Skip to main content

Chapter 3: Step-by-Step Reasoning

🎯 Learning Objectives​

  • Decompose complex tasks into manageable steps
  • Create verification checkpoints
  • Implement self-correction prompts
  • Build systematic problem-solving frameworks

Task Decomposition​

Break complex tasks into smaller, manageable steps:

Before (Monolithic)​

"Write a complete business plan for a mobile app."

After (Decomposed)​

"Let's create a business plan for a mobile app, 
step by step:

Step 1: Define the problem being solved (2-3 sentences)
Step 2: Describe the target audience (demographics, needs)
Step 3: Outline the solution and key features (bullet points)
Step 4: Analyze competition (3 competitors, brief comparison)
Step 5: Revenue model (how the app makes money)
Step 6: Marketing strategy (3 main channels)
Step 7: Key metrics for success (5 KPIs)

Complete each step before moving to the next."

Verification Steps​

Add self-checking to improve accuracy:

"Solve this problem, then verify your answer:

Problem: [problem]

1. Show your solution
2. Check: Does each step follow logically?
3. Check: Is the final answer reasonable?
4. Check: Can you verify with a different method?
5. Final confirmed answer: ___"

Self-Correction Prompts​

Encourage the AI to catch its own mistakes:

"After completing your analysis:
- Review for any logical gaps
- Identify any assumptions you made
- Note any areas of uncertainty
- Suggest what additional information would help"

Framework: STAR Problem Solving​

Use the STAR framework to analyze this situation:

S - Situation: What's the context?
T - Task: What needs to be accomplished?
A - Action: What steps should be taken?
R - Result: What outcome is expected?

Then review: Are the actions likely to achieve the result?

Progressive Refinement​

"Draft 1: Write a quick first version
Draft 2: Improve clarity and structure  
Draft 3: Polish language and tone
Final: Check against all requirements"

Key Takeaways​

Summary
  1. Decompose complex tasks into clear steps
  2. Add verification checkpoints for accuracy
  3. Enable self-correction by prompting review
  4. Use frameworks (STAR, etc.) for structure
  5. Progressive refinement improves quality

← Previous: Role-Playing and Personas | Take the Quiz β†’