auto-news

Why we included this project

Keeping up with a scattered mix of RSS feeds, Reddit threads, tweets, and YouTube channels is usually a time sink. This project pulls those sources into one pipeline, runs each item through an LLM via LangChain to generate summaries and insights, and drops anything that doesn't match your stated interests, so you're left with a short, readable briefing. It can transcribe YouTube videos that lack a transcript before analyzing them. You can choose between OpenAI, Gemini, or a local Ollama backend, which helps if you want to keep inference off external servers. The results land in Notion, and since the backend runs with docker-compose or Kubernetes, self-hosting is a real option.