once clean balance sheets and not-quite-illegal monopolies is a sensation, with companies trying to remove people so management can get their bonuses before the customer satisfaction index (CSI) scores get reported. Author’s Note: I am not cynical at all.

Given consumer attitudes toward the current dealer vehicle purchase experience provided by the bushels of “bad apples,” dealers cannot ignore the lack of trust or the resulting regulatory attention. Think of California’s CARS Act, FTC legal lashings and nasty grams to 97 dealer groups.

Dealers will need to facilitate this A2A process as much as practicable, since polls show a positive outlook from consumers interacting with AI bots for major portions of the digital sales process already. At least that is what the AI vendors are polling.

As of now, dealer success at AI adoption has been uneven, and attributing a return on investment (ROI) to AI’s ability to scale has been difficult. The deciding factors include training and readiness, properly documented business processes to be “AI’ed,” and data brought together from multiple silos in a consistent format.

Success does not come from management’s deep desire to deploy cutting-edge AI demos that, like historical military plans, fail upon contact with the enemy.

What Agent to Agent Means for a Dealership

Automotive has a dedicated, published specification: the Auto Agent Protocol (AAP), a profile built on top of Google's open Agent2Agent (A2A) standard specifically for franchise dealers. Think of it as a shared set of instructions for exchanging dealership information. AAP defines five standardized transactions:

• Dealer Information
• Inventory Search
• Vehicle Detail
• Inventory Facets — the filters used to narrow a vehicle search
• Lead Submission

The automotive A2A specification is designed to move real inventory, real pricing and valid leads between a shopper's AI agent and a dealership's own systems. Notably, the spec requires that the price an agent receives be the advertised vehicle price, including mandatory dealer charges but excluding tax, title and registration, a rule its authors describe as aligned with FTC guidance against bait-and-switch pricing as well as California’s CARS Act.

It also allows a submitted lead to convert into ADF/XML, the structured lead format most dealer CRMs already accept — unless they are trying to limit outside CRM usurpers.

AAP is early and voluntary, so no OEM or dealer group has mandated it (yet). Its existence provides a foundation for agent-to-agent interactions built on vehicle identification numbers (VINs) and available inventory. Those information exchanges and lead submissions are not, by themselves, a completed vehicle purchase. In short, the infrastructure is taking shape.

“In short, this is happening...” In Automotive and Beyond

Automotive dealers can take this trend seriously by looking at how fast AI A2A has moved into other industries characterized by high-priced purchases. In travel, Sabre, PayPal and others announced a partnership in February 2026 aimed at end-to-end agentic booking across more than 420 airlines and two million hotel properties. Google has also been building agentic hotel booking directly with Booking.com, Expedia, Marriott, IHG and Choice Hotels.

Yet as of June 2026, fewer than one in eight hotel companies could actually complete a booking through an agent in real time. The gap wasn't consumer demand, industry analysts note, but how much of each hotel's inventory and pricing existed in structured, agent-readable data…sound familiar?

In residential real estate, the National Association of Realtors' 2025 technology survey found the vast majority of agents already use some form of AI in their workflow. Multiple listing services are now treating structured, machine-readable data as core infrastructure rather than a back-office concern, with protocols such as the Model Context Protocol connecting AI systems to data and tools.

In general retail, more than a million Shopify merchants such as Glossier and Vuori, alongside sellers on the separate Etsy marketplace, are already transacting through ChatGPT Shopping. Stripe's Agentic Commerce Suite counts Coach, Kate Spade, and Ashley Furniture among early adopters.

The pattern repeats in every vertical examined: agents transact fastest, and most accurately, where a business's inventory, pricing and terms already exist in a consistent, machine-readable form, without a human translating them first. Where that data still lives in silos, PDFs, spreadsheets or a salesperson's head, agents may skip the business or, worse, answer for it inaccurately. Yikes!

The Age-Old Question in Automotive Retail — Who Owns the Data Behind the AI Answers?

This is where the A2A trend becomes a bottom-line issue rather than an IT curiosity. Whichever system holds the structured data — the compliant offer language, inventory, clear pricing, trade-in value and credit-status prompts — supplies the substance an agent uses to answer a shopper.

