I still manually synthesize information from multiple sources(docs,blog posts and treads).LLM help with sumaries but merge into a reliable decision step is very human-heavy.