{{item.videoDuration}}
{{item.title}}
{{item.text}}
{{item.videoDuration}}
{{item.text}}
Designing for new rules of discoverability and transaction
We’ve explored how AI is reshaping commerce by creating a decision layer between customer intent and transaction. The next question is practical: If AI agents increasingly influence what customers see, compare, and buy, how should retailers design their commerce systems to increase exposure?
From PwC’s perspective, two capabilities now sit at the center of agentic commerce:
- discoverability designed for generative AI
- transaction infrastructure designed for interoperability
Rewriting discoverability
In traditional ecommerce, brands competed for search engine ranking and optimized the visual appeal of their shelf space. Customers manually browsed links, scanned pages, and narrowed options.
But AI-mediated experiences compress the process of discovery.
AI systems synthesize answers. They reduce a wide range of possible products to a handful of recommended options. Brands that aren’t surfaced in the synthesized response are likely to be entirely excluded from consideration.
This shift introduces an urgent need for retailers to optimize for generative engines.
GEO (generative engine optimization) isn’t simply SEO under a new name. It requires:
- Structured, authoritative product data
- Clear and extractable explanations of claims and benefits
- Consistent taxonomy and metadata
- Trust signals that AI systems can interpret and reuse
AI models are increasingly capable of reasoning, comparison, and workflow orchestration, but they rely on structured inputs and predictable interfaces. As such, brands optimized only for human browsing risk becoming invisible in AI-generated summaries.
Discoverability now depends on being included, not just ranking first. Commerce should become agent-addressable.
Yet, even when AI systems surface a product, today’s commerce stacks often break at the moment of action. That’s because of the fragmented nature of so many commerce environments. Retailers often deploy custom stitching and human intervention to compensate for disconnected systems across product, cart, identity, checkout, payments, and fulfillment. Agents, however, don’t work around brittle checkout flows. Instead, they’ll simply abandon the cart.
To support scalable, agent-led transactions, commerce systems should take predictable, structured actions. Without common protocol, agents can’t reliably reason, transact, or recover. To enable agentic commerce at scale, the ecosystem requires a universal commerce protocol (UCP).
UCP is an open-source standard that creates a common language for commerce between AI agents and businesses. It standardizes product discovery, cart actions, identity linking, checkout, and order execution while preserving merchant-of-record ownership.
Think of UCP as a layer of the transaction, allowing brands to:
- Turn on AI-powered shopping on their own websites
- Enable transactions inside AI experiences without re-platforming
- Expose product, pricing, and fulfillment logic in an AI-readable format
- Maintain governance, compliance, and policy controls
Instead of building one-off integrations for each surface, enterprises integrate once and connect across agents.
Conversational commerce on owned properties
Transaction rails alone are not sufficient. The interaction layer should also evolve.
Google Cloud’s Gemini Enterprise for Customer Experience (GECX) is designed to bring conversational buying directly to brand websites. GECX uses AI to guide customers through product discovery, answer questions in real time, and move them from intent to purchase. And it all happens seamlessly within the brand’s own digital environment. GECX:
- Interprets customer intent
- Guides product selection
- Connects to existing catalog and checkout systems
- Executes transactions within the conversational flow
This does not require re-platforming. It sits on top of a retailer’s existing commerce backbone while preserving merchant-of-record control.
Agentic commerce, therefore, spans two environments simultaneously:
- Owned brand experiences, where conversational guidance transforms browsing into structured, AI-assisted buying
- AI-native environments, where structured transaction protocols enable secure, agent-initiated purchases
Both are critical. Designing for one without the other limits scalability.
The strategic imperative
In an agent-driven world, being invisible to AI increasingly means being invisible to customers. Fragmented commerce stacks are a competitive liability. And interoperability is a growth enabler.
The organizations that move early can shape how agents interpret, recommend, and transact across their categories. Those that wait may find that demand is routed elsewhere.
Learn more about how PwC and Google Cloud can help your organization make that early move to shape your agents for success.
Want to learn more? Join PwC at Google Cloud Next
April 22-24, 2026 | Las Vegas
{{filterContent.facetedTitle}}
{{contentList.loadingText}}
{{contentList.loadingText}}
Follow us
