Nice writing. I found that regular RAG doesn't always give good response, just wonder if GraphRAG is really better. I know the quality of response also has something to do with the LLM, for example, ChatGPT can give better answers than Llama using the same workflow. Anyway, the author provides sample code to use GraphRAG (kudos to that), I'll try to test if it's better for my use case.