RecoFeed

RecoFeed targets developers and product teams who want the engagement benefits of a recommendation engine without shipping their users' behavioral data to a third-party service. The pitch is straightforward: embed the feed, let users pick or confirm interests (the demo includes AI-generated category seeds), and the ranking runs locally against a vector index that lives on the device. Because the index is local, personalization still works in real time while the data stays with the user.

The product is built on CloseVector, a cross-platform vector database that supports browser, server and desktop environments. That shared substrate is what allows the same recommendation logic to run in a web page and still resolve similarity queries quickly. The site includes ready-made example feeds — for instance an English-language explanation of Tang poetry and a Chinese version of the same — to show what a compiled feed looks like out of the box.

Availability is the main caveat today: RecoFeed describes itself as being in closed beta, with sign-up required for access. Pricing follows a freemium shape — the basic recommendation feed is committed to remain free, while more advanced capabilities such as AI workflows are where fees come in. There is published documentation for teams that want to evaluate the integration before requesting beta access.