everyone is talking about RAG, so I went down the rabbit hole and found the boring part that actually makes it work: ret...

everyone is talking about RAG, so I went down the rabbit hole and found the boring part that actually makes it work: retrieval and chunkingover the last months I've been studying and experimenting with vector databases and local LLMs, and I turned that into a three–part series on the Storyblok blogtoday the last and most challenging piece is out: a step-by-step guide to building a fully local RAG pipeline with @weaviate + @ollama in the article I show, with real code and a full repo linked:• why hybrid search (BM25 + vectors) beats pure vector search• how bad chunking quietly ruins most RAG systems• why a smaller model + good retrieval often beats a huge (expensive) model with bad context• how a structured CMS (like Storyblok) basically gives you chunking for freestack:• Weaviate for vectors and hybrid search• Node.js for the glue• Qwen 3.5 on Ollama running locally (but this works with cloud models too)if you work on docs, DX, or AI features for content-heavy products, this might be a...