Topic Analysis: Chapter 5 - Tools and Workflows

Topic Analysis: Chapter 5 - Tools and Workflows

Metadata

  • Syllabus Reference: Part 2, Chapter 5
  • Primary Sources: Article 6 (Tools I Use)
  • Secondary Sources: Interview Q12-13, Article 2, Article 5
  • Analysis Date: 2025-11-28
  • Status: Complete

1. Source Materials

1.1 Primary Sources

From Article 6: "Tools I Use"

Tool division:

"Perplexity - roughly 90% of my searches (estimate) Zed + Claude - programming and technical things ChatGPT - non-technical things, household, garden"

Perplexity advantages/limitations:

"Advantages: Quick summary with source links, Can easily copy entire response with links, Helps with categorization Limitations: Often lacks latest data, Sometimes 'invents' sources that don't exist, Tends to be superficial on technical topics"

Perplexity workflow:

"1. Give short question with context (5-10 seconds) 2. Get summary + links 3. Verify dates and primary sources 4. Take only relevant parts (not everything!) 5. Insert as context into next tool"

Zed + Claude:

"Zed advantages: Excellent UX for context work, Can edit prompts during execution, Easy to add/remove files from context, Claude model is 'intelligent' - can find relevant files"

When Claude fails:

"Files with similar names in different directories, Too much context (more than 3-4 files), When it doesn't know which library version we use"

ChatGPT use cases:

"Best for: Writing and editing texts, Home projects (garden, recipes, repairs), Explaining things to kids, Brainstorming and ideation"

Tool combinations:

"Perplexity + ChatGPT: Research via Perplexity, Processing and writing via ChatGPT Perplexity + Claude: API documentation via Perplexity, Implementation via Claude What I never combine: ChatGPT + Claude (doesn't make sense, they have different uses)"

Cost breakdown:

"Perplexity Pro: Free via Revolut Premium package Zed Pro: Usage-based pricing, roughly $50-100/month ChatGPT Plus: $20/month Total: $70-120/month"

Tool selection advice:

"1. Start with one - not all at once 2. Give it a month - you need time to develop habits 3. Look for fit - not every tool suits everyone 4. Don't pay immediately - try free versions 5. Specialize - each tool for different thing"

1.2 Secondary Sources

From Interview Q12-13 (Tools)

"Perplexity on research, getting information (roughly 90% of searches) Zed+Claude for programming and technical things ChatGPT for non-technical things, home, garden, kids"

"Combinations: Claude+Zed with Perplexity less often, Perplexity+ChatGPT more often, never ChatGPT+Claude"

From Article 5: "AI Won't Steal Your Job"

"I use: Perplexity for research, getting information (roughly 90% of searches), Zed+Claude for programming and technical things, ChatGPT for non-technical things, home stuff, garden and for kids. Each tool has its purpose. It's not about which is 'best.' It's about which is best for a specific thing."

1.3 External Citations

None specific - focus is on practical tool selection.


2. Content Extraction

2.1 Key Concepts

  1. Tool Division by Task Type

    • Definition: Different tools for different contexts
    • Perplexity: Research, information gathering
    • Claude/Zed: Technical work, programming
    • ChatGPT: Non-technical, creative, home/family
    • Source: Article 6, Interview Q12
  2. Research Workflow

    • Definition: Perplexity → extract → next tool
    • Steps: Short question → summary + links → verify → extract relevant → use in Claude/ChatGPT
    • Source: Article 2, Article 6
  3. Coding Workflow

    • Definition: Zed + Claude with context management
    • Features: Edit prompts mid-work, add/remove files, AI finds relevant files
    • Source: Article 6
  4. Tool Combinations

    • Works: Perplexity + ChatGPT (research + writing)
    • Works: Perplexity + Claude (docs + implementation)
    • Never: ChatGPT + Claude (different domains)
    • Source: Article 6, Interview Q13
  5. Tool Selection Principles

    • Start with one, give it a month
    • Find what fits you
    • Free first, pay when hitting limits
    • Specialize each tool
    • Source: Article 6

2.2 Key Examples

  1. Perplexity Research Flow

    • Context: Any research task
    • Process: 5-10 second question → summary → extract relevant → next tool
    • Source: Article 6
  2. Zed + Claude for Bug Fix

    • Context: Linear task with file references
    • Process: Task description → Claude finds files → edit prompt if needed → fix
    • Source: Article 6
  3. ChatGPT for Home Projects

    • Context: Garden planting, electricity analysis
    • Examples: What to plant, how deep, when to water; solar panel ROI analysis
    • Source: Article 5, Article 6
  4. Tool Cost Analysis

    • Context: Monthly spending on AI tools
    • Breakdown: Perplexity (free via Revolut), Zed ($50-100), ChatGPT ($20)
    • Total: $70-120/month, saves hours weekly
    • Source: Article 6

