🎯 Lesson Objectives

By the end of this lesson, you will:

Understand that first prompts rarely give perfect results
Master the iteration loop for refining AI output
Know effective follow-up phrases that improve results
Avoid common iteration anti-patterns

📖 The Reality of Prompting

Here's a truth that might be comforting:

Prompting Reality

Your first prompt will rarely give you exactly what you want.

And that's completely okay! Professional prompt engineers expect to iterate.

Why First Prompts Often Miss:

You might not know exactly what you want until you see it
AI might interpret your words differently than intended
Complex requirements are hard to fully capture upfront
You'll discover new requirements as you go

The Goal:

Not to write the perfect first prompt, but to refine efficiently through conversation.

🔄 The Iteration Loop

Think of prompting as a conversation, not a single command:

┌─────────────────┐
│  1. PROMPT      │ ← Your initial request
└────────┬────────┘
         ↓
┌─────────────────┐
│  2. REVIEW      │ ← Look at AI's response
└────────┬────────┘
         ↓
┌─────────────────┐
│  3. IDENTIFY    │ ← What's missing or wrong?
└────────┬────────┘
         ↓
┌─────────────────┐
│  4. REFINE      │ ← Ask for specific changes
└────────┬────────┘
         │
         ↓ (repeat until satisfied)
┌─────────────────┐
│  5. DONE        │ ← Result matches your needs
└─────────────────┘

🔘 A Quick Example: Button Iteration

Step	Your Action	Result
Initial	"Create a button"	Basic gray button
Review	"Hmm, I wanted blue and bigger"	-
Refine	"Make it blue with white text and 20px padding"	Blue button, better
Review	"Nice, but needs hover effect"	-
Refine	"Add hover effect that darkens slightly"	Blue button with hover
Done!	3 prompts, 2 minutes	Exactly what you wanted

💬 Powerful Follow-Up Phrases

Here are proven phrases that guide AI to better results:

For Simplification:

Text

"Make it simpler"
"Reduce complexity"
"Write a more basic version"
"Can you simplify this?"

For Addition:

Text

"Add [feature]"
"Include [element]"
"Also make it [quality]"
"Can you add validation for..."

For Removal:

Text

"Remove [element]"
"Don't include [feature]"
"Simpler, without [thing]"
"Take out the [component]"

For Clarification:

Text

"Explain why you did [thing]"
"What does this part do?"
"Walk me through the logic"
"Why this approach?"

For Alternatives:

Text

"What's another way to do this?"
"Show me a different approach"
"What if we used [alternative]?"
"Any other options?"

For Style:

Text

"Make it look more [adjective]"
"Use [color/style] instead"
"Match this design: [description]"
"Make it mobile-friendly"

🎮 Full Example: Building a Countdown Timer

Let's see a complete iteration session:

Prompt 1 (Initial):

Create a countdown timer

AI Creates: Basic countdown in Python (console-based) Problem: I wanted web-based!

Prompt 2 (Refine):

Make it a web-based countdown timer using HTML, CSS, and JavaScript

AI Creates: Web countdown with basic styling Problem: No way to set the time!

Prompt 3 (Add Feature):

Add an input field where users can enter minutes to count down

AI Creates: Timer with input field Problem: Looks plain

Prompt 4 (Style):

Make the design modern with dark background, large numbers, nice start button

AI Creates: Styled timer with dark theme Problem: Should make sound when done!

Prompt 5 (Edge Case):

Add a beep sound and visual flash when timer reaches zero

AI Creates: Timer with audio alert Review: Perfect! ✅

Total: 5 prompts → Polished, fully-featured timer!

❌ Anti-Patterns to Avoid

These common mistakes slow down iteration:

Anti-Pattern 1: Starting Over for Every Change

❌ Wrong Way

✅ Right Way

'Create timer' → 'Create timer with input' → 'Create timer with input and dark theme'

'Create timer' → 'Add input field' → 'Make it dark themed'

Each time you start fresh, losing context

Build on previous responses

Fix: Keep the conversation going. AI remembers context.

Anti-Pattern 2: Accepting First Result Without Review

❌ Wrong

✅ Right

AI output → Immediately use → Find problems later

AI output → Review carefully → Refine → Confident

Fix: Always spend 30 seconds reviewing before accepting.

Anti-Pattern 3: Vague Feedback

❌ Vague

✅ Specific

'Make it better'

'Make the button larger and add more padding'

'This isn't right'

'The calculation should subtract, not add'

'Fix it'

'Fix the hover state — it should turn blue'

Fix: Be specific about what's wrong and what you want instead.

Anti-Pattern 4: Changing Too Many Things at Once

❌ Too Much

✅ Manageable

'Make it bigger, change color, add animation, fix spacing, add shadow, change font, make responsive'

'Make it bigger and add dark mode' → Then: 'Add animation and fix spacing' → Then: 'Make responsive'

Fix: Change 1-3 things at a time. Easier to verify each change worked.

🌳 Iteration Decision Tree

Use this when deciding your next step:

Does the output match what you wanted?
│
├── YES → ✅ Done! Accept the result
│
└── NO → What's the main issue?
         │
         ├── Wrong technology → "Use [correct tech] instead"
         │
         ├── Missing feature → "Add [feature]"
         │
         ├── Unwanted elements → "Remove [element]"
         │
         ├── Wrong style → "Make it [description]"
         │
         ├── Doesn't work → "This gives error: [error]. Fix it"
         │
         └── Don't understand → "Explain [confusing part]"

🧪 Practice Exercise

Try this iteration exercise:

Start with:

Start Here

Create a simple card component

ChatGPT Claude

Then iterate to add (one or two at a time):

An image at the top
A title below the image
A short description
A "Learn More" button
A shadow effect
Rounded corners
A hover animation

Challenge: Achieve this in as few prompts as possible!

📋 The Iteration Checklist

✅ Before accepting any AI output:

0/6

✅ Lesson Summary

Key Takeaways

First prompts rarely perfect — Expect to iterate
Keep the conversation going — Build on previous responses
Be specific in feedback — Tell AI exactly what to change
One to three changes at a time — Easier to verify

The Iteration Loop:

Prompt → Review → Identify Issues → Refine → Repeat

📝 Mini Quiz

📝 Check Your Understanding

1/3

How many prompts should you expect for a complex task?