Three months after launching its Universal Commerce Protocol — the open standard that lets AI agents shop on behalf of consumers — Google has shipped a round of updates that move UCP from proof-of-concept to something that looks a lot like production-ready infrastructure.
The updates, detailed on Google's blog, address three of the biggest limitations that held back the January launch: single-item-only carts, stale product data, and broken loyalty experiences.
What's New
Multi-item carts are the headliner. UCP now lets shopping agents save multiple items to a cart at once — the way actual humans shop. The January launch required agents to process one item at a time, a limitation that made the technology feel like a demo rather than a tool.
Real-time catalog access is the second major addition. A new Catalog capability lets agents retrieve live product details — variants, inventory, pricing — directly from a retailer's systems. No more making purchase decisions based on stale data that was last refreshed hours ago.
Identity Linking rounds out the update, connecting shoppers' loyalty accounts across UCP-integrated platforms. That means a consumer who triggers an AI-assisted purchase through Google can still earn loyalty points, access member pricing, and receive free shipping — benefits that would otherwise be lost when shopping through an intermediary.
The Checkout Endgame
Google isn't being subtle about where this is headed. UCP will soon power a checkout feature on eligible product listings in AI Mode in Search and the Gemini app, according to 9to5Google. The implication: a consumer could ask Gemini to find a specific product, compare prices across retailers, check inventory, and complete the purchase — all without ever visiting a retailer's website.
More than 20 companies across retail and payments have endorsed UCP, including Shopify, Etsy, Wayfair, and Target, per TechCrunch. Google is also simplifying the onboarding process through Merchant Center, according to Search Engine Land, aiming to make UCP accessible to retailers of all sizes — not just enterprise-scale operators with dedicated integration teams.
The eBay Counterpoint
Not everyone is on board. As we've covered, eBay explicitly banned autonomous AI "buy for me" agents from its platform in February, prohibiting LLM-driven bots from placing orders without human review. eBay's concern is straightforward: if bots drive prices down efficiently, the commission-based marketplace loses revenue. There are also legitimate worries about AI agents hoarding products, manipulating bidding, and distorting pricing dynamics.
The contrast is revealing. Google is building the plumbing for a world where AI agents handle routine commerce on behalf of consumers. eBay is drawing a hard line between AI as a discovery tool and AI as a transaction engine. Both positions are rational — they just reflect fundamentally different business models and risk profiles.
What Retailers Should Watch
The practical question for retailers isn't whether agentic commerce is coming — the Adobe data showing AI traffic converting 42% better than human shoppers has settled that debate. The question is which protocol stack to bet on.
Google's UCP has first-mover advantage and distribution. But it also means accepting Google as an intermediary in the transaction. For brands that have spent years building direct-to-consumer channels, that's a meaningful strategic concession. The coming months will reveal whether UCP's convenience outweighs the control retailers give up — or whether alternatives emerge that offer the same AI-agent capabilities without the platform dependency.