If that substance lives inside the dealership's own dealer management system (DMS) and/or CRM, the dealer has a direct place to manage those answers. If, on the other hand, the dealer's A2A platform lives inside a third party's proprietary software-as-a-service (SaaS) environment, the vendor's business model can influence what gets said on the dealer's behalf. During a legal event, the FTC would expect a lottasplaining to do by all involved.

The dealer may still face the consequences of a non-compliant answer. That is why vendor contracts need to address responsibility, data access and risk, rather than leave those questions until something goes wrong. The FTC Safeguards Rule also makes oversight of service providers a consideration.

Have the hard conversation now: what happens if the vendor changes its terms, pricing, ownership or AI stack? What would give you grounds to end the relationship, and what could you take with you?

How I Am Reacting — Two Ends of the Pipeline, Not Quite an Ouroboros

I am approaching this shift from two ends of the sales process through businesses I founded: ImagineMyDreamCar.com and DealerEFX. Here is how I am putting “agent-ready” into practice. And no, the pipeline is not quite an ouroboros, the snake eating its own tail.

On the consumer research and discovery side, I built ImagineMyDreamCar.com (IMDC), an AI-driven, hyper-personalized vehicle lead generator aimed at high-net-worth luxury buyers. It generates a unique, persistent URL that AI systems can read for every AI vehicle render, carrying structured metadata on model, trim, color and feature preferences. Those preferences are catapulted into buyers’ social networks whenever they share a photograph of themselves with their AI-rendered dream cars.

I designed IMDC's architecture to be agent-amenable ahead of any formal protocol requirement. The same structured data that feeds search engine optimization and answer-engine optimization could, in principle, give a shopping agent a useful preference signal, rather than generic inventory, copy-paste content or self-aggrandizing reviews.

That is how I see IMDC fitting into the emerging A2A ecosystem and its automotive-specific Auto Agent Protocol extension. It is a directional opportunity, not an industry-wide compliance deadline like those associated with the FTC Safeguards Rule or Red Flags Rules.

On the lead-response side, my work with DealerEFX addresses what happens after an inquiry arrives, whether it comes from an autonomous AI agent or a human-assisted process. We build and maintain OEM-compliant, year-make-model-specific email templates with calls to action, including trade-in valuations, credit applications and service promotions. They are designed to deploy and run inside the CRM rather than a separate, proprietary platform.

That distinction matters for the ownership question above: the aim is to keep content, disclaimers and a consistent brand voice in the dealer's CRM, as the dealer's asset, regardless of which AI Email Response platform, search engine or agent protocol ultimately delivers the customer.

Neither of my companies currently plugs directly into ACP, UCP or AP2. The Auto Agent Protocol itself remains an early, community-published specification without an announced industry mandate. What I am working toward is a practical definition of readiness: structured, portable, dealer-owned data and content, built now, ahead of any formal requirement to have it.

Avoid Being Gutted Like a Tauntaun, Metaphorically Speaking

New and CPO/CUV dealers do not need to chase a “Betamax vs VHS” protocol war that titans in payment networks and AI platforms are still fighting among themselves. They do need to ask harder questions of every vendor touching their content, inventory feed or customer data:

Where does your compliant lead-response and deal content actually live — in your CRM, or inside a vendor's platform? Would you lose access if the contract ended?

Can you export your structured data in a usable form, or is it locked in a proprietary format?

If a customer's AI agent asked about your inventory and pricing today, could it get an accurate answer, or would it have to guess from a stale feed of unclear pricing and old inventory data?

Who, contractually, owns the preference and engagement data generated at the moment of highest buyer intent, when a customer submits a lead form?

This is critical. Know who owns the data and how easily you can access it after the contract ends.

None of these questions require a dealership to bet on a winner between ACP, UCP, AP2 or the Auto Agent Protocol. They require dealers to retain control of the substance behind their website ads: vehicle inventory, clear and compliant pricing, documented business processes, and the reasons a customer should buy from them rather than the store two exits down the highway.

That is still the manna from heaven that comes with good business practices: knowing what your dealership can offer, and making those benefits clear to customers.