Design, log, compare, and score prompt experiments so users can systematically improve outputs instead of guessing.
