Skip to content
ATOMZ.

Fix the catalog

Taxonomy and metafields: how to make a catalog AI can read

By Ankit Minocha. Founder, Atomz. Updated May 25, 2025.

A readable catalog has two things: every product mapped to the Shopify Standard Product Taxonomy, and the attributes for each product written to metafields. That is the layer an agent reads.

The taxonomy

The Shopify Standard Product Taxonomy is an open classification of categories and the attributes that belong to each. Mapping to it is what makes your products legible to agents and search engines that already expect that structure.

The metafields

Metafields are where structured attributes live in Shopify, things like material, fit, use case, and compatibility. They stay in your store, owned by you and readable by every tool that reads metafields.

This is a stub article. Replace the body with the full piece from your content files.

Subscribe to The Agentic Operator

Catalog and agentic commerce, for Shopify operators. No fluff. Unsubscribe anytime.

Keep reading

Shopify infers your product attributes. Here is why that is a problem.

Shopify Catalog now auto-infers material, style, and occasion from your product data, marked as Inferred with varying accuracy. When you do not fill the real attributes, agents shop on Shopify's guess.

Every Shopify store now serves an agents.md. Here is what to do with it.

Shopify now auto-serves /agents.md, /llms.txt, and /llms-full.txt on every store, the canonical files AI agents read to learn how your store works. Most merchants will leave the defaults. That is a mistake.

Sitemap.xml for AI discovery: helping agents find every product

A sitemap is how an agent finds the products that are not linked from your homepage. Here is how Shopify's sitemaps work and how to make sure your full catalog is discoverable.