I just wanted to organize my Obsidian. I ended up writing two indexes, semantic search, and RAG

The author shares their experience developing Vault Audit AI, a tool for automating note management in Obsidian. Initially conceived as a simple AI audit for a knowledge base, the project evolved into a comprehensive system. It now features two types of indexes, semantic search, a 'similar notes' function, duplicate detection, and a RAG (Retrieval-Augmented Generation) implementation for local storage. The article details the system's architecture, the technical challenges encountered, and lessons learned from the testing process. Despite having nearly 700 tests, real-world usage in Obsidian presented unexpected hurdles, highlighting the complexity of integrating AI tools into local knowledge management systems. This article is a valuable resource for developers interested in implementing LLMs and vector databases into personal workflows.