The New Product Feed: How ACP and UCP Are Changing Commerce Data Strategy

In September 2025, OpenAI and Stripe launched the Agentic Commerce Protocol (ACP), an open standard that determines how AI agents, merchants, and payment systems work together to create agentic shopping experiences. Before ACP, every AI assistant needed a custom integration with each merchant and payment provider. The protocol replaced those bespoke integrations with a shared language for validating credentials, confirming orders, and processing transactions. It powers ChatGPT's shopping experiences and is already supported by Salesforce, BigCommerce, Wix, Squarespace, WooCommerce, and PayPal.

In January 2026, Google and Shopify launched the Universal Commerce Protocol (UCP), built for Google AI Mode, Gemini, and Google Shopping. UCP covers the full purchase journey, from search through payment and order tracking. It’s protocol-agnostic, supporting REST, MCP, and A2A, and is endorsed by more than 20 organizations including Mastercard, Visa, Walmart, Target, and Best Buy.

These aren't competing standards. They're parallel infrastructure for overlapping ecosystems. A brand selling through ChatGPT needs ACP; a brand appearing in Google AI Mode needs UCP. Most mid-to-enterprise retailers will need both, which makes a unified, composable data foundation essential.

A third standard, the Machine Payments Protocol (MPP), emerged at Stripe Sessions 2026 in April to enable agent-to-agent payments for programmatic use cases like API access and B2B transactions. It's early stage, but it confirms the direction of travel is towards more open protocols, not fewer.

ACP and UCP aren't just checkout protocols, they're data protocols, and what they require is meaningfully different from what traditional ad channels require.

Stripe has been explicit that product feeds are being built for businesses to broadcast real-time pricing and availability, not a nightly batch export. Google has framed UCP's role as extending into product data standardization first, ahead of checkout integration or payment token configuration.

Consider the scenario Stripe uses to illustrate multi-item cart support: a user tells ChatGPT to "order everything I need for taco night," and the agent has to build a cart across multiple SKUs and execute it as a single transaction. That requires accurate real-time inventory and pricing across every SKU involved, confirmed at the moment of query. A feed sufficient for a Google Shopping ad isn't sufficient for that. __This is less like an ad feed and more like a live product API, expected to respond accurately and instantly to whatever the agent asks. __

In agent-mediated commerce, the behavioral data stream starts at the add-to-cart moment. Discovery, consideration, and comparison all happen inside ChatGPT or Gemini now, which means attribution collapses, personalization breaks, and traditional retail media signals go dark.

That means feed quality doesn't only affect which recommendations you win, it also affects the only data you'll get back from the channel. Brands with richer, structured feeds generate cleaner agentic transaction data; brands with thin feeds get a purchase event stripped of everything that made it interpretable.

Stripe's retrospective on early agentic commerce found that most real-world edge cases trace back to data quality: prices that don't match, inventory that isn't current, attributes that don't answer the query that surfaced the product in the first place. Brands that solve the data problem first win the recommendations, and accumulate better data from those recommendations, which compounds the advantage over time.

As an ACP launch partner, Orium has seen this directly: the merchants who move fastest through protocol integration aren't the ones with the most development resources. They're the ones whose data is already clean, structured, and API-accessible.

The Stripe Agentic Commerce Suite reduces the integration lift— handling discovery, checkout, payments, and fraud detection across agent surfaces through a single integration—but it doesn't enrich your catalog for you. A simpler integration with a thin catalog is still a thin catalog.

Catalog API access. Is your product catalog accessible in real time via API, or is it batch-only?

Feed structure. Are your feeds structured to answer agent queries, including conversational attributes and intent-ready descriptions, or just for ad relevance scoring?

Multi-item transaction readiness. Do you have inventory and pricing accurate enough for multi-item cart validation?

Protocol mapping. Have you mapped your current feed against ACP and UCP data requirements? The gaps are usually larger than expected.

Integration strategy. Are you evaluating the Stripe Agentic Commerce Suite as your implementation path, or building endpoints from scratch?

Ownership. Who owns protocol readiness? If the answer is "nobody yet," that's step one.

Protocol implementation and conformance testing are measured in months, not weeks, so starting in 2026 is not early. Talk to our team to understand how your current data architecture aligns with ACP and UCP requirements.