More and more shoppers ask an AI assistant instead of a search engine: "Find me a rain jacket under €150 that ships to Germany." For the assistant to answer with real products, real prices and real stock, it needs a way to talk to shops. The Model Context Protocol, MCP for short, is the most widely used way to do that. This article explains what MCP is, what it means for online shops and what you can do today.
What is the Model Context Protocol?
MCP is an open standard that Anthropic introduced in November 2024. It defines how an AI application connects to external tools and data sources. Instead of building a separate integration for every assistant, a service offers one MCP server, and every MCP-compatible client can use it.
A common comparison is USB-C: one plug instead of a different cable for every device. For a shop that means one interface through which different AI assistants can search the catalog, check prices or prepare a cart.
How MCP works, in plain terms
- MCP server: runs on the service side, for example a shop or a deals platform, and offers tools such as "search products" or "get product details".
- MCP client: the AI application, such as an assistant or an agent, that calls these tools when a user asks something.
- The model decides: based on the user's question, the AI chooses which tool to call and turns the answer into a reply.
Why MCP matters for online shops
Until now, being found meant ranking in search results or marketplaces. With AI assistants a second path appears: being chosen by an assistant that can query your products directly. Shops whose data is reachable and complete can be suggested; shops that are not reachable are simply not part of the answer.
Shopify, MCP and UCP
According to Shopify's developer documentation, every Shopify store has its own Storefront MCP server through which AI agents can search the catalog, manage carts and answer questions about policies. Shopify states that its catalog tools implement the Universal Commerce Protocol (UCP), a commerce standard that Google and Shopify presented in January 2026. More on UCP in Google UCP: how to get your store discovered by AI agents.
For WooCommerce and other systems there is no built-in equivalent yet; plugins and third-party services fill the gap.
What you can do today
- Complete your product data. Title, description, GTIN, brand, price, availability, images and variants. An assistant can only recommend what it can read. See Structured data cheat sheet.
- Keep prices and stock current. An assistant that recommends an out-of-stock product loses the shopper, and you lose the next recommendation.
- Write clear policies. Shipping, returns and delivery times are questions shoppers ask assistants.
- Check your visibility. Our free UCP visibility test shows whether your shop exposes a readable catalog.
- Be where assistants look for deals. Platforms that offer MCP access to offers put your products in front of assistants without your own integration.
How Heartly uses MCP
Heartly runs its own MCP server for deals. AI assistants that connect to it can list current deals, filter them by category and look up a single offer, each with a link to the merchant's product page. Flash sales from the Heartly app and offers listed through Heartly Listing appear there automatically, alongside the Heartly app and deals.heartly.io.
Frequently Asked Questions
What does MCP stand for?
Model Context Protocol, an open standard for connecting AI applications to external tools and data, introduced by Anthropic in November 2024.
Do I need my own MCP server as a shop owner?
Not necessarily. Shopify provides a Storefront MCP server for each store. Complete product data and current prices matter more than your own server.
What is the difference between MCP and UCP?
MCP is a general protocol for connecting AI applications to tools. UCP is a commerce-specific standard that describes catalog, cart and checkout; it can be offered over MCP.
Which AI assistants support MCP?
Many current assistants and developer tools can act as MCP clients. Support differs by product and plan, so check the assistant you care about.
How do my products get recommended by AI assistants?
By being reachable and complete: a readable catalog, accurate prices and stock, clear policies, and presence on platforms that assistants query.