By category

AI agents, by what you sell

What stops an AI shopping agent depends a good deal on what you sell. Apparel fails at the size picker, food and drink at stock that moved since the feed was written, home and furniture at a delivery cost nobody publishes. These pages describe the failure each category runs into structurally, rather than how often it happens.

• AI agents and apparel stores

  Apparel is the category where the buy stage fails most structurally, because almost every purchase requires choosing a size and a colour before anything can go in a basket. A control that a person operates with a mouse and an eye is the single point an agent has to get through, and it is the point most often built without a form field behind it.

• AI agents and beauty stores

  Beauty catalogues are wide, shallow and heavy on shades, which makes them easy for an agent to read and hard for it to choose from. The recurring failure is not the buy path but the evaluate stage: an agent asked for a foundation for a given skin tone needs shade and ingredient information as data, and it is usually published as pictures and prose.

• AI agents and electronics stores

  Electronics is the category where identifiers matter most. An agent comparing a specific model across stores needs a GTIN or an MPN to know it is looking at the same thing, and a listing without one is a listing it cannot confidently include in a comparison. The buy path is usually simpler here than in apparel; the evaluate and compare stages are harder.

• AI agents and food and drink stores

  Food and drink is dominated by things that change: stock that turns over daily, dates, and subscriptions. An agent reads a page once and acts on what it says, so a catalogue whose truth has a short half-life punishes any gap between the feed and the page harder than a category where stock sits still.

• AI agents and health and wellness stores

  Health and wellness carries the most constrained language of any retail category: what a product may claim is regulated, so the useful facts are stated carefully and often as images. That combination, high-stakes questions and low-structure answers, is what makes the evaluate stage the hard one here.

• AI agents and home and furniture stores

  Home and furniture is where delivery is part of the product. A sofa an agent cannot get a delivery cost or a lead time for is one it cannot compare, and those facts sit outside the product page more often than in any other category. Dimensions matter for the same reason and are usually in an image.

• AI agents and jewellery and accessories stores

  Jewellery is small-catalogue, high-consideration and heavily photographed, which is the combination that produces the most beautiful pages an agent can read least. Materials, sizes and hallmarks are what a shopper asks about, and on most stores they are in the photography and the prose rather than in anything structured.

• AI agents and outdoor and sports stores

  Outdoor and sports combines apparel’s variant problem with electronics’ specification problem: a boot has sizes and widths, and a tent has weights, seasons and capacities. Both halves have to be readable for an agent to answer the question it was sent with, and stores usually get one of the two right.