Agentic commerce has an infrastructure problem

There’s no shortage of compelling demos of agentic commerce. Instead of browsing and checking out manually, shoppers interact with an AI agent that discovers products, compares options, and completes purchases on their behalf.

But when these experiences meet real commerce systems, the reality looks different. Pricing, inventory, fulfillment, fraud checks, and payment authorization all run on infrastructure built for a different model of commerce — one where a human navigates a website and completes a single checkout event. Agentic commerce changes that model.

Agents query systems continuously. They check pricing, availability, shipping options, and product details throughout a session. In more advanced cases, they may maintain context across multiple interactions and complete transactions autonomously.

Most commerce infrastructure wasn’t designed for this type of continuous interaction.

At Adyen, we sit at the layer where AI interfaces connect to enterprise commerce systems. Over the past several months we’ve been working on infrastructure at this layer, speaking with more than 200 global enterprise merchants and engaging with many of the platforms building agentic experiences. One pattern has become clear: infrastructure is what’s limiting agentic commerce today.

In our new paper, Agentic Commerce Has an Infrastructure Problem, we outline five structural constraints that must be addressed before this model can scale:

• Protocol fragmentation across emerging AI commerce interfaces

• Product data systems that machines can’t reliably query

• Enterprise commerce stacks designed for linear checkout flows

• Fraud and trust frameworks built for human-initiated transactions

• Merchant onboarding models that don’t scale across platforms

None of these challenges are insurmountable. But solving them will require changes across the ecosystem — from AI platforms and payment networks to merchants and commerce infrastructure providers.

For merchants, the takeaway is straightforward: preparing for agentic commerce is less about integrating with a specific AI platform and more about strengthening the systems that power commerce today — product data, APIs, inventory, and identity. These investments improve existing channels today while positioning merchants for new interaction models tomorrow.

Read the full paper: Agentic Commerce Has an Infrastructure Problem.