Absolutely True not only for Local LLMs but for cloud ones too. Clarifying the intention, the type of output we want improves the model's response multiple folds.