Why Automotive Needs a Commerce Protocol

What happens after the conversation ends

Over the last year, I have spent a lot of time looking closely at how AI conversations move through dealership systems. Not just the AI itself, but what happens after the conversation ends. And something keeps showing up.

A customer might have a 15 to 20 message interaction with a dealership's AI assistant. In that conversation they share the vehicle they want, a trade-in with a remaining payoff, a target monthly payment, financing questions, and a preferred appointment time. From the customer's perspective, it feels like a real conversation with someone who understands them.

But once that interaction enters dealership systems, all of that context collapses into a simple lead record. Name. Phone number. Email. Vehicle of interest. The details that actually matter -- the ones that would make the next conversation feel like a continuation rather than a restart -- disappear somewhere between systems.

The more I looked at it, the clearer it became that this is not an AI problem. It is an infrastructure problem.

The industry was built around a different type of interaction

For almost two decades, automotive lead infrastructure has relied on ADF, the Auto-Lead Data Format. ADF created a common language so dealership websites, CRMs, and third-party tools could exchange lead information. At the time, most leads were simple form submissions: name, email, phone, vehicle. ADF solved a real interoperability problem, and the industry adopted it widely.

But the nature of the interaction has changed. Today a lead often is not a form submission at all. It is a conversation. And a conversation carries far more context than a traditional lead record. Intent signals. Payment expectations. Trade-in details. Timing signals. Behavior patterns.

When that context is lost between systems, the dealership ends up working with a partial view of the customer. The AI captured the information. But the infrastructure was not designed to carry it. So the next system in the chain starts again from the surface.

The missing layer

If automotive software is going to support conversational commerce -- whether that is AI assistants, messaging, or new digital retail tools -- the industry needs a better way for systems to pass interaction context between each other. Not just basic lead fields. The actual structure of the interaction itself: conversation transcripts, customer intent signals, deal stage indicators, trade-in details, payment targets, appointment signals.

Something closer to how modern platforms exchange structured interaction data. This is the layer that does not exist yet.

What ACP is designed to solve

This is what led me to start building the Automotive Commerce Protocol. ACP is an open, JSON-based specification designed to carry richer interaction data between automotive systems while staying backward compatible with existing ADF pipelines.

ACP structures every interaction into an envelope that carries the full conversation context: what the customer said, what they want, what stage they are at in the buying process, and what the next system needs to know to pick up where the conversation left off. The envelope travels with the lead through every system in the stack.

The goal is not to replace existing systems or standards. And it is not about one company owning the infrastructure. ACP is open, vendor-neutral, and designed to work alongside the formats already in place. It reflects the reality that today's customer interactions are much more than simple lead forms.

This is a Design phase problem

In the AI Operator Framework, the first phase is Design. Before you deploy any tool, you audit the system and map where AI creates real leverage. The infrastructure gap in automotive is a textbook Design phase problem. Dealers are deploying AI tools without first designing the data architecture those tools need to be effective.

Better AI on top of broken infrastructure does not produce better outcomes. It produces faster context loss. The system has to be designed before the tools get deployed.

The question for anyone building or buying

AI tools will continue to improve. But if the systems carrying the interaction still reduce everything to a handful of fields, the industry will keep discarding the most valuable part of what AI actually produces.

If automotive retail is moving toward conversational commerce, the underlying infrastructure has to evolve with it. That is probably a conversation worth having as an industry. And it is one I intend to keep pushing.