The first wave of AI in commerce focused on evolution—better recommendations, smarter personalization, and more responsive chatbots. What’s emerging now is categorically different. We’re entering an era where AI agents don’t just assist the shopping journey; they shape it, orchestrate it, and increasingly carry it out on the user’s behalf.

But the real transformation isn’t the agents themselves. It’s the invisible architecture forming beneath them, a new substrate for digital commerce that rewrites how discovery, evaluation, and decision-making happen.

This shift is subtle, structural, and easy to underestimate. Yet it will redefine how brands compete, how consumers behave, and how the entire commerce ecosystem measures success.

From Clickstreams to Intent Streams

For decades, digital behavior has been measured through clickstreams, the familiar trail of taps, scrolls, searches, sessions, and pageviews. Clickstreams worked because the interface was the journey, actions were observable, and every touchpoint left a trace.

AI agents break this paradigm as agents don’t click; they reason, negotiate, retrieve, and decide.

A user might express a single prompt, “Find me a good carry-on suitcase under $300,” and the agent performs dozens of micro‑actions behind the scenes:

None of these steps is visible in a traditional analytics model. They don’t show up as clicks or pageviews. They exist as intent shifts inside the agent’s reasoning loop.

This means the next generation of analytics must evolve from tracking what users do to understanding how agents interpret what users want.

That includes:

The unit of analysis is no longer the “session.” It’s the conversation, not between user and interface, but between user and agent, and increasingly, between agents themselves.

This is a profound change, and it requires new instrumentation, new mental models, and new ways of defining what a “journey” even is.

The Rise of the Interoperability Layer

If AI agents are the new interface, interoperability is the new infrastructure.

Google’s Universal Commerce Protocol (UCP) is one of the earliest and clearest signals of this shift. It’s a foundational shared language that allows agents to access product data, inventory, pricing, and checkout across platforms.

Why does this matter?

Because without a common protocol, agents remain siloed. UCP and the standards that will inevitably follow act as connective tissue across the commerce ecosystem. They turn fragmented touchpoints into a continuous, cross-platform experience. They enable agents to:

This is the difference between a chatbot that answers questions and an agent that can do things.

Interoperability is the missing layer that makes agent-mediated commerce viable at scale. It’s the equivalent of what HTML did for the early web by being a shared foundation that unlocks entirely new behaviors.

Agents Are Becoming the New Interface

As agents take on more cognitive loads, the interface becomes less important. The “journey” stops being a sequence of clicks & views, and starts becoming a sequence of decisions.

A shopper might begin with a vague prompt: “I need a gift for a friend who loves cooking, is in her mid 30s, and prefers baking.”

From that moment, the agent handles the heavy lifting:

The user sees only the final recommendations and not the dozens of microsteps that led to it. This compression of the journey has several implications:

In an agent-mediated world, the most important interactions happen between agents, not between users and interfaces.

The New Competitive Landscape

If agents become the primary decision‑makers, the competitive dynamics of commerce change dramatically.

1. Brands compete for agent trust, not user attention.

Search rankings, ad placements, and homepage features matter less when the agent is the one choosing what to show the user. The question becomes:

How do you become the product an agent recommends?

That depends on:

2. Product data becomes a strategic asset.

In an agent-mediated ecosystem, messy or incomplete product data is a growth limiter. Agents need:

The brands that invest in data hygiene will outperform those that don’t.

3. Journeys become dynamic, contextual, and non-linear.

There is no “funnel” when the agent can jump from discovery to evaluation to purchase in a single reasoning loop. The journey becomes:

This requires new frameworks for understanding behavior, ones that treat the journey as a fluid conversation rather than a fixed sequence.

4. Analytics must evolve from tracking clicks to interpreting intent.

The next generation of analytics will need to answer questions like:

This is a shift from behavioral analytics to cognitive analytics, understanding not just what happened, but why.

A New Lens for Understanding Commerce

We’re moving toward a world where the shopping journey is not a path but a conversation mediated by agents, shaped by interoperability, and driven by intent rather than interaction.

To understand this world, we need new mental models:

The rewrite of digital commerce is already underway, and the companies that thrive will be the ones that understand the hidden architecture forming beneath the surface and adapt before it becomes the new default.

Pretty interesting, right? There’s a lot more to unpack, and I’ll dive into it next. More on this soon.