If you update a GPT’s instructions, the new instructions will only apply to conversations initiated after the update and won’t affect existing ones. For some GPTs, like role-playing games, user progress is crucial, and many such GPTs with complicated instruction may require frequent updates and iterations. Given the current technological conditions, is there a way to optimize this issue?
Hi!
Have you tried to put your instructions in a knowledge file or get them via a function call from an external source?
I’m sure the second option could resolve your issue.
That is correct. This is normal, expected, behavior and by design.