Excerpt from TechCrunch:
Google on Sunday announced a new open standard called the Universal Commerce Protocol (UCP) for AI agent-based shopping, at the National Retail Federation (NRF) conference. The standard, developed with companies like Shopify, Etsy, Wayfair, Target, and Walmart, lets agents work across different parts of customer buying processes, including discovery and post-purchase support.
The core idea is that the standard could facilitate these various parts of the process instead of requiring connections with different agents. Google said that it also works with other agentic protocols, such as Agent Payments Protocol (AP2) — which Google announced last year — Agent2Agent (A2A), and Model Context Protocol (MCP). The company specified that Agents and Businesses can pick and choose specific extensions of the protocol that suits their needs.
The company said that it will soon use UCP for eligible Google product listings in AI mode in search and the Gemini apps to let shoppers check out directly from U.S.-based retailers while researching a product. Users will be able to pay using Google Pay and pass on the shipping information saved in the Google Wallet. Google said that it will soon support PayPal as a payment option.
…and this is the Agentic Commerce Protocol from OpenAI (launched last September):
Merchants pay a small fee on completed purchases, but the service is free for users, doesn’t affect their prices, and doesn’t influence ChatGPT’s product results. Instant Checkout items are not preferred in product results. When ranking multiple merchants that sell the same product, ChatGPT considers factors like availability, price, quality, whether a merchant is the primary seller, and whether Instant Checkout is enabled, to optimize the user experience.
1) What each protocol is really optimizing for
OpenAI + Stripe: ACP = “commerce inside ChatGPT”
ACP is the plumbing behind Instant Checkout in ChatGPT. Merchants publish a product feed and expose checkout endpoints; ChatGPT renders the checkout UI; the merchant stays Merchant of Record and charges via their PSP; OpenAI/Stripe provide a secure delegated-payment mechanism (Stripe’s Shared Payment Token is the first implementation mentioned).
Strategically, ACP is built to make the conversation the store, as follows:
discovery → selection → checkout happens where the user already is (ChatGPT).
Google: UCP = “commerce as a universal layer across surfaces”
UCP is positioned as an open, modular standard that can model much more of the funnel/lifecycle. Not just checkout, but also identity linking and order management at launch, with roadmap items like multi-item carts, loyalty account linking, tracking/returns. It’s meant to be consumed by many “surfaces” (agents, apps, chat, voice, etc.) and plugs into existing protocols like AP2 / A2A / MCP.
Strategically, it’s very “Google” this move. It makes the ecosystem interoperable so the winner is the player with the best distribution + discovery + monetization machine (Search/Ads/Android/Gemini).
2) Practical differences that matter (not just philosophy)
A. Scope
ACP: “agent-ready checkout + payment delegation” (plus product feed for discovery in ChatGPT experiences). It explicitly frames 3 core flows: product feed, checkout/order handling, delegated payments.
UCP: designed for “end-to-end shopping journey” and broader lifecycle primitives; initial core capabilities: Checkout, Identity Linking, Order Management (with post-purchase concepts clearly in scope / roadmap).
Concluding, ACP can feel like “a fast lane into ChatGPT commerce”. UCP is “a commerce operating system layer” that (if adopted) reduces everyone’s need for one-off integrations.
B. Where the UX lives
ACP: checkout experience is embedded in OpenAI UI (ChatGPT renders it), while the merchant keeps state + payment processing on their systems.
UCP: supports native checkout integrations and an embedded checkout option (iframe-based) for bespoke flows/branding (at least for approved merchants).
Concluding, UCP is trying to preserve merchant brand/control and work across many surfaces; ACP is optimized for a consistent ChatGPT-native buying flow.
C. Payments + consent model
ACP: Delegated Payment Spec → OpenAI prepares a one-time delegated request; a PSP returns a token; merchant completes charge. Stripe’s Shared Payment Token is the first compatible implementation mentioned.
UCP: emphasizes compatibility with Agent Payments Protocol (AP2) and talks about “provable” authorizations backed by cryptographic proof of user consent, plus modular “payment handler” design.
Concluding, ACP is payment-rail pragmatic (Stripe is the wedge). UCP is pushing an ecosystem-wide consent/audit posture (very attractive to platforms and regulators if it holds up in practice).
D. Go-to-market + monetization pressure
OpenAI: Reuters reports merchants pay OpenAI a fee per completed purchase via Instant Checkout (new revenue stream beyond subscriptions).
Google: UCP is paired with commerce + ads tooling in “AI Mode”, including merchant attributes and “Direct Offers”, and Google is openly shifting monetization into conversational shopping surfaces.
Implications:
OpenAI monetizes transactions (take-rate logic).
Google monetizes intent + discovery (ads/promotions layered into shopping journeys), and UCP makes that intent legible and action-ready.
3) Stakeholder implications
For merchants
ACP path: faster “distribution into ChatGPT”, but you’re playing on OpenAI’s surface and (today) ChatGPT availability is partner-gated for Instant Checkout.
UCP path: more implementation surface area, but you get a surface-agnostic contract. Your capabilities can be consumed by many agents/platforms, and Google explicitly sells it as keeping you the Merchant of Record with customer/data control.
Concluding, ACP is “plug into this AI mall”. UCP is “become compatible with every AI mall”.
For AI platforms / assistants
ACP strengthens the ChatGPT “walled garden”. Even as an open standard, the flagship implementation is ChatGPT Instant Checkout, so the gravitational pull is ChatGPT distribution.
UCP tries to commoditize integrations. If everyone speaks UCP, the platform differentiators shift to ranking, UI, trust, and distribution (Google’s home turf).
For payments players
ACP is a Stripe-friendly wedge (delegated tokenization spec + first implementation).
UCP is “payment-handlers + AP2 compatibility” and explicitly rallies networks/PSPs around interoperability.
ACP could accelerate “Stripe as the default agentic payments bridge”. UCP could turn payments into a modular plug-in where Google Wallet/PayPal/etc. compete at the surface level.
For competition
Your framing matches what Google is signaling: UCP is universal on purpose. If no one has a structural integration moat, the winner becomes the best aggregator of demand (Search + Gemini + Ads).
Meanwhile, OpenAI is betting that the user’s primary interface is the conversation. Keep the entire journey inside ChatGPT and take a cut.
My bullish take is that Google has already shown how things will unfold. Seeing how it is eating daily from the Chat GPT market share by leveraging its over 3 billion Android devices, I am firmly convinced that Google will remain relevant also in the next search era.
“Where is my moat?” is designed for individual readers, though the occasional forward is absolutely fine. If you’d like to set up multiple subscriptions for your team with a group discount (minimum 5 seats), reach out to me directly.
Thanks for your support & have a wonderful day!
Thank you for being a supporter of WIMM!
