Question
One analyst can test an allowed model in Launchpad, and the same AI Core instance accepts API calls through its service key. A colleague can also save prompts, but this analyst has only experimenter roles and sees no Save action. Which change fixes the analyst's problem without altering runtime governance?