AI Dictionary › Prompting

Self-Refine

Self-refine is a loop where the same model generates a first answer, critiques it against explicit criteria, and rewrites an improved version, repeating until the result is satisfactory. No external tools are needed: generation, feedback and revision all live in the prompt. Example: "Step 1: write the customer reply email. Step 2: list 3 flaws regarding clarity, tone and completeness. Step 3: rewrite the email fixing them." The model produces, self-evaluates and improves, often with a clear quality jump over the first draft.

Definition

Use it when the first answer is good but improvable: writing, code, arguments. It works best with precise critique criteria; otherwise the model makes only cosmetic edits. Mind the cost: each cycle spends tokens.

Related terms

• Reflexion
• Chain of Thought
• Self-Ask
• Prompt Iteration
• Chain-of-Density

More in Prompting

• Prompt
• Prompting
• Role Assignment
• Zero-shot
• Few-shot
• Chain of Thought
• Prompt Engineering
• Output Format

Put it into practice

• Real scenario: apply this concept with an instant AI verdict.
• Daily gym: short drills to train dictionary concepts.
• Playoff game: challenge other players, elimination-style.

📱 Download the Android app (beta) iOS coming soon

Say what you mean. Get what you need.

Grace Certified, the AI coach that trains and certifies your prompt engineering, by Agora Intelligence.