Headline 2030 analyst projections for "AI-driven" or "agentic" commerce range from 10% to 25% of US ecommerce (e.g., Bain, Morgan Stanley, Gartner) 3 sources. However, these estimates broadly include transactions that are merely initiated or influenced by AI but completed on a retailer's own site, as well as retailer-hosted agents (like Amazon's Rufus). Because this strict definition requires the checkout to occur entirely outside the retailer's own website or app, a significant h

When estimating "fully-agentic checkout," it is critical to separate it from the broader "agentic commerce" category. While analysts project broad agentic commerce—where AI initiates or influences purchases but humans still confirm them—to reach 10–25% of US ecommerce by 2030 2 sources, fully-agentic checkout is a much stricter subset. It requires that the human neither visits the retailer's site nor reviews the specific transaction. Consequently, fully-agentic volume will be only a small

The strict resolution criteria for "agentic checkout" require that the transaction occurs completely off the retailer's own site or app. This definition effectively excludes the single most common pattern in the market today: AI-assisted discovery followed by a click-through to a retailer's site. Because most headline forecasts for "agentic commerce"—such as Bain's 15-25% projection bain.com or EMARKETER's broader estimates emarketer.com—bundle in AI-influenced sales, the genuinely agent-exec

The strict requirement of "no site visit AND no human review" heavily constrains this market. In the UK, broad agentic commerce infrastructure is advancing, but the fully autonomous subset is severely limited by a documented trust deficit. Worldpay data indicates UK shoppers expect roughly 7% of online purchases to flow through AI agents by 2030, but crucially, only 6% of consumers would permit an agent to complete a transaction without prior approval ecommercenews.uk. Factoring out the high-tru