The following is my understanding of using vector data with Pinecone. I'm not sure if it's correct or not.
If we store user chat context data in Pinecone and retrieve it every time a new chat is input to pass as context to ChatGPT, what is the difference between storing the data in Pinecone and storing it on our own server?
If we save all chat records from the beginning of using the chat tool to Pinecone, and then retrieve the top 10 (customizable) most similar records based on the user's latest input from Pinecone, is it correct to understand that this retrieval process searches through all chat content and then passes the obtained data as context to ChatGPT?