2.3 Key Quotes

  1. "The answer isn't about which is 'the best' overall. It's about which is best for a specific thing." - Article 6

    • Use for: Opening principle
  2. "This division isn't random. It developed naturally based on what worked best for me where." - Article 6

    • Use for: Natural specialization
  3. "The best tool is one that: You use daily, You understand its limitations, You can effectively combine" - Article 6

    • Use for: Tool selection criteria
  4. "ROI is huge. I save hours weekly." - Article 6

    • Use for: Cost justification

2.4 Data/Statistics

  • 90% Perplexity vs 10% Google for searches
  • 5-10 seconds for initial Perplexity query
  • $70-120/month total AI tool cost
  • 3-4 files max context before Claude struggles

3. Gap Analysis

3.1 Content Gaps

  • [x] Tool division covered
  • [x] Perplexity workflow covered
  • [x] Zed + Claude covered
  • [x] ChatGPT uses covered
  • [x] Combinations covered
  • [x] Cost covered
  • [ ] Could add alternatives (Cursor, Copilot brief mention)

3.2 Clarity Issues

  • None - very practical chapter

3.3 Depth Assessment

  • Strong personal experience
  • Clear decision framework
  • Good cost transparency

4. Structure Proposal

4.1 Chapter Outline

Chapter 5: Tools and Workflows

Section 5.1: Choosing the Right Tool

  • Main point: Different tools for different contexts
  • Content from: Article 6
  • Include: Division by task type, not "best overall"

Section 5.2: Research Workflow

  • Main point: Perplexity → extract → next tool
  • Content from: Article 2, Article 6
  • Include: Step-by-step flow, what to extract

Section 5.3: Coding Workflow

  • Main point: Zed + Claude with context management
  • Content from: Article 6
  • Include: Prompt editing, file context, when it fails

Section 5.4: Tool Combinations

  • Main point: What works together, what doesn't
  • Content from: Article 6, Interview Q13
  • Include: Perplexity+ChatGPT, Perplexity+Claude, never ChatGPT+Claude

4.2 Opening Hook

"People ask me which AI tool is best. Wrong question. The right question is: which tool is best for what?"

4.3 Key Takeaways

  1. No single "best" tool - specialize each for different task types
  2. Perplexity for research → extract relevant → feed into working tool
  3. Claude/Zed for technical, ChatGPT for non-technical
  4. Start with one tool, give it a month, find your fit
  5. ROI: $70-120/month saves hours weekly

4.4 Transition

"Now that you have the tools, let's see them in action with real-world examples across different task types."


5. Writing Notes

5.1 Tone/Voice

  • Practical, personal experience
  • Opinion-based but justified
  • Cost-transparent

5.2 Audience Considerations

  • Useful for beginners choosing tools
  • Developers will relate to Zed/Claude
  • General audience to ChatGPT examples
  • Cost breakdown helps decision-making

5.3 Potential Visuals

  1. Tool Division Diagram

    • Perplexity = Research
    • Claude = Technical
    • ChatGPT = Non-technical
  2. Research Workflow Flow

    • Question → Summary → Extract → Next tool
  3. Tool Combinations Matrix

    • What works with what
  4. Cost Breakdown Table

    • Tool, cost, use case

6. Prepared Citations

Internal

  • [A2] Article "What Good Context Looks Like"
  • [A5] Article "AI Won't Steal Your Job"
  • [A6] Article "Tools I Use"
  • [I12] Interview Q&A, Question 12 (Tools)
  • [I13] Interview Q&A, Question 13 (Combinations)

External

  • None specific to this chapter

7. Open Questions

  1. Include alternatives like Cursor, Copilot?

    • Decision: Brief mention of why not used, focus on what is used
  2. Detailed settings/configurations?

    • Decision: Keep high-level, mention key settings only
  3. Include pricing links?

    • Decision: No, prices change - just note approximate ranges

Article Details

Category
context engineering new topic analysis
Published
November 28, 2025
Length
1,386 words
8,783 characters
~6 pages
Status
Draft Preview

More from context engineering new topic analysis

Topic Analysis: Chapter 1 - Why Your AI Prompts Fail

# Topic Analysis: Chapter 1 - Why Your AI Prompts Fail ## Metadata - Syllabus Reference: Part 1, Chapter 1 - Primary Sources: Article 1 (Prompt Is Not Enough), Article 9 (Vibe Coding) - Secondary Sou...

Read article

Topic Analysis: Chapter 2 - What Good Context Looks Like

# Topic Analysis: Chapter 2 - What Good Context Looks Like ## Metadata - Syllabus Reference: Part 1, Chapter 2 - Primary Sources: Article 2 (What Good Context Looks Like), Article 7 (Teams) - Seconda...

Read article