CarGenius AI's Corey Lydstone joins the Greenfield podcast

Our co-founder and CEO Corey Lydstone joined Steve Greenfield on the Greenfield podcast to discuss the next major platform shift in automotive retail and what it means for dealers.

Listen to or watch the full episode here.

Corey has seen a shift like this before. He built Autolist during the move from desktop to mobile, grew it into one of the leading mobile car shopping platforms, and later sold the business to CarGurus. His view is simple. AI will reshape car shopping even more than mobile did.

Here is what the conversation covered.

Consumers are already shopping with AI

The behavior shift is already underway.

Shoppers are using ChatGPT, Claude, Grok, and other AI tools to research vehicles. They ask which third row SUV fits their family. They compare trims. They try to understand features, pricing, incentives, and what they should do next.

In the old journey, a shopper visited several marketplaces, watched YouTube reviews, read forums, compared dealer sites, and then submitted a lead. AI compresses that journey. A shopper can now ask one question and get a clear, personalized answer.

That changes where influence happens. It also changes which businesses sit closest to the customer.

Marketplaces were built around trust, transparency, and aggregation. Those are still valuable. But AI is starting to take over more of that role. When a consumer trusts an AI assistant to help choose the right vehicle, compare options, evaluate dealers, and understand price, the traditional intermediary has to work much harder to stay central to the funnel.

Why marketplaces face a harder road

Marketplaces still deliver real ROI today. Dealers should not ignore them. That is where much of the audience still is.

But the long term pressure is real.

Consumer acquisition is getting harder. SEO is more competitive. AI answers are changing how consumers discover information. As more shopping starts inside AI interfaces, some of the traffic that used to flow through search engines and marketplaces will move elsewhere.

There is also a structural issue. A marketplace has to put its own brand forward, which means the dealer's brand is secondary. That creates tension. Dealers want leads and visibility, but they are understandably cautious about giving more data, more access, and more control to a platform that may also be competing for the customer relationship.

That is not only a business model problem. It is a product problem. The best shopping experience needs richer information, tighter integrations, and real time access to dealer specific tools. That is hard to build when dealers do not fully trust the intermediary.

What CarGenius AI is building

CarGenius AI connects dealers directly to AI agents. We do that through MCP servers.

MCP stands for Model Context Protocol. It is an open standard introduced by Anthropic that helps AI assistants connect to external systems, data, and tools in a structured way. In plain English, an MCP server lets an AI understand what a system can do and then use it.

For dealers, that matters because their websites already contain the information AI agents need: inventory, vehicle details, dealership policies, lead paths, service information, financing content, and trade-in options. The problem is that this information is usually trapped inside websites built for humans, not for AI agents.

We build an MCP server for each dealer. It lives on the dealer's domain, for example mcp.dealership.com. That server gives AI agents access to approved tools, including:

• Inventory search, so an agent can find matching vehicles in seconds instead of scraping a website
• Dealership information, so an agent understands the store, its policies, and its process
• Vehicle details, so an agent can retrieve the full VDP
• Lead handling, so a shopper can start a real conversation with the store

AI agents are information hungry, and dealers already have much of the best information. A Carfax report is a simple example. You cannot always get that from a marketplace, but many dealers already have it on their own VDPs. With our Carfax integration, that report can travel with the vehicle directly to the AI agent.

This is only the beginning. We are working with early dealers to connect financing, trade-in, and service. Service matters because a large share of dealer website traffic is service related. An AI can already answer a question like what a dashboard light means. The next step is helping the customer schedule the right appointment.

AI search on the dealer's own site

We also deploy AI search directly on the dealer's website.

Think of it as ChatGPT for the dealership's own domain. It is not a traditional chat widget, and it is not gated behind a lead form. It answers real shopper questions using the dealer's inventory, website content, and approved tools.

Dealers tell us it works like their dream salesperson. It handles specific questions that traditional chat tools struggle with. In one case, an international shopper asked whether they could reserve a vehicle without US credit. The AI answered clearly, explained the next steps, and turned the conversation into a lead for the sales team.

There is another benefit: visibility. Dealers finally get to see how consumers interact with AI. They can see every query. They learn what shoppers ask, how the AI responds, where the dealership's information is clear, and where it is missing or wrong.

We have already found real issues this way. Some website providers leave important pages out of their sitemap, which means AI tools may never see them. We identify that quickly and help fix it. That helps the dealer today with AI visibility, and it can improve how Google understands the site too.

Where this is heading

Right now, when a shopper asks ChatGPT for help buying a car, the AI often has to rely on scraped information from marketplaces and editorial sites. That is slow, incomplete, and it may miss the best source of truth, which is the dealer's own information.

As more dealers turn on MCP servers, that changes.

We are not just helping one dealer expose data to AI. We are standardizing how AI can search across connected dealers. Ask for a gray Porsche 911, and the system knows that Chalk may be a match. Ask for a family SUV with captain's chairs, and it searches inventory using the language consumers actually use, not rigid filters.

That is the future. Structured dealer data, made usable by AI agents, with richer context than scraping can provide. It is better for the consumer, better for the dealer, and better for the AI platforms.

That creates a flywheel. As the data gets richer, AI shopping gets better. As AI shopping gets better, more consumers use it. As more consumers use it, dealers need to make their information accessible there.

Agents buying cars is coming. When it arrives, many old tactics will stop working the way they used to. Dealers will compete more on process, trust, transparency, and long term customer relationships. That is good news for strong operators. Most great dealers already want to win that way. They just need the infrastructure to meet consumers where they are going.

The first step for dealers

Corey's advice is simple. Start now.

Ask ChatGPT, Claude, or Grok about your dealership. Ask about your inventory. Ask about your service department. Ask why someone should buy from you instead of a nearby competitor. Then ask yourself whether the answer is accurate, complete, and useful. If it is not, you have work to do.

The next step is to get an MCP server set up for your store. CarGenius AI will set one up free for any dealer, and you do not pay until AI actually uses it. We want every dealership in the country to have this infrastructure, because that is how the industry starts learning what is coming.

The five year vision

Corey described the vision this way. CarGenius AI wants to be the Stripe of automotive commerce inside AI.

Stripe powers online payments in the background. You do not see its brand at checkout. It just works. That is the role we want to play for automotive retail.

We are not building our own consumer agent. We are not competing with dealers. We are not becoming another marketplace. Our job is to be the infrastructure layer that lets consumers and dealers act inside AI. As models improve, we use the best ones available. As AI platforms evolve, we connect dealers to them. As consumers shift their behavior, we make sure dealers are ready.

Listen to the full conversation on the Greenfield podcast.

Then take the two steps that matter. Test your dealership in ChatGPT today, and get your free MCP server set up. Email Corey at corey@cargenius.ai.

‍