Agentic commerce and the evolution of retail purchase behaviour

The way commerce works is undergoing its most significant transformation since the inception of the internet.. For two decades, brands optimised for search engine algorithms. Now they face a more fundamental challenge: becoming intelligible to AI.

Discovery is shifting from links to conversations. 

Products are no longer found through search results alone but surfaced, evaluated, and recommended within platforms like ChatGPT, Google AI Overviews, Gemini, and Perplexity. For the first time, AI is interpreting product catalogues, comparing attributes, and influencing purchasing decisions.

This isn't speculative. McKinsey estimates AI agents could orchestrate up to $1 trillion in US B2C retail by 2030, with global projections reaching $3 to 5 trillion. This is one of the largest channel shifts in retail history.

The Evolution of the AI Buyer

To understand where commerce is heading, it helps to map how AI's role is expanding across three phases.

Phase 1: The Researcher (Now)

Today, AI acts as a capable advisor. When a consumer asks for "running shoes for marathon training under $150," the model synthesises structured data, reviews, and specifications to generate recommendations.

This is where Generative Engine Optimisation (GEO) becomes critical. If your catalogue isn't machine-readable, you're invisible at the moment of recommendation. You're no longer competing for page rank. You're competing for interpretability.

Phase 2: The Assistant (Emerging)

The next phase is already unfolding. Agents like OpenAI's Operator can execute tasks. A user can ask for a product, and the agent handles discovery, filtering, comparison, and even cart creation.

The human still approves the purchase, but the agent has done the work. The battleground shifts from visibility to selection. The question is no longer "Will they find us?" but "Will the agent choose us?"

Phase 3: The Autonomous Buyer (Next)

In the near future, agents will manage purchases end-to-end. They will restock essentials, replace products, and act on predictive signals without human prompts.

This requires new infrastructure: secure payment rails, authenticated agent protocols, and trust frameworks. Early standards like Shopify's Universal Commerce Protocol signal how agents will query inventory and transact without ever visiting a website.

The paradigm shift in brand visibility

For years, brands optimised for search engines using keywords and backlinks. That paradigm is shifting.

In an AI-mediated world, the gatekeeper is a reasoning model. These models parse structured product data to answer questions, compare options, and validate claims.

If your catalogue lacks structure, such as unclear descriptions, inconsistent attributes, or outdated data, AI cannot validate your product. And if it cannot validate it, it will not recommend it.

Initiatives like Shopify's Catalogue API and emerging protocols are enabling AI systems to query inventory, access verified specifications, and surface products directly within conversational interfaces.

You are no longer building a website for human visitors alone. You are now building for systems that query, verify, and act on your catalogue.

This requires:

• Structured product attributes that machines can compare
• Real-time inventory and pricing that can be verified
• Clear relationships between products for contextual understanding
• Explicit differentiation that helps models determine relevance

Consumers are asking AI to solve problems, not return lists of links. The brands that answer those problems at the data layer will win.

The implications of this for leadership are immediate. Selection is now governed by infrastructure. Catalogue optimisation becomes a core growth lever, not a backend task.

The window is narrower than it appears. AI models are being trained on today's data. 

Missing this moment means missing the foundation of how your brand is understood.

Trust becomes the moat. Brands with consistent, verifiable data will be the ones agents default to. When AI recommends your product inside a private conversation, traditional tracking models lose relevance.

The brands that win will recognise a fundamental shift: the buyer is no longer just human, and the catalogue is no longer just a website. The transition from AI as a researcher to an assistant to an autonomous buyer is already underway. 

The question isn't whether this shift will happen. It's whether your brand will be legible when it does. If your catalogue isn't legible to AI, it isn't in the conversation.

Glu helps Shopify brands become structured, interpretable, and recommendable inside AI systems. Start with a free audit.