Show HN: Gretel – Topic specific feeds built from YouTube recommendations

Category: ai-ml

Tags: youtube, recommendation-system, embeddings, desktop-app, tauri

Score: 6.8/10 (Innovation: 6, Technical: 7, Documentation: 8, Utility: 6)

Gretel is a local desktop app that builds focused YouTube feeds by leveraging YouTube's own recommendations through YT.js, then uses embeddings to filter and adapt content based on user-defined topics and channels. It offers separate profiles without extra Google accounts, local SQLite storage, and optional managed or bring-your-own OpenRouter keys for embeddings. The project is interesting for its novel combination of YouTube recommendation scraping, embedding-based relevance filtering, and a privacy-conscious local-first approach.

Target audience: end users seeking customized YouTube feeds, privacy-conscious users, and developers interested in recommendation systems

Repository: https://github.com/Relic-a/Gretel/ · TypeScript · Apache-2.0 · 1 stars

View on Hacker News