By Michael Quigley, President & Chief Strategy Officer, Co-Founder at Impel
The current AI conversation in automotive is about dealers deploying AI to talk to customers. That conversation is already dated. The next wave — agentic AI in automotive retail — is fundamentally different. The customer will have their own AI agent too. And when both sides of the transaction are AI-mediated, the rules of engagement change entirely. The question isn’t “how should we use AI?” It’s “is our agent smarter than theirs?”
The buyer’s AI is already in the room
According to the 2025 Cox Automotive Car Buyer Journey Study, 19% of all buyers and 25% of new-vehicle buyers are already using AI tools, including ChatGPT and Google AI Overviews, during the vehicle shopping process. These buyers arrive more informed, with specific questions, with expectations already shaped by an AI that has already researched and comparison-shopped for them.
In fact, the front door of consideration (and even negotiation) has moved. And most dealers don’t know it yet.
This isn’t a minor behavioral shift. It’s a structural change in how the purchase journey begins. The buyer’s agent is conducting research, filtering options, and forming preferences, all before the first touchpoint with your team or your AI. By the time they reach out, the consideration set may already be narrowed. If you weren’t in it, you lost the deal before it started.
Agentic AI in automotive retail: already in motion
This evolution is even more significant. AI agents don’t just advise anymore. They act. The consumer’s agent queries multiple dealerships simultaneously, compares responses, evaluates tone and responsiveness, and surfaces the best option. All before the human buyer makes a single phone call.
Furthermore, Gartner predicts that by 2028, 60% of brands will use agentic AI as persistent digital concierges spanning marketing, sales, and customer service. The infrastructure is here. The behavior is forming fast.
The data points to where this is going. But one dealer in Salt Lake City has already gone there. As written about in Car Dealership Guy, in May 2026 a general manager at Mark Miller Subaru built an AI agent to shop his own store without telling his sales team. His staff spent four hours negotiating with a machine, and they didn’t know it. That wasn’t a proof of concept. That was a live transaction. And by his estimate, AI agent-to-agent deals already represent about half a percent of automotive volume, roughly where internet deals were in the late 1990s.
The dealerships treating this as a future problem are making the same mistake as the ones who thought internet leads were a fad.
Your first response is now a machine-readable signal
Here’s where it gets concrete. When the buyer’s agent is evaluating your dealership, your first response isn’t a message to a person. It’s a signal to an algorithm.
Response speed. Relevance. Accuracy. Personalization. These aren’t soft metrics anymore. They’re machine-readable criteria that determine whether your dealership stays in the consideration set or gets routed around entirely. The buyer’s AI is scoring your dealership before any human is involved.
That reframes everything about how dealers should think about their first response. It’s not just about sounding friendly. It’s about being fast, accurate, and contextually relevant in a way that satisfies both the human and the agent evaluating on their behalf.
The dealer whose AI responds fastest, most precisely, and most helpfully wins the transaction before a human is ever involved. The dealer without an agent, or with a generic one, doesn’t even make the shortlist.
The asymmetry
This is the competitive question the industry hasn’t fully confronted yet.
For decades, the advantage in automotive retail was location, inventory, brand, and relationships. Dealers held all the cards. Those variables still matter. But in the agentic era, a new variable enters the equation: who has the smarter agent on their side of the table.
Deeper data. Better customer context. Greater sentiment and intent decoding. More sophisticated agent behavior: goals, tone, escalation logic, brand voice, all deliberately configured. As I explored in The High-Stakes Reality of AI Exposure, dealerships that aren’t AI-enabled are already paying more for the same output. AI-fluent competitors are pulling further ahead — and that margin gap doesn’t stay constant.
Deloitte’s 2026 Tech Trends research found that while 38% of organizations are piloting agentic AI, only 11% are actively running it in production. The gap between exploring and executing is where competitive advantage is won or lost. The dealerships on the right side of that gap are already pulling ahead.
We’re already seeing this play out with many of the retailers Impel serves. Hundreds of dealerships are actively programming their AI agent behavior, defining how it responds, when it escalates, what it prioritizes. That’s not AI deployment. That’s agent strategy. And it’s the early signal of what will become table stakes.
