Google Just Told Retailers: Get Ready for AI Shoppers

In January 2026, Google stood on stage at the National Retail Federation's annual conference and announced the Universal Commerce Protocol — an open standard that lets AI agents browse, negotiate, and buy from any retailer that implements it. Months earlier, OpenAI and Stripe had already shipped the Agentic Commerce Protocol, powering checkout inside ChatGPT.

Two protocols. Two visions. One unmistakable message: the age of AI-powered shopping is here, and businesses that aren't ready will be invisible.

If you run an online business, this isn't a trend to watch — it's infrastructure you need to understand right now. (New to agent commerce? Start there for the full picture.) This guide focuses on the two protocols reshaping how agents transact.

What is the Agentic Commerce Protocol (ACP)?

ACP is an open standard co-developed by OpenAI and Stripe. It defines how an AI assistant can act as the shopping interface: discovering products, presenting options to a buyer, and completing a purchase — all without the customer ever leaving the chat.

How ACP works

1. A shopper asks their AI assistant (currently ChatGPT) for a product recommendation.
2. The assistant queries ACP-enabled merchants via a REST API to retrieve product catalogues, pricing, and availability.
3. The shopper confirms a purchase inside the conversation.
4. Stripe processes the payment through its existing rails.

ACP is deliberately focused. It optimises for a single-agent experience — one user, one AI assistant, one checkout flow. Merchants who already use Stripe can integrate with minimal code changes.

Who's already on ACP?

Shopify has announced ACP support across its platform. Salesforce has integrated it into Commerce Cloud. The merchant base is growing fast — Stripe's existing install base means any Stripe merchant can onboard with minimal effort, which is driving adoption far faster than typical protocol launches.

What is the Universal Commerce Protocol (UCP)?

UCP is Google's answer — a protocol designed not just for checkout, but for the entire shopping journey: discovery, evaluation, negotiation, purchase, and post-purchase support.

How UCP works

1. Merchants publish their commerce capabilities at a /.well-known/ucp endpoint on their own domain — similar to how robots.txt signals to search crawlers.
2. Any AI agent (not just Google's) can discover these capabilities and negotiate a transaction.
3. Payments flow through the Agent Payments Protocol, which supports Google Pay, PayPal, and traditional payment methods.
4. Post-purchase actions like returns, tracking, and support are all part of the protocol.

UCP is protocol-agnostic. It works with REST APIs, MCP (Model Context Protocol), and Google's Agent-to-Agent (A2A) framework. It's designed for a multi-agent future where shoppers might use different AI assistants for different tasks.

Who's backing UCP?

Google co-developed UCP with Shopify. The backing coalition includes Adyen, American Express, Best Buy, Etsy, Flipkart, Macy's, Mastercard, Stripe, Target, The Home Depot, Visa, Walmart, Wayfair, and Zalando, among others.

UCP vs ACP: A side-by-side comparison

Feature	ACP (OpenAI + Stripe)	UCP (Google)
Launched	Late 2025	January 2026
Scope	Checkout-focused	Full shopping journey
Discovery	Centralised (merchants register via Stripe)	Decentralised (/.well-known/ucp endpoint)
Architecture	REST API, single-agent optimised	Protocol-agnostic (REST, MCP, A2A)
Payments	Stripe	Agent Payments Protocol (Google Pay, PayPal, others)
Agent model	Single-agent (ChatGPT)	Multi-agent (any AI assistant)
Integration effort	Low (Stripe merchants can integrate quickly)	Higher (publish capabilities, handle negotiation)
Open standard	Yes (OpenAI + Stripe governance)	Yes (broad industry coalition)

What this means for your business

1. You need both protocols

ACP gets you into ChatGPT faster — potentially with minimal code. UCP gives you broader reach across every AI assistant that supports it. They're not competing standards; they're complementary layers. Most major retailers are supporting both.

2. Structured data is the foundation

Both protocols rely on machine-readable product information: prices, descriptions, availability, shipping terms. If your catalogue isn't structured, neither protocol can work with it. This isn't new advice — it's been the SEO best practice for years — but now it directly affects whether AI agents can sell your products.

3. Discovery just changed

With UCP's /.well-known/ucp endpoint, AI agents can discover your commerce capabilities the same way search engines discover your sitemap. Businesses that publish these endpoints gain a new distribution channel. Those that don't remain invisible to agent-based shoppers.

4. Checkout needs to work without a browser

Traditional e-commerce assumes a human is clicking through pages. Agentic commerce assumes an API call completes the transaction. If your checkout flow requires CAPTCHA solving, multi-step form filling, or JavaScript rendering to display prices, AI agents will skip you entirely.

5. Post-purchase matters more than ever

UCP includes post-purchase interactions as part of the standard. Returns, order tracking, customer support — all accessible to agents. Businesses that expose these capabilities will retain customers; those that don't will lose them to competitors whose agents can handle the full lifecycle.

Where to start

If this feels overwhelming, it isn't. The steps are concrete:

1. Audit your structured data. Do you have Schema.org markup for your products? Are prices, availability, and descriptions machine-readable? This is the single most impactful thing you can do today.

2. Check your API surface. Do you have APIs that expose product catalogues and pricing? Can an external system place an order without a browser?

3. Evaluate your payment stack. If you're on Stripe, ACP integration is straightforward. For UCP, check whether your payment provider supports the Agent Payments Protocol.

4. Publish a UCP endpoint. Even a minimal /.well-known/ucp file that describes your capabilities signals to AI agents that you're open for business.

5. Scan your site. The Zeodyn Scanner™ checks your website across six dimensions of AI commerce readiness — Discovery & Access, Structured Data, Commerce Data, Protocol Support, Security & Trust, and Technical Performance — and gives you a Zeodyn Score™ out of 100 with specific recommendations.

The bottom line

Google and OpenAI don't agree on much, but they agree on this: AI agents are going to buy things. The only question is whether they'll buy from you.

The businesses that move now — structuring their data, exposing their APIs, implementing these protocols — will capture the agentic commerce wave. The rest will wonder why their traffic dried up.

Not sure where to start? Start with your score: zeodyn.com/scan