Agentic Commerce Protocol (ACP)

Definition

The Agentic Commerce Protocol (ACP) is Shopify’s open protocol that enables AI agents to perform commerce actions - searching for products, reading details, adding items to a cart, and initiating checkout - on behalf of a human user. Built on top of the Model Context Protocol (MCP), ACP adds a commerce-specific layer that turns any Shopify store into a machine-readable shopping destination.

ACP was announced by Shopify CEO Tobi Lutke in early 2025 as part of Shopify’s push to make its merchants visible in AI-driven shopping experiences. Rather than waiting for AI companies to figure out commerce, Shopify built the protocol itself and made it available to all Shopify merchants.

The protocol exposes store data through structured MCP tools. An AI agent can call tools like searching a product catalog, retrieving detailed product information including variants and pricing, and generating checkout links. The store owner does not need to build anything custom - ACP works with existing Shopify product data.

Why It Matters

ACP is the first major ecommerce protocol designed specifically for AI agents. Before ACP, AI assistants could only recommend products by linking to websites. With ACP, they can actually browse a store’s catalog, compare products, check availability, and guide the user through to purchase - all within the AI conversation.

For Shopify merchants, ACP is automatically available. Every Shopify store has an MCP endpoint that AI agents can query. This means millions of stores are already participating in the agentic commerce ecosystem, whether their owners realize it or not.

The competitive significance is substantial. Shopify moved first among major ecommerce platforms to build native AI agent support. Merchants on other platforms - WooCommerce, PrestaShop, Magento - do not yet have equivalent protocol-level access for AI agents. This gives Shopify merchants an early advantage in AI-driven product discovery.

However, ACP is Shopify-specific. It only works with Shopify stores. This has led to the development of competing approaches like UCP (Universal Context Protocol) that aim to work across all platforms.

How It Works

ACP exposes several core capabilities through MCP tools:

Product Search lets AI agents query a store’s catalog using natural language or structured filters. The agent can search by product type, price range, availability, and other attributes.

Product Details provides full product information including descriptions, images, variants (size, color), pricing, and inventory status. This gives the AI agent everything it needs to make an informed recommendation.

Cart and Checkout generates checkout links that the user can follow to complete their purchase. The AI agent does not handle payment directly - it creates a link to Shopify’s secure checkout flow where the user enters their payment details.

A typical ACP interaction looks like this: a user asks an AI assistant to find running shoes. The AI discovers the store’s ACP endpoint, searches the catalog, retrieves detailed information about matching products, presents options to the user, and when the user decides, generates a checkout link. The user clicks through and completes the purchase on Shopify’s checkout page.

For merchants, the key lever is product data quality. ACP serves whatever data is in your Shopify store. If your product titles are vague, your descriptions thin, or your attributes incomplete, that is exactly what AI agents see. Rich, structured product data translates directly into better AI-driven discovery and conversion.

Related Terms

• Model Context Protocol (MCP) - The underlying protocol that ACP is built on
• Universal Context Protocol (UCP) - A competing protocol aiming to work across all ecommerce platforms
• Agentic Commerce - The broader movement ACP enables
• Product Feed - The structured product data that ACP exposes to AI agents