So I have a custom Assistant linked to an iOS / Android App.
I am very impressed with the new Apps-SDK and want to move a variant of the app inside ChatGPT & I’m trying to work out the best architecture.
At present, my app goes back and forth with the Assistant - via an in app chat view, and then the final produced output content (a text file) can be saved and manipulated inside the app and separate backend DB.
The text file is initially created and then modified by the assistant before being finalised.
This latter part - where the content is finalised away from the Assistant would naturally port over to a ChatGPT app, but I’m not clear on whether you’d move the assistant to the app - with the associated system instructions, or use a separate companion agent, and how that might work.
Any thoughts?