commercetools has published a comprehensive guide explaining three core agentic commerce protocols and their distinct purposes. MCP (Model Context Protocol), created by Anthropic, standardizes how AI agents securely connect to external data sources, tools and APIs, enabling agents to access live inventory, loyalty program eligibility and customer preferences while respecting privacy boundaries. ACP (Agentic Commerce Protocol), developed by OpenAI and Stripe with commercetools support, focuses on secure checkout in GenAI environments; after OpenAI paused Instant Checkout in ChatGPT in March 2026, the protocol now supports product discovery with checkout completing on merchant sites, while OpenAI tests checkout plugins with approved partners. UCP (Universal Commerce Protocol), built by Google with partners including Walmart and Target, standardizes the entire shopping lifecycle from discovery and comparison through checkout, order tracking and post-purchase support, powering Google's Universal Cart live since October 15, 2026 for US shoppers.
For commerce practitioners, the choice of which protocols to implement depends on business goals and target AI channels. MCP enables custom AI shopping experiences and back-office agent work. UCP provides broad discoverability across Google's AI surfaces without multiple integrations. ACP remains relevant for the evolving OpenAI ecosystem, though its checkout approach continues to develop. Merchants can support autonomous agents managing the customer journey end-to-end while remaining the merchant of record, with AI agents able to manage carts, check inventory and complete transactions securely.
Rather than building and maintaining integrations for each emerging protocol independently—a resource-intensive undertaking requiring specialized expertise—commercetools offers abstraction layers through its Agentic Commerce platform and MCP implementations, allowing retailers to stay current with protocol evolution without heavy engineering lift.