I built a self-hosted news digest that re-ranks itself based on what you actually read

Description and use

CondenseIt is a self-hosted daily news digest that fetches, summarizes with a local LLM, and re-ranks articles based on user interactions, providing transparent explanations and running efficiently on modest hardware.

Problem

RSS readers are stateless and repeat the same feed order daily, offering no personalization or memory of user preferences.

Outcome

CondenseIt delivers a personalized, self-hosted daily news digest that learns from explicit and implicit user interactions, providing transparent ranking explanations and running efficiently on a Mac Mini without a discrete GPU.

Workflow

• Configure source feeds via admin panel
• Schedule daily run
• Fetch articles from RSS, YouTube, Reddit, Hacker News, GitHub, Google News, and website change detection
• Summarize each article locally with Ollama (or OpenRouter)
• Score articles using 11 signals (keyword overlap, bigram matching, TF-IDF similarity, category/source averages, implicit engagement, explicit ratings)
• Apply exponential decay to ratings
• Render ranked list in web app with transparency panel
• Allow user to rate articles and adjust preferences

Evidence

I use RSS readers constantly. The problem I kept running into was that they are stateless.

Source

CondenseIt is a self-hosted daily news digest. You point it at your sources, it runs on a schedule, summarises everything using a local LLM, and serves the results as a web app you can read in the browser.

Source

Summaries run through Ollama locally, Metal-accelerated on Apple Silicon. No discrete GPU needed.

Source