Hey all,
Last month OpenAI updated their APIs, and my company InfiniteUp is just updating now.
I see there is a feature called 'Assistants' (API docs here) that does exactly what we want it to do. It passes each chat message into a data repository that than builds up to be referenced in future queries: cool right.
So if it's a Physical Trainer AI, each message gets added to the knowledge base so that the GPT has a better odds of referencing correctly. In our current integration we were working off summarizing data and sending with each message.
If anyone would like to share their 2 cents I'd appreciate it, I feel like there are multiple ways to work at this and I'd like to be mindful of our choices.