How do evaluation metrics guide improvements in agent development?

• A. By hard-coding the best approach and not adapting.
• B. By training all models to maximize a single metric without comparisons.
• C. By quantifying performance across prompting strategies, tool sets, and configurations to compare and improve.
• D. By eliminating user feedback.

Answer: (C)