Google Universal Commerce Protocol: Multi-Item Carts

What Is Google Universal Commerce Protocol

Google Universal Commerce Protocol (UCP) is an open API specification that defines the communication layer between AI shopping agents and merchant commerce systems. Where traditional e-commerce integrations are built point-to-point — a merchant integrates with each platform separately — UCP establishes a single standard that any compliant AI agent can use to discover products, verify availability, build orders, and initiate checkout.

The protocol emerged from a recognition within Google that the transition to agentic shopping would break the existing web-scraping and product feed model. AI agents need structured, queryable data in real time. Static feeds updated once per day cannot support a shopping agent deciding whether to buy a product right now at the current price with current stock levels. UCP was designed to bridge that gap.

The original UCP focused on single-product discovery and checkout. Merchants could expose a product catalog endpoint and a checkout endpoint, and AI agents could find products and purchase one at a time. That foundation proved the model but immediately revealed the limitation: real shopping trips almost always involve multiple items. The March 2026 expansion addresses that gap directly.

Multi-Item Cart Support Explained

The core addition in the March 2026 UCP expansion is the cart session object: a persistent, stateful representation of a consumer's intended purchase that an AI agent builds incrementally before handing off to checkout. Where the original protocol treated each product interaction as a discrete event, the new cart session allows an agent to add items, modify quantities, apply discount codes, and review the full order before initiating payment.

This changes the practical capability of AI shopping agents dramatically. A Gemini-powered shopping assistant can now accept a consumer request like "build me a home office setup under $2,000" and construct a multi-item cart across monitor, keyboard, mouse, desk accessories, and cable management products — checking inventory and pricing for each item in real time, applying relevant promotions, and presenting the complete order for consumer approval before checkout.

The cart session object includes explicit conflict resolution mechanics for situations where items become unavailable or prices change between initial query and checkout. The protocol defines three resolution strategies: block (prevent checkout until the consumer acknowledges the change), substitute (offer an alternative product), and remove (drop the unavailable item with notification). Merchants can configure which strategy applies to different product categories, giving them control over the checkout experience even when an AI agent is driving it.

Real-Time Catalog Queries and Product Feeds

Alongside multi-item cart support, the March 2026 UCP update introduced the catalog query layer: a set of API endpoints that allow AI agents to query merchant product data in real time rather than relying on batch-updated product feeds. This is a fundamental architectural shift in how AI agents access merchant inventory.

Traditional product feeds — whether Google Merchant Center uploads, XML sitemaps, or JSON-LD markup — are batch documents. A merchant uploads a feed file, and Google indexes it. The index reflects product data as it existed when the feed was generated, which may be hours or days old. For static products with stable pricing, this works adequately. For merchants with dynamic inventory, frequent price changes, or flash sales, stale feed data is a persistent problem that leads to consumer disappointment and failed conversions.

The catalog query endpoints support three main query types: product lookup by GTIN or merchant SKU, category search with filtering parameters, and availability check for a specific product-variant-quantity combination. Merchants expose these endpoints and UCP handles the authentication, rate limiting, and response caching on the protocol layer. The performance requirement is sub-500ms response time at the 95th percentile, which means merchants need to optimize their catalog API for this latency target.

20+ Global Partners and the Open Standard

One of the most significant aspects of UCP that distinguishes it from Google's previous commerce initiatives is the breadth of the partner ecosystem that co-developed the specification. Previous Google commerce tools were largely Google-designed with external adoption expected after the fact. UCP was developed with active input from Visa, Mastercard, Stripe, and Shopify from the beginning, plus more than fifteen additional global partners across payment processing, logistics, and commerce platforms.

The multi-stakeholder development model means UCP is genuinely portable across the ecosystem rather than a Trojan horse for Google Lock-in. Merchants implementing UCP gain access to AI agents powered by Gemini, but also to any third-party AI shopping assistant that implements the public specification. This is the same network effect logic that made HTTP successful: a universal standard benefits every participant more than any proprietary alternative. For context on how this fits within the broader evolution of agentic commerce protocols, UCP and ACP are complementary rather than competing, with UCP stronger on Google-native shopping surfaces and ACP covering broader multi-agent orchestration scenarios.

Merchant Implementation Guide

UCP implementation has three layers: product data quality, API endpoint creation, and authentication setup. Merchants who already run well-maintained Google Merchant Center feeds are starting from a better position than those without, but all merchants need to add the API layer on top of their existing product data infrastructure.

Platform-hosted merchants on Shopify have the most straightforward path: Shopify is building native UCP support into its platform as an official partner. Custom platform merchants face more work but have the advantage of full control over their implementation. Mid-market merchants on platforms like WooCommerce or BigCommerce should watch for official plugins from those platforms before committing to a custom build. For a detailed look at how Shopify's agentic commerce and UCP implementation is coming together, that post covers the Shopify-specific technical path in depth.

How AI Agents Use UCP Shopping Flows

Understanding how AI agents actually traverse UCP endpoints helps merchants design their implementations for real-world traffic patterns. Google's Gemini-powered shopping agent follows a structured decision loop that UCP is designed to support efficiently.

Payment Security and Compliance

Payment security in an agentic commerce context introduces requirements beyond standard e-commerce PCI-DSS compliance. When an AI agent is initiating a purchase on behalf of a consumer, the authentication chain must prove not only that the consumer authorized the purchase but that the agent is operating within the consumer's explicitly granted permissions. UCP addresses this through a layered authorization model.

Strategic Implications for Merchants

UCP is infrastructure-layer work with significant competitive implications. Merchants who implement UCP early will have their products discoverable by AI shopping agents as those agents gain consumer adoption. Merchants who wait will face a harder catch-up problem as agent traffic concentrates among compliant merchants.

The competitive dynamics are particularly acute for high-consideration purchases where consumers benefit most from AI assistance — home electronics, apparel, furniture, sporting goods. These are categories where an AI agent's ability to query across merchants, compare specifications, and build complete carts provides genuine consumer value. Early UCP adopters in these categories will see the highest incremental traffic from agentic shopping surfaces. For a full assessment of how our team approaches eCommerce strategy including emerging protocol adoption, we work with clients on both the technical implementation and the go-to-market positioning of new commerce capabilities.

Current Limitations and Roadmap

UCP is a significant advance in agentic commerce infrastructure, but it carries real constraints that merchants should understand before committing resources to implementation. The protocol is still evolving, the consumer-facing AI surfaces are in limited rollout, and the engineering investment is non-trivial.