Chain-of-Thought Reasoning

The Prompt

Solve this problem using explicit chain-of-thought reasoning:

Problem: [YOUR QUESTION OR PROBLEM]

Follow this thinking structure:

**Step 1 — Understand:**
- Restate the problem in your own words
- Identify: what's given, what's asked, what's ambiguous
- Ask: what type of problem is this? (calculation, decision, creative, analytical)

**Step 2 — Decompose:**
- Break into sub-problems
- Identify which sub-problem to solve first
- Note dependencies between sub-problems

**Step 3 — Solve each sub-problem:**
For each:
- State your approach
- Show your work (every logical step)
- Check: does this intermediate result make sense?
- Note assumptions made

**Step 4 — Synthesize:**
- Combine sub-problem solutions
- Check for consistency between parts
- Verify: does the overall answer satisfy all original constraints?

**Step 5 — Verify:**
- Try to disprove your answer (what would make it wrong?)
- Sanity check: is the magnitude reasonable?
- Alternative approach: can you get the same answer a different way?

**Final Answer:** [Clear, concise]
**Confidence:** [High/Medium/Low with reasoning]
**Key Assumption:** [The one thing that if wrong, changes the answer]

How To Use It

Use this for any problem that requires multi-step reasoning — math, logic, analysis, complex decisions. The explicit structure prevents the AI from jumping to conclusions and catches errors through the verification step.

Why It Works

AI models produce better answers when forced to show their work. Each reasoning step becomes a checkpoint where errors can be caught before they compound. The verification step catches approximately 30% of errors that get through the initial reasoning.

Variations

Think step by step (minimal):

Think step by step. For each step, explain your reasoning. After reaching a conclusion, try to find a flaw in your reasoning. Correct if needed.

Structured debate:

Consider this question from three perspectives (optimist, pessimist, realist). Let them debate. Then synthesize into a balanced conclusion.

Rubber duck reasoning:

Explain your reasoning as if to a smart 12-year-old. If you can't explain a step simply, you probably don't understand it well enough — dig deeper.