I am dabbling with Azure's OpenAI service, and I want to be able to store previous conversations for future recall. I know that I can take each prior prompt-and-answer and send them as part of the new prompt, but obviously this will eventually hit my limit for tokens. I also realize I can also asked the AI to summarize the previous conversations, and then only pass on that more condensed set of information. In either case, there's still always the risk of running out of space or losing information.
Is there any way to take prior conversations, store them in an external resource, and then tell the AI to use that resource when forming its replies?
