True Norma launch announcement

I set out to build a DTC store almost entirely with AI.

It almost worked.

The build was faster than I expected. Category research, brand strategy, the Shopify store, product pages, schema. Weeks of work compressed into a few days. AI handles the foundational work well when what you’re feeding it is accurate.

What I was feeding it wasn’t always accurate.

Dropshipping suppliers have a data quality problem. Specs incomplete. Warehouse locations listed but not verified. Compliance documents thin. Travel toys, the first category, didn’t make it through sourcing. I scrapped it and rebuilt around home wellness accessories.

The second test was harder to get a signal on.

True Norma is designed so AI agents can actually shop on it: structured product data, spec copy written as field:value pairs, llms.txt, crawlers allowed. An agent asked for a heated eye massager with confirmed US shipping shouldn’t have to infer anything.

Whether agents actually use it that way is still open. The structure is right. Discovery, trust, conversion. Unproven.

For DTC operators watching agentic commerce: what experiment should I run next?