Skip to main content

📋 Planning Prompts

Planning prompts teach the AI to think ahead and create a structured plan before taking any action. Instead of jumping straight to an answer, the AI first outlines what it will do, why, and in what order.

It's like asking a builder to draw blueprints before laying bricks. The plan keeps everything organized and on track.

🎯 Why This Matters​

Without a plan, AI agents tend to:

  • Jump to conclusions without considering all aspects
  • Miss important steps in complex tasks
  • Lose track of what they've done and what remains
  • Produce inconsistent or incomplete results

Planning prompts solve these problems by forcing strategic thinking before execution. This leads to better, more reliable outputs for any complex task.

📖 The Concept in Detail​

What Makes a Good Plan?​

A good AI-generated plan has these elements:

  1. Goal clarity — What exactly needs to be achieved
  2. Step breakdown — Specific actions in logical order
  3. Dependencies — Which steps depend on others
  4. Success criteria — How to know each step is complete
  5. Contingencies — What to do if something goes wrong

Types of Planning Prompts​

TypePurposeBest For
Goal DecompositionBreak a big goal into sub-goalsLarge projects
Sequential PlanningCreate ordered step-by-step plansProcesses with dependencies
Strategic PlanningEvaluate options before choosingDecision-making tasks
Conditional PlanningPlan with if-then branchesUncertain situations
Resource PlanningIdentify what's needed for each stepProject management

The Plan-Then-Execute Pattern​

Phase 1: PLAN
- Understand the goal
- Identify all required steps
- Order steps logically
- Identify potential blockers

Phase 2: EXECUTE
- Follow the plan step by step
- Check off completed steps
- Adjust plan if needed

Phase 3: REVIEW
- Verify all steps completed
- Check if goal was achieved
- Note any lessons learned

💡 Prompt Examples​

Goal Decomposition Prompt​

You are a planning agent. Before doing anything, you MUST create a 
detailed plan.

Goal: Create a complete marketing strategy for launching a new 
mobile fitness app.

Instructions:
1. First, break this goal into 3-5 major sub-goals
2. For each sub-goal, list 2-4 specific action steps
3. Identify dependencies between steps
4. Estimate relative effort for each step (low/medium/high)
5. Present the plan in a structured format

Do NOT start working on the strategy yet. Only output the plan.
I will review it and say "Execute" when ready.

Strategic Thinking Prompt​

You are a strategic planning assistant. I need help deciding the 
best tech stack for a new e-commerce platform.

Before recommending anything:
1. List the key factors to consider (scalability, cost, team skills, etc.)
2. Identify 3 possible approaches with pros and cons
3. Score each approach on the key factors (1-10)
4. Recommend the best approach with justification

Think strategically. Consider both short-term and long-term impacts.
Present your analysis as a structured plan before giving any 
recommendation.

❌ Bad Example​

Help me migrate our company's database from MySQL to PostgreSQL.

Why it fails: No planning instruction. The AI might jump straight into SQL conversion commands without considering data mapping, downtime planning, backup strategy, testing, or rollback procedures.

✅ Improved Example​

You are a database migration planning agent. 

Goal: Migrate our company's production database from MySQL to 
PostgreSQL with zero data loss.

Before suggesting ANY migration steps, create a comprehensive plan:

1. ASSESSMENT PHASE
   - What information do I need about the current database?
   - What are the key risks?

2. PREPARATION PHASE  
   - What needs to happen before migration starts?
   - What tools and resources are needed?

3. MIGRATION PHASE
   - What is the step-by-step migration process?
   - What is the estimated downtime?

4. VALIDATION PHASE
   - How do we verify data integrity after migration?
   - What tests should we run?

5. ROLLBACK PLAN
   - If something goes wrong, what's the recovery process?

Output the plan first. Mark each step with priority (P1/P2/P3).
Wait for my approval before proceeding to execution.

Why it works: Forces the AI to plan across multiple phases, considers risks and rollback, prioritizes steps, and requires approval before execution.

🧪 Try It Yourself​

🧪 Try It Yourself

Edit the prompt and click Run to see the AI response.

🏋️ Practice Challenge​

Challenge

Write a planning prompt for an AI agent that needs to organize a company hackathon for 50 employees. The plan should include:

  • Timeline (4 weeks before the event)
  • Budget allocation
  • Team formation strategy
  • Judging criteria
  • Logistics (venue, food, prizes)

Make sure the AI creates the full plan before executing any steps.

🌍 Real-World Scenario​

Scenario: You're a startup CTO and need to take your product from MVP to production-ready.

A planning prompt would ask the AI to:

  1. Audit the current MVP architecture
  2. Identify technical debt and security gaps
  3. Prioritize improvements by impact and effort
  4. Create a 3-month roadmap with milestones
  5. Define success metrics for each milestone

The key is: plan first, then execute. This prevents costly mistakes from rushing to code without thinking through the full picture.

🎤 Interview Question​

Interview Question

Q: Why is planning important in AI agent design, and how do you prompt for it?

A: Planning prevents agents from acting prematurely on incomplete information. Without planning, agents may skip critical steps, waste resources, or produce inconsistent results. To prompt for planning, you explicitly instruct the AI to create a structured plan before taking any action. This includes breaking the goal into sub-goals, ordering steps logically, identifying dependencies, and defining success criteria. The "plan-then-execute" pattern — where the AI outputs a plan for human review before proceeding — is the most reliable approach.

📝 Summary​

Key Takeaways
  • Planning prompts force the AI to think strategically before acting
  • A good plan includes goals, steps, dependencies, success criteria, and contingencies
  • Use goal decomposition to break big problems into manageable sub-goals
  • The plan-then-execute pattern lets you review the plan before the AI acts
  • Always include a rollback or contingency section for critical tasks
  • Planning dramatically improves the quality and reliability of complex AI outputs