What Are AI Commerce Protocols?
AI commerce protocols are the open standards behind agentic commerce. They define how AI agents discover products, read merchant data, complete checkout and handle what happens after the purchase.
Three cover the commerce side:
- ACP (Agentic Commerce Protocol): OpenAI and Stripe, September 2025. Product feed and checkout for ChatGPT and ACP-compatible surfaces such as Copilot.
- UCP (Universal Commerce Protocol): Google and Shopify, January 2026. The full journey from discovery to post-purchase for Google AI Mode.
- MCP (Model Context Protocol): Anthropic, now Linux Foundation, November 2024. General-purpose connection between AI models and external systems, including live catalog queries.
A second group covers payment trust: AP2 from Google, Visa TAP and Mastercard Agent Pay. These verify that an agent is who it says it is and that the shopper authorised the purchase. They arrive through your payment processor rather than as something you build.
How the Protocols Work Together
MCP connects data, ACP handles the ChatGPT feed and checkout, UCP covers the Google journey. They layer rather than compete.
The protocols sit at different layers:
Data (MCP). Lets an agent query live product data, inventory, pricing and orders from merchant systems. Both ACP and UCP can use it.
Transaction (ACP). Defines the product feed ChatGPT reads and the checkout flow, with payment through Stripe. Since March 2026, ChatGPT sends shoppers to the merchant to buy, and ACP continues behind the feed and retailer-built ChatGPT Apps.
Journey (UCP). Covers discovery, browsing, cart, checkout (with AP2), tracking, returns and service for Google AI Mode. Designed to work with MCP for data.
Scroll horizontally to read the full table →
| Protocol | Developer | Launched | Primary Function | Key AI Platforms |
|---|---|---|---|---|
| ACP | OpenAI / Stripe | Sep 2025 | Product feed and checkout | ChatGPT, Copilot |
| UCP | Google / Shopify | Jan 2026 | Full commerce lifecycle | Google AI Mode |
| MCP | Anthropic / Linux Foundation | Nov 2024 | Data connectivity | Claude, ChatGPT, Gemini, others |
What Retailers Actually Have to Do
Less than the protocol count suggests. Your commerce platform and processor carry most of it. Your job is the product information they transmit.
The protocol list looks like an engineering backlog. For most retailers it is not:
- Your commerce platform handles the protocols. Shopify enrols merchants for ChatGPT and built UCP with Google. Other platforms are following. Check what yours already supports before planning any integration.
- Your processor handles payment trust. AP2, Visa TAP and Mastercard Agent Pay arrive through Stripe, Adyen, Fiserv and the card networks. Confirm support with your processor.
- Your feeds carry your products. Google Merchant Center and the OpenAI product feed are how AI Mode and ChatGPT see your catalog. Keep them complete and current.
- Your product information decides the outcome. Every protocol transmits the same attributes, descriptions and Q&A. If those are thin, the protocol delivers a thin product to the assistant.
The fourth point is where the work is, and it is the work Paz handles. A Paz expert investigates missed recommendations and identifies product information gaps, fixes them from your sources, our experts check the changes against agreed guidelines, and the fix reaches supported store and feed destinations. See how it works.
FAQ
Do I need to support all three protocols?+
Are the protocols competing?+
Will there be more protocols?+
Related terms
How AI-ready are your products?
Check one product page for the information gaps that keep it out of AI answers and search results. If you want the fixes handled, Paz experts do that work for ecommerce teams.