EdgeReco

On-device search and recommendations

EdgeReco moves product search, ranking, and recommendations out of a paid backend and into each shopper's browser. A store publishes one signed catalog bundle; the browser verifies it, stores it locally, and answers queries from SQLite with FTS5 and vector search.

What runs locally

Why it matters

No per-query server cost, no catalog API latency, and no upload of shopper behavior. The engine is open source and the bundle is content-addressed and signature-verified.

Try the live storefront · Source on GitHub

Frequently asked questions

Is there really no backend?

After the signed catalogue is synced, search and recommendations run in the browser. There is no per-query server call.

Does it work offline?

Yes. The catalogue and engine are stored on the device; after the first sync the storefront keeps working offline.

What data leaves the browser?

No shopper activity is sent to a server. The catalogue is fetched from the store's origin and verified locally.