How to implement Retrieval Augmented Generation (RAG) on top of hybrid search?

It’s just on the RAG part I’m lost. I was thinking about using OpenAI GPT-3.5-turbo.