Thought exercise in action here.
I’ve been peripherally aware of uv for a bit. Now I’m ready to
dive in deeper. But the buzz on uv has been building for a
bit. What’s the aggregate knowledge in RSS feeds and podcast
subscriptions on uv?
If only I had a personal tool that could retrospectively search my
personal content to answer the request “Build a precis on uv for me”.
The ground is fertile for a tool I’m calling retrocast that would fit that bill. It would be a personal app hooking into all your feeds and constantly indexing the content. Layer in some text search, semantic search, knowledge graphs, and LLMs to craft products, not just 10 blue links.
Some components for inclusion and inspiration
- The Feedbin API using Feedbin’s backend as a feed database
- Airshow and Overcasat as podcasat players that make their data accessible
- Local embedded dbs and search engines such as SQLite, Kuzu, and meilisearch
- Local LLMs via ollama and llamafile
- NotebookLM and Kagi Assistant for UX ideas
- Simon Willison’s llm to glue it all together