Back to prompts
AI Masteryadvanced
4.7

Chain-of-Thought Designer — Make AI Think Step-by-Step for Complex Problems

Transform any complex question into a chain-of-thought prompt that forces AI to reason through problems systematically instead of jumping to conclusions.

Copy & Paste this prompt
You are a prompt engineering specialist focused on reasoning and chain-of-thought techniques. Help me get better AI reasoning.

My Problem:
- What I'm asking the AI to do: [DESCRIBE THE TASK]
- Why it's complex: [What makes this hard for AI?]
- Current output quality: [What's wrong — hallucinations? Logical errors? Missing steps?]
- Model I'm using: [GPT-4 / Claude / Gemini / etc.]

Design my chain-of-thought prompt:

**1. PROBLEM DECOMPOSITION**
Break my complex task into sequential thinking steps:
- Step 1: [What should AI think about first?]
- Step 2: [What depends on Step 1?]
- Step 3: [Building on previous steps...]
- Step N: [Final synthesis]

**2. THE CHAIN-OF-THOUGHT PROMPT**
Rewrite my task as a structured reasoning prompt with:
- Explicit instruction to "think step by step"
- Intermediate checkpoints (verify before proceeding)
- Self-correction triggers ("Check: does this make sense?")
- Final answer clearly separated from reasoning

**3. FEW-SHOT EXAMPLES**
Provide 1-2 worked examples showing:
- The input
- The step-by-step reasoning (showing the 'thought process')
- The final answer
- How errors are caught and corrected mid-chain

**4. REASONING PATTERNS**
Which reasoning strategy fits my problem best?
- Linear chain (A→B→C→Answer)
- Tree of thought (explore multiple paths, pick best)
- Self-consistency (solve 3 ways, check agreement)
- Recursive decomposition (break into sub-problems)
- Adversarial checking (argue both sides)

Recommend which to use and why.

**5. VERIFICATION LAYER**
Add a self-check at the end:
- "Review your reasoning. Are there any logical errors?"
- "Does your conclusion follow from your premises?"
- "What's the strongest argument AGAINST your answer?"
- "Rate your confidence 1-10 and explain why."

**6. TEMPERATURE & SETTINGS**
- Recommended temperature for this task
- Whether to use system prompt or user prompt for the chain
- Max tokens consideration
- When to split into multiple calls vs. one long chain

The goal: AI that REASONS through problems rather than pattern-matching to the most likely-sounding answer.
Use as Template
#chain-of-thought#reasoning#prompt-engineering#AI-techniques#problem-solving

Works with

chatgptclaudeany

💡 Pro Tips

  • •Chain-of-thought works best for logic, math, and multi-factor decisions — less needed for creative tasks
  • •Ask follow-up: 'Now apply this chain-of-thought to MY specific decision: [your problem]'
  • •Adding 'Think step by step' alone improves accuracy by 10-15% on reasoning tasks

✨ Example Output

🧠 PROBLEM: 'Should I accept this job offer?' (complex multi-factor decision)

❌ NAIVE PROMPT: 'Should I take this job?'
Result: Generic pros/cons list

✅ CHAIN-OF-THOUGHT PROMPT:

"I need help making a career decision. Think through this step by step.

Step 1: List all factors that matter in a job decision for someone at my career stage (5+ years experience, considering long-term trajectory).

Step 2: For EACH factor, score both my current job and the new offer on a 1-10 scale. Show your reasoning for each score.

Step 3: Weight the factors by importance to ME based on what I've told you about my priorities.

Step 4: Calculate the weighted score for each option.

Step 5: BEFORE giving your recommendation — identify what assumptions you made. Which scores are you least confident about? Flag these.

Step 6: Consider what could go WRONG with each choice. What's the worst realistic outcome?

Step 7: Now give your recommendation with confidence level (1-10).

IMPORTANT: If your confidence is below 7, tell me what additional information would raise it."

📊 REASONING PATTERN: Tree of Thought (multiple evaluation paths converging)
- Path 1: Financial analysis
- Path 2: Career trajectory analysis  
- Path 3: Lifestyle/values alignment
- Convergence: Which path matters most given stated priorities?

🧠 Why This Works

AI models are pattern matchers — they generate the most likely-sounding response, which isn't always logically correct. Chain-of-thought prompting forces the model to show its work, catch errors mid-reasoning, and arrive at conclusions through logic rather than vibes. Research shows CoT improves accuracy by 15-40% on complex reasoning tasks.

📅 When to Use This Prompt

For any task requiring logical reasoning, multi-step calculations, complex decisions with trade-offs, tasks where AI tends to hallucinate or make logical errors, or problems that require weighing multiple factors simultaneously.

🎯 What You'll Get

A structured prompt that produces step-by-step reasoning with self-correction, appropriate for your specific problem type. You'll get more accurate, verifiable AI outputs — and when the AI is wrong, you can see exactly WHERE in the chain the error occurred.

🔗 Related Prompts