Inference, not segmentation
Weighted rules over request and behavior signals decide who is probably shopping—no account, no questionnaire, no stored profile required.
Commerce
Aisles
A storefront that reads the intent behind each visit and generates the category page to fit — same URL, same products, four different layouts.
- Status
- Live — available to use now
- Type
- site
- Access
- public
- Updated
- 2026-08-12
Live demo
What this work does
Aisles reads the intent behind each visit—search terms, referrer, behavior on the page—and generates the category page to fit one of four inferred shopper types. Same address, same products, four different layouts, each checked against a fixed schema before it renders. It draws its catalog live from the Kibble & Co. store.
Generated, but never invalid
The AI composes each page from a small fixed set of building blocks, and a schema check rejects anything outside it—so a generated page can differ, but it cannot break.
Pointed at a real store in a day
Connecting the engine to the Kibble & Co. catalog took one brand configuration file and two settings, because all catalog access already passed through one function.
Explicit relationships