If you have spent the last couple of years wiring up Python AI applications with Semantic Kernel, you probably have a healthy respect for how messy retrieval-augmented generation (RAG) and agent memory can get when database SDKs start bleeding into your application logic. Rather than tearing up...
