If you have sat through a vendor pitch about AI shopping this year, you have heard the acronyms. UCP, ACP, MCP, AP2. Each one comes with a confident claim that it is the future of commerce. For a Shopify Plus brand trying to decide what to do on Monday, most of that noise can be filtered down to three standards worth understanding and a short list of actions worth taking.
This is the plain-language version.
Why AI agents need commerce protocols at all
An AI agent cannot learn the quirks of millions of individual stores. To find a product, check the price, build a cart and complete a purchase reliably, it needs a standard way to ask every store the same questions. That is all a protocol is, a shared language between agents and merchants.
Without one, agents scrape pages and guess. With one, they can ask your store directly and get a clear answer.
What is MCP?
The Model Context Protocol is an open standard for connecting AI applications to external systems, such as databases, business tools and stores. Anthropic introduced it in late 2024, and in December 2025 donated it to the Agentic AI Foundation, a new Linux Foundation project co-founded with Block and OpenAI. It is supported across ChatGPT, Gemini, Microsoft Copilot and many other tools.
Think of MCP as the plumbing. It is not specific to commerce. It is how an agent connects to anything, including your store. Shopify's Storefront Catalog MCP, for example, lets an agent search a merchant's products, look up variants and fetch product details.
What is UCP?
The Universal Commerce Protocol is an open standard for the whole shopping journey, from product discovery through cart, checkout and post-purchase. It was announced in January 2026, developed by Google with Shopify, Etsy, Wayfair, Target and Walmart, and endorsed by more than twenty other companies, including Adyen, Mastercard, Stripe and Visa. UCP can run over standard APIs, over MCP or over Google's Agent2Agent protocol.
For Shopify merchants, UCP arrived alongside Agentic Storefronts, which let you manage how your products appear across AI channels such as ChatGPT, Microsoft Copilot and Google's AI Mode from the Shopify admin. In practice, Shopify does most of the protocol work for stores on its platform. Your job is the data and the decisions, not the plumbing.
What is ACP?
The Agentic Commerce Protocol was created by OpenAI and Stripe and launched in September 2025 alongside Instant Checkout, which let shoppers buy inside ChatGPT. In March 2026 OpenAI retired Instant Checkout after about six months, refocused ChatGPT shopping on product discovery and sent checkout back to merchants' own sites. ACP continues as an open standard, mainly for sharing product information with AI platforms.
The lesson is useful. The first version of buying inside an AI assistant did not stick, but discovery through AI assistants did. Any plan built around one assistant's checkout feature was a plan built on sand.
What about AP2 and agent payments?
Payments have their own layer. Google's Agent Payments Protocol, AP2, was developed with more than sixty organizations, including Mastercard, PayPal and Adyen. It uses signed records called mandates to prove what a shopper authorized an agent to buy. Visa and Mastercard have their own programs for recognizing trusted agents. We cover what this means for fraud, chargebacks and returns in when an AI agent buys from your store.
What a Shopify Plus brand should do now
In order of value, not order of excitement.
- Get your product data right. Every protocol reads from the same catalog. Complete attributes, clean variants and current prices matter more than any integration. Our product data checklist for AI shopping agents is the place to start.
- Review your AI channels in the Shopify admin. Understand which AI channels your products are available on, what information is shared and how it appears.
- Make sure checkout is ready for the traffic. With checkout back on merchant sites, agents increasingly send ready-to-buy shoppers to your store. Accurate stock, clear shipping costs and a fast checkout convert that traffic.
- Measure AI referrals. Track sessions and orders that arrive from AI assistants so you can see whether this channel matters for your brand yet.
- Check your headless storefront, if you have one. If you run a custom front end on NextJS, make sure product data, structured data and policies stay consistent between your storefront and what Shopify shares with AI channels.
What can wait
- Building your own shopping agent. Few mid-market brands need one yet, and the protocols are still settling.
- Custom protocol integrations. If your platform supports UCP for you, building your own version rarely pays off.
- Chasing every new acronym. New standards will keep arriving. The data foundation underneath them will not change much.
If you are weighing whether a headless build makes sense in this new landscape, our Vercel and Shopify webinar on headless commerce covers what changes and what stays with Shopify. Our article on choosing a front-end framework goes deeper on the build side.
Common questions
Is MCP the same as UCP?
No. MCP is a general standard for connecting AI tools to any system. UCP is a commerce-specific standard for the shopping journey. UCP can use MCP as one of the ways agents and stores talk to each other.
Do Shopify Plus merchants need to implement UCP themselves?
Generally no. Shopify co-developed UCP and handles much of the implementation for stores on its platform. What merchants control is the quality of their product data and which AI channels they participate in.
Is shopping in ChatGPT over?
Buying directly inside ChatGPT through Instant Checkout ended in March 2026. Discovering products through ChatGPT did not. Shoppers still research there and then complete the purchase on the merchant's site.
Does a headless storefront help or hurt with AI agents?
It can help, because an API-first stack is easier for agents to read and connect to. It can also hurt if your custom storefront drifts from the data your platform shares. The architecture matters less than keeping one consistent source of truth.
Which protocol will win?
Probably more than one, each doing a different job. That is why the safest investment is in your data and your checkout, which every protocol relies on.
Where to start
Start with your catalog and your checkout, then decide which AI channels are worth your attention. If you want a second opinion on your stack or your roadmap, get in touch. For more on why architecture matters here, read why AI agents need a composable architecture.
