Projects

LornAI.com

Connect product catalogs to AI shopping channels

Mayin JoshiNov 2025lornai.com

Several solutions exist for syncing product catalogs to channels like Amazon and TikTok Shop (Channel Engine, Feedonomics, and others), but none existed yet to easily sync and connect product catalogs from major retailers to AI channels like ChatGPT, Perplexity, and Gemini. And the retailers that build this integration out, and consistently send high quality product data, descriptions, and availability, will be prioritized by AI channels as they serve personalized recommendations to customers.

So singamreddy.com and I built out the platform, called Lorn. Lorn is the modern-day Feedonomics: we enable any retailer to connect their real-time product catalog to different AI channels, each with its own agentic commerce protocols and requirements. We build and maintain the integrations to all AI shopping channels, making becoming discoverable on the greatest new channel for shopping as easy as a couple of clicks.

Lorn translated existing merchant data and checkout systems into the formats AI shopping agents require.
Lorn translated existing merchant data and checkout systems into the formats AI shopping agents require.
  • Syncing product catalogs on a recurring schedule into several divergent protocol formats, including OpenAI's ACP, Google's AP2, and MCP, without a per-channel rebuild for the merchant.
  • Normalizing heterogeneous sources (storefronts, files, PIMs, order-management systems) into one canonical product model, then validating the required fields for each protocol.
  • Keeping price and availability fresh across every channel so AI agents never recommend stale inventory.
  • Routing delegated payments to the merchant's existing processor, like Stripe or PayPal, so Lorn never holds the money and completed orders land in the systems the business already uses.

The result is one technical integration instead of a separate project for every new AI surface: ingest the catalog without a migration, enrich the data, format it for agent protocols, publish and continuously refresh the feed, and receive orders through the existing stack.