I built a RAG pipeline that retrieves live web data before generating answers. The goal was simple. I wanted a pipeline that could work with information that changes, instead of relying on a static collection of documents. Here's how I approached it. 🧵