The shift from AI-assisted to AI-executed
The first wave of AI in commerce was AI-assisted: a model helps a human find, evaluate, and decide, but the human still clicks "buy," fills the form, and completes the transaction. The next wave — already beginning — is AI-executed: an agent acts on the user's behalf, completing the full purchase, booking, or configuration flow without the human touching the site. For businesses, this is a discontinuous change. The buyer is no longer a person navigating your interface; the buyer is an AI agent that needs to interact with your systems programmatically.
WebMCP: giving agents a structured way in
WebMCP is an emerging protocol that lets a website expose its capabilities — product search, add-to-cart, checkout — as structured actions that AI agents can discover and call. Rather than an agent scraping HTML or simulating browser clicks, WebMCP provides a defined interface: here are the things you can do on this site, here are the parameters each action accepts, here is how to authenticate. Think of it as an API designed specifically for AI agents rather than for developers. A site that publishes a WebMCP manifest is telling every agent-equipped AI system: you can transact here, programmatically, without a human in the loop.
x402: machine-native payments
x402 is a payment protocol built for AI agents — a way for an agent to pay for something without a human entering payment details at a checkout screen. Named for the HTTP 402 "Payment Required" status code, x402 lets a server request payment and an agent to fulfil it in a single round-trip. For GEO, x402 matters because a site that cannot accept machine-native payment cannot complete an agentic transaction. An agent that reaches your checkout page and encounters a human-optimised form will abandon the purchase; an agent that reaches a site supporting x402 can complete it autonomously.
A wide-open frontier
Agent integration is the least-developed dimension of GEO across the commercial web. Analysis of over a hundred audited domains found that agent integration scores near zero across the board — not because the technology is impractical, but because almost no one has implemented it yet. This makes it the highest-asymmetry opportunity in the current GEO landscape: the technical bar is real, but the competitive field is almost entirely empty. The brands that build agentic commerce capability in the next 12 to 24 months will have established a durable first-mover position before the practice becomes standard.