MCP from CDK & Reynolds and other DMS companies

[Alex Snyder]

How long do you think it will be before CDK and Reynolds give dealers an MCP connection? Someone told me Tekion is forgoing some dashboards they promised for an MCP upcharge. I have to think that will pressure the other DMS companies.

What's your over/under on time for this?

 

[Jeff Kershner]

This sure would be nice. As for a time frame... you know they are going to holdout as long as they can.

 

[Alex Snyder]

This sure would be nice. As for a time frame... you know they are going to holdout as long as they can.

I found your LinkedIn post about this thread funny. All the responses are from vendors overcomplicating the question. A few sounded offended that the question was even posed. I wonder what they fear

 

[Ron Clark]

would be curious to hear your thoughts on what MCP is at this point. lots of variations I have heard.

 

[Alex Snyder]

would be curious to hear your thoughts on what MCP is at this point. lots of variations I have heard.

I think the MCP connection is quite clear as the AI solutions define it.

MCP stands for Model Context Protocol. It’s an open standard that lets AI applications connect to external tools, data, and software through a common interface.

Think of it as a universal connector for AI: instead of building a separate custom integration for every AI assistant and business system, developers can expose a system’s capabilities through MCP so compatible AI applications can discover and use them.

How it works

• The AI application is where you make a request.
• An MCP client within that application connects to an MCP server.
• The server exposes capabilities—such as searching documents, looking up a customer, or executing an action—and describes how to use them.

For example, a CRM’s MCP server could let an assistant look up customer information or prepare an update using natural-language requests. What the assistant can actually do depends on which capabilities the server exposes.

How it differs from an API

An API provides access to a particular service. MCP provides a standardized way for AI applications to discover and use capabilities, often by calling that service’s existing API behind the scenes. It doesn’t replace the API; it adds a consistent, AI-facing interface.

MCP is the connection standard—not the AI model or the agent deciding what to do.

For a dealer scenario, I like to think of it in levels of data exposure. One level is where the numbers the dealer can already see in a dashboard are received via MCP to help build a larger dealership dashboard and make month-end reporting more efficient.

Things get deeper when the data behind the numbers is available, thus giving the dealer the ability to act on it. That can be as simple as a list of customers, cars, parts, services, or dollars, depending on what the system specializes in.

And then there is deeper access, like how customers engaged, click paths, pricing changes, etc. This is where dealers can be empowered to vibe-code their own solutions.

@David Steinberg just released a very cool MCP at FourEyes. He's given dealers deep enough data levels to build solutions.