CatalogAI

The product content linter for AI discovery

What it is

CatalogAI scans a Shopify store's product catalog and gives it an AI Discoverability Score from 0 to 100. Think of it like a health checkup for your product data — it tells you what's broken, what's weak, and fixes it in one click.

The end goal is simple: make stores visible to AI shopping agents like ChatGPT Shopping, Perplexity, and Google AI Mode. If an AI can't understand your products, it can't recommend them.

The story behind it

I kept seeing the same thing in Shopify communities — merchants confused about why their traffic was dropping even though their SEO looked fine. The answer hit me when I started digging into how AI agents actually shop.

They don't browse. They query structured data. And most stores have terrible structured data.

When someone asks ChatGPT "find me a blue yoga mat under $50 with good grip," it's not crawling websites — it's reading product attributes, metadata, taxonomy mappings. Most merchants score 15-35 out of 100 on their first audit. They had no idea.

Then Shopify dropped the Catalog API in Winter 2026 and I realized — the infrastructure to fix this just became available, and nobody's built the tool yet. "AI Discoverability" isn't even a category on the App Store. I want to be the one who names it.

How it actually works

The audit engine pulls every product, variant, and metafield from a store via Shopify's Admin API and the new Catalog API. Then it runs about 30 checks per product:

Structured attributes — can a machine read your product properties?
Description quality — is there enough for an AI to reason about the product?
Image alt text — AI agents care about this more than you'd think
Category mapping — does the product fit Shopify's taxonomy correctly?
SEO metadata — titles and descriptions optimized for how AI queries work

The fix engine is the fun part. It uses the Anthropic API to rewrite metadata, but it's not doing generic SEO copywriting — it's structuring data so machines can actually reason about products. There's a real difference between "great for yoga" and having the right attributes in the right fields.

Large catalogs can have thousands of products, so audits run as background jobs through BullMQ and Redis. Merchants see a progress bar, then get a full report with every issue prioritized by how much it hurts their discoverability.

The business side

The free tier is the whole strategy. Every merchant gets their score and the full issue list for free — no credit card, no trial. Once they see a score of 23 out of 100 and a list of 47 fixable issues, the paid tiers ($49-$399/mo) sell themselves.

I'm betting on the score being shareable. Merchants posting "went from 21 to 78!" is the growth engine.