The transition from browsers to bots

The current state of online commerce is built for the human eye. Websites are collections of visual assets, tracking pixels, and complex UI flows designed to capture and keep human attention. This architecture is a significant barrier for AI agents. When a large language model attempts to buy a product, it has to scrape HTML, navigate fragile DOM structures, and interpret unpredictable checkout forms. The Commerce Operations Foundation (COF) exists to move commerce from this human-centric session model to a machine-readable protocol model.

Founded in 2024 and structured as a 501(c)(6) not-for-profit industry body, the COF is not a technology vendor. Instead, it is a standard-setting organization. Its primary output is the Commerce Protocol, a set of open standards that define how an AI agent should ask for a product price, check inventory, add items to a cart, and execute a payment. By formalizing these interactions, the foundation enables a world where an agent can complete a transaction in milliseconds via a structured API call rather than minutes of simulated web browsing.

A counterweight to platform lock-in

The organizational structure of the COF is a deliberate choice. By operating as an independent foundation, it attempts to solve a coordination problem between merchants, AI labs, and payment processors. In the current retail environment, companies like Amazon and Google exert massive control over discovery and checkout. If AI agents are to become the primary interface for shopping, there is a risk that a single AI provider could become a new kind of gatekeeper, charging 'agent taxes' or favoring specific merchants.

COF offers an alternative: an open, interoperable layer that any merchant can implement. This is similar to how the Linux Foundation or the PCI Security Standards Council operates. They provide the neutral ground where competitors can agree on the plumbing of the internet. For merchants, implementing COF standards means their inventory becomes 'agent-ready' without having to build custom integrations for every new LLM or agent framework that hits the market.

The technical architecture of intent

At the core of the foundation’s work is the definition of commerce 'intents.' This involves mapping the complex, often messy world of retail—discounts, shipping logic, tax calculations, and SKU variants—into a predictable JSON-based format. The foundation works closely with the Model Context Protocol (MCP) ecosystem, ensuring that the way agents 'see' a store is consistent across different platforms.

This protocol-first approach also addresses the security and identity challenges inherent in autonomous commerce. The foundation's standards cover how an agent proves it has the authority to spend a user's money and how a merchant provides a verifiable receipt. By standardizing these handshakes, the COF reduces the surface area for fraud and errors in autonomous transactions. While the organization is headquartered in London, its reach is global, reflecting the borderless nature of both commerce and the AI agent stack.