If a user asks a vague question like "Tell me more about that," how do you ensure the retriever finds relevant documents from past conversation history?

If the retriever returns zero relevant documents, how do you prevent the LLM from making up an answer?

How do you handle a "Delete" request in your vector database if a user wants their data removed (Right to be Forgotten)?

Why might you use a "Cross-Encoder" re-ranker after your initial vector retrieval? What is the trade-off in terms of latency?