Topic Analysis: Chapter 3 - Before You Prompt - Preparation

Metadata

• Syllabus Reference: Part 2, Chapter 3
• Primary Sources: Articles 1-4, Presentation
• Secondary Sources: Interview Q1-5
• Analysis Date: 2025-11-28
• Status: Complete

---

1. Source Materials

1.1 Primary Sources

From Article 1: "Prompt Is Not Enough"

Car seat example with WHY:

"I know ADAC tests are relevant... For me, safety is a higher priority than having to take the seat out of the car once a year. So I gave Perplexity this request: 'My son is 120cm tall. Get safety ratings from ADAC tests (not overall ratings!) and create a table. Exclude marketplace sellers (Amazon, etc.)'"

From Article 3: "Why You Give Much and Get Little"

Breaking down tasks:

"Good Task (broken down):

1. Create HTML table with 3 columns: amount, category, date
2. Add validation - amount must be number
3. Store data in localStorage
4. Add basic CSS styling
5. Create delete button for each row"

Junior test:

"Before every prompt, ask yourself: 'Could a junior developer/colleague who doesn't know my project fulfill this task?' If not, you're missing context."

Define success:

"'Done = test passes without errors' - Define success"

From Article 4: "Think Like an Engineer"

Atomization principle:

"When I break a big problem into smaller parts: For each task, I can quickly provide the right context (e.g., the right code file). The task description is shorter and clearer. AI has an extremely high chance of correct results on first iteration."

Practical workflow:

"1. First analysis (read-only): 'Analyze this task and break it into steps' 2. Then atomic tasks: 'Do only step #1' - I add context for that step 3. Record progress: I write analysis to markdown file 4. Every 8-12 interactions: Ask for SUMMARY + 3 TODO"

From Presentation: 10 Advices

1. Give the WHY, not just WHAT
2. Break into atomic parts
3. One example beats 1000 words
4. Say what to EXCLUDE
5. Define what DONE looks like

1.2 Secondary Sources

From Interview Q1-2 (Research Process)

• Always give reason for what you want
• AI can advise (e.g., car seat category by height)
• Iterate and specify based on initial results

From Interview Q5 (Task Breakdown)

• Depends on content complexity
• Look at features, data, context scope
• If information is massive, divide into smaller tasks

1.3 External Citations

None specific - focus is on practical preparation principles.

---

2. Content Extraction

2.1 Key Concepts

1. Give the WHY, Not Just WHAT

   • Definition: AI doesn't know your priorities without explanation
   • Example: Car seat - safety > convenience, so sort by safety not overall
   • Source: Article 1, Presentation

2. Break Into Atomic Parts

   • Definition: Smaller tasks = better results
   • Example: Expense tracker broken into 5 steps vs one big request
   • Source: Article 3, Article 4

3. One Example Beats 1000 Words

   • Definition: Show AI what you want instead of describing
   • Example: "Here are my previous 2 articles as style examples"
   • Source: Article 1, Presentation

4. Say What to EXCLUDE

   • Definition: AI includes everything unless told not to
   • Example: "Exclude marketplace sellers (Amazon, etc.)"
   • Source: Article 1, Article 8

5. Define What DONE Looks Like

   • Definition: Clear success criteria make verification possible
   • Example: "Done = test passes without errors"
   • Source: Article 3, Presentation

6. The Junior Developer Test

   • Definition: If junior can't complete it, AI can't either
   • Application: Before every prompt, ask if new colleague could do it
   • Source: Article 3

2.2 Key Examples

1. Car Seat with WHY

   • Context: Finding right car seat
   • Without WHY: Generic list by overall rating
   • With WHY: Sorted by safety specifically because that's priority
   • Source: Article 1

2. Expense Tracker Breakdown

   • Context: Building simple app
   • Before: "Build expense tracker" (500 lines garbage)
   • After: 5 atomic steps, each clear
   • Source: Article 3

3. Article Writing with Examples

   • Context: LinkedIn article request
   • Key addition: "Here are my previous 2 articles as style examples"
   • Impact: AI matches your voice/style
   • Source: Article 1

