Private RAG over my notes, PDFs and contracts

I embed my Obsidian vault, PDFs and old contracts into a local vector store and query them through Qwen3 30B A3B, so I can ask 'find every note that mentions X' or 'what's the login procedure for Y' without a single file leaving my machine. This is the on-ramp use case that pulls a lot of professionals into local AI - once you see private RAG work on your own documents, going back to uploading files to a cloud chatbot feels wrong. Setup: Ollama + a vector DB (Qdrant/Chroma) + AnythingLLM or a small custom script.