During the retrieval process, we will simply retrieve the k nearest neighbours from the embedded question to add it at the context of our LLM prompt.