4. Exclude Specification

   • Context: Any search/selection task
   • Example: "Exclude Amazon products"
   • Why it matters: AI will include everything by default
   • Source: Article 1, Article 8

2.3 Key Quotes

1. "If you can't explain it to a junior in a minute, it's too big" - Article 3

   • Use for: Task size heuristic

2. "Each step has a clear goal. AI knows exactly what you want. You get exactly that." - Article 3

   • Use for: Atomization benefit

3. "Giving AI a large task often takes more time than breaking it into small steps" - Article 4

   • Use for: Counter-intuitive truth

4. "AI includes everything unless told not to" - Derived from examples

   • Use for: Exclusion principle

2.4 Data/Statistics

• Break down complex tasks: ~5 atomic steps typical
• Junior test: "If not explainable in 1 minute, too big"
• Every 8-12 interactions: Summary checkpoint

---

3. Gap Analysis

3.1 Content Gaps

• [x] WHY principle covered
• [x] Atomization covered
• [x] Examples principle covered
• [x] Exclusion principle covered
• [x] Success criteria covered
• [ ] Could add more non-technical examples

3.2 Clarity Issues

• None - principles are clear and practical

3.3 Depth Assessment

• Strong practical guidance
• Good concrete examples
• Principles applicable across domains

---

4. Structure Proposal

4.1 Chapter Outline

Chapter 3: Before You Prompt - Preparation

Section 3.1: Give the WHY, Not Just WHAT

• Main point: AI doesn't know your priorities without explanation
• Content from: Article 1, Interview Q1-2
• Include: Car seat example, Perplexity category selection

Section 3.2: Break Into Atomic Parts

• Main point: Smaller tasks = dramatically better results
• Content from: Article 3, Article 4, Interview Q5
• Include: Expense tracker breakdown, atomization principle

Section 3.3: One Example Beats 1000 Words

• Main point: Show AI what you want
• Content from: Article 1, Article 2
• Include: Style examples, format examples, attach previous work

Section 3.4: Say What to EXCLUDE

• Main point: AI includes everything unless told otherwise
• Content from: Article 1, Article 8
• Include: Amazon exclusion, specific constraints

Section 3.5: Define What DONE Looks Like

• Main point: Clear success criteria enable verification
• Content from: Article 3
• Include: "Test passes" example, verifiable outcomes

4.2 Opening Hook

"You wouldn't hand a new colleague a sticky note saying 'fix the app' and expect perfect results. Yet that's exactly how most people approach AI."

4.3 Key Takeaways

1. Always explain WHY you need something - AI can then optimize for your actual priority
2. Break big tasks into atomic steps - each step gets precise context
3. Show examples instead of describing - one example beats 1000 words
4. Explicitly exclude what you don't want - AI defaults to including everything
5. Define verifiable success criteria - "done = X" enables clear evaluation

4.4 Transition

"Now that you know how to prepare before prompting, let's look at what happens during and after - how to iterate effectively when AI doesn't get it right."

---

5. Writing Notes

5.1 Tone/Voice

• Practical, actionable
• Clear principles with examples
• Before/after comparisons

5.2 Audience Considerations

• 5 clear principles anyone can follow
• Technical and non-technical examples for each
• Checklist format for quick reference

5.3 Potential Visuals

1. Five Principles Checklist

   • WHY, Atomic, Example, Exclude, Done

2. Task Breakdown Diagram

   • One big task → 5 atomic parts

3. Before/After Task Examples

   • Side-by-side for each principle

---

6. Prepared Citations

Internal

• [A1] Article "Prompt Is Not Enough"
• [A3] Article "Why You Give Much and Get Little"
• [A4] Article "Think Like an Engineer"
• [I1-2] Interview Q&A, Questions 1-2 (Research process)
• [I5] Interview Q&A, Question 5 (Task breakdown)
• [P] Presentation - 10 Advices

External

• None specific to this chapter

---

7. Open Questions

1. Should 5 principles be numbered or treated as checklist?

   • Decision: Both - numbered for structure, checklist for reference

2. How much overlap with Chapter 2 templates?

   • Decision: Chapter 2 = what context looks like, Chapter 3 = how to prepare it

3. Include workflow diagram?

   • Decision: Yes, simple visual of preparation steps