Commerce is complex, shaped by billions of transactions and millions of merchants. Shopify has spent over two decades understanding this landscape, recognizing that variations in payment options, fulfillment methods, and discount rules create a labyrinth of challenges for retailers. To address this, Shopify, in collaboration with Google, developed the Universal Commerce Protocol (UCP). This open standard allows AI agents to connect with merchants flexibly, enabling them to define their capabilities and negotiate transactions based on their unique needs. UCP distinguishes itself by separating responsibilities into layers, a strategy that promotes adaptability. The core transaction primitives are defined by the shopping service, while capabilities like Checkout and Orders are independently versioned. Extensions allow merchants to customize their offerings without being constrained by rigid protocols. For example, fulfillment can involve various methods—shipping, local delivery, or split shipments. Merchants can create extensions tailored to their specific requirements, ensuring the protocol evolves alongside commerce without breaking existing functionalities. The protocol facilitates negotiation between agents and merchants, similar to how HTTP operates with content types and accept headers. Each party publishes profiles declaring their capabilities, allowing for dynamic interaction. Notably, UCP uses reverse-domain naming for extensions, eliminating the need for central approval processes. This model fosters an open ecosystem where capabilities can evolve freely, reflecting the diverse needs of commerce. UCP also addresses situations where human intervention is necessary. Certain transactions might require human input due to regulatory constraints or complex requirements. The protocol incorporates a state machine that tracks the checkout process, ensuring that no transaction is left incomplete. Embedded Checkout Protocol (ECP) enhances this by allowing seamless handoffs between agents and merchants, maintaining structured communication and context throughout the transaction. This collaborative approach ensures that as agents' capabilities grow, they can handle more transactions autonomously while still providing a pathway for human assistance when required.