The asymmetry compounds
The dealership with the smarter agent doesn’t just win today’s transaction. It builds a structural advantage that grows over time. In Centaurs, Chess, and the State of Vertical AI in 2026, I explored why the human-agent combination consistently outperforms either alone. That advantage only holds when the agent is trained, configured, and running. Every interaction makes it smarter. Every month it isn’t running is a month it isn’t learning.
That’s the compounding layer most dealers aren’t accounting for. As I argued in The Down Escalator Problem, the cost of delay isn’t linear. The dealership without an agent, or with a dumb one, loses out before they even know there was an opportunity. And by the time they catch up, the gap may already be structural.
The provocation
For decades, the customer walked in. You had the home field advantage: the lot, the showroom, the relationship, the closer.
In the agentic era, the first interaction happens between machines. Your agent versus theirs. And if yours loses that exchange, if it’s slower, less accurate, less contextually intelligent, the human buyer may never walk in at all.
The question isn’t whether agentic commerce is coming to automotive. It’s already here, at the edges, and will move fast toward the center. The question is whether your dealership is ready to compete.
Because the buyer’s agent isn’t waiting for you to catch up.
Want more perspectives on vertical AI, automotive, and strategy? Subscribe to Drive the Future.
What dealers are asking
What is agentic AI in automotive retail?
Agentic AI in automotive retail refers to AI systems that don’t just respond to questions — they take autonomous actions on behalf of buyers or dealerships. On the buyer side, an AI agent researches vehicles, compares dealerships, and narrows the consideration set before any human is involved. On the dealer side, an AI agent handles inbound inquiries, qualifies leads, books appointments, and follows up — all without staff intervention. When both sides of a transaction are AI-mediated, the dynamics of automotive retail change fundamentally.
Are buyers already using AI agents to shop for cars?
Yes. According to the 2025 Cox Automotive Car Buyer Journey Study, 19% of all buyers and 25% of new-vehicle buyers are already using AI tools — including ChatGPT and Google AI Overviews — during the vehicle shopping process. These buyers arrive with research already done, preferences already formed, and a consideration set already narrowed. In some cases, the dealer that didn’t make that list lost the deal before the first touchpoint.
What is agentic commerce and how does it affect car dealerships?
Agentic commerce is when AI agents act — not just advise — on behalf of consumers. In automotive, a buyer’s AI agent can query multiple dealerships simultaneously, compare response speed and relevance, evaluate tone, and surface the best option before the human buyer makes a single call. Gartner predicts that by 2028, 60% of brands will use agentic AI as persistent digital concierges across marketing, sales, and service. For dealerships, that means the first impression is no longer made to a person — it’s made to an algorithm.
What makes one dealership’s AI agent smarter than another’s?
The differentiating factors are data depth, configuration, and runtime. A smarter agent has access to more customer context — vehicle history, prior interactions, sentiment signals, intent data — and is deliberately configured with goals, tone, escalation logic, and brand voice. Crucially, it’s also been running long enough to learn. Impel’s AI Operating System gives dealerships the ability to program agent behavior at that level of specificity, across every stage of the customer lifecycle. An agent that hasn’t been configured and running isn’t just behind — it isn’t learning.
Has AI-to-AI car buying actually happened?
Yes. In May 2026, a general manager at Mark Miller Subaru built an AI agent to shop his own store without telling his sales team. His staff spent four hours negotiating with a machine and didn’t know it. By his estimate, AI agent-to-agent deals already represent roughly half a percent of automotive volume — approximately where internet leads were in the late 1990s. Dealers who treated internet leads as a fad learned an expensive lesson. The pattern is repeating.
How should dealerships prepare for the agentic AI era?
The starting point is having a configured, running AI agent — not a generic one. A buyer’s AI is evaluating response speed, accuracy, relevance, and personalization as machine-readable criteria. Dealerships that don’t have an agent, or have one running on default settings, won’t make the shortlist. Impel’s AI Operating System enables dealerships to program how their agent responds, when it escalates, what it prioritizes, and how it represents the brand — across voice, chat, sales, and service. That’s not AI deployment. That’s agent strategy.