Building a Fabric knowledge hub using AI and Obsidian by Tom Martens

6,000+ articles from curated sources.
An automated content pipeline that never sleeps.
A vector store that understands entire articles, not fragments. And an honest-first architecture designed to catch errors before they compound.
This session walks through building a personal RAG system for Microsoft Fabric — from content gathering to retrieval to synthesis — and the architectural decisions that matter more than the technology: source hierarchy, bias avoidance, and why error laundering is real.