Why Spatial Intelligence Is a Core Layer of Agentic Commerce
Agent checkout specifications record who paid and what was approved, but not how the item looks on the person buying it.
Open specifications published by payment networks and technology companies now let a software agent buy on a person’s behalf. Together they define a product feed, a checkout session, a payment credential limited to a single merchant and cart, and a signed record of what the buyer approved. For goods that a part number, a price and a delivery date describe completely, that set is close to sufficient: a replacement filter does not need to be seen before it is bought. Apparel and beauty purchases end at a different question: how the item will look on the person buying it. In the checkout formats examined here, appearance arrives as the merchant’s catalog images, and none has a field for how the item looks on the buyer. Bain & Company forecast in December 2025 that considered purchases such as apparel will follow spec-driven ones to agents as trust grows. This essay reads the specifications at field level, alongside returns data, and argues that for goods bought by eye the purchase depends on an input the rails do not carry, a view of the item on the buyer. Producing that view is a task for spatial intelligence, the area of AI that perceives and reasons about people and objects in space. That makes spatial intelligence a core layer of agentic commerce.
That view is the shopper’s side of the purchase, and it is FaceCake’s field: FaceCake builds visual commerce software for beauty and fashion, and as a leader in AR try-on it already shows shoppers products on themselves. Any such view has to meet three conditions: it has to show this item on this buyer rather than a catalog photograph, rest on the shopper’s consent and hold for every buyer. The essay closes with FaceCake’s answer to each.
