/powerpoint-automation

Common Use Cases

Detailed examples of how to use PowerPoint Automation for various scenarios.

1. Tech Blog → Presentation

Scenario: Convert a Zenn or Qiita technical article into a presentation

Input:

• URL: https://zenn.dev/example/articles/azure-copilot-guide
• Article length: 3000 words, 5 images, 3 code blocks

User Request:

"Create a 15-slide presentation about Azure Copilot from this Zenn article:
https://zenn.dev/example/articles/azure-copilot-guide"

Workflow:

TRIAGE → User provides URL
    ↓
PLAN → System proposes: 15 slides, template D, standard detail
    ↓
EXTRACT → Fetch article content + download images
    ↓
TRANSLATE → (Skip if Japanese source)
    ↓
BUILD → create_from_template.py
    ↓
REVIEW → PPTX Reviewer checks quality
    ↓
DONE → Open PowerPoint

Output:

• 15-slide PPTX with:
  • Title slide
  • Agenda (3 sections)
  • 10 content slides (with code blocks)
  • 3 image slides (screenshots)
  • Summary slide

Time: ~3-5 minutes

2. English PPTX → Japanese Translation

Scenario: Translate Microsoft Ignite session slides to Japanese

Input:

• File: input/BRK252_DataSecurity.pptx (120 slides, English)
• Requirement: Maintain original design and layout

User Request:

"Translate this Microsoft presentation to Japanese, keeping the same design"

Workflow:

TRIAGE → Detect PPTX input
    ↓
PLAN → User confirms: Full translation (120 slides), method A (template inheritance)
    ↓
PREPARE_TEMPLATE → Diagnose + clean template
    ↓
EXTRACT → reconstruct_analyzer.py (PPTX → content.json)
         → extract_images.py (save all images)
    ↓
TRANSLATE → Localizer agent (content.json → content_ja.json)
    ↓
BUILD → create_from_template.py (inherit master from original)
    ↓
REVIEW → JSON Reviewer (translation quality)
       → PPTX Reviewer (layout, overflow, notes)
    ↓
DONE → 120-slide Japanese PPTX

Output:

• Japanese PPTX maintaining:
  • Original colors and fonts
  • Slide master design
  • Image positions
  • Speaker notes (translated)

Key Features:

• Template inheritance (method A)
• Automatic layout selection
• AutoFit control (prevent text spacing issues)
• Position adjustment for Japanese text

Time: ~15-20 minutes for 120 slides

3. Report Generation from Notes

Scenario: Create quarterly report presentation from Markdown notes

Input:

• Markdown notes with bullet points
• 5 charts/graphs (PNG images)
• Goal: Formal business report

User Request:

"Create a quarterly report presentation from my notes.
Include these 5 charts and make it professional."

Workflow:

TRIAGE → Detect mixed input (text + images)
    ↓
BRAINSTORM → Interactive Q&A
           → "Who's the audience?" → "Executives"
           → "Duration?" → "30 minutes"
           → "Tone?" → "Formal"
    ↓
        proposal.json generated
    ↓
PLAN → User confirms: 25 slides, template (business style)
    ↓
EXTRACT → Convert notes to content.json
        → Map images to slides
    ↓
BUILD → create_from_template.py
    ↓
REVIEW → PPTX Reviewer
    ↓
DONE → Business report PPTX

Output:

• 25-slide report with:
  • Executive summary
  • 5 sections with charts
  • Key metrics and insights
  • Action items and next steps

Brainstormer Interaction: The Brainstormer agent asks:

1. Who's the audience?
2. Presentation duration?
3. Key message/goal?
4. Tone (formal/casual)?
5. Any topics to avoid?

Time: ~10 minutes (including brainstorming)

4. Technical Documentation → Training Material

Scenario: Convert API documentation into training slides

Input:

• API reference (Markdown/HTML)
• Code examples (JSON, cURL)
• Architecture diagram

User Request:

"Create training material for our REST API.
Include authentication, endpoints, and code examples."

Workflow:

TRIAGE → Detect documentation input
    ↓
PLAN → User confirms: 20 slides, method B (pptxgenjs for code blocks)
    ↓
EXTRACT → Parse documentation
        → Extract code snippets
        → Convert architecture diagram
    ↓
BUILD → create_pptx.js (pptxgenjs)
    ↓
REVIEW → PPTX Reviewer (code readability)
    ↓
DONE → Training material PPTX

Output:

• 20-slide training with:
  • API overview
  • Authentication flow (diagram)
  • Endpoint reference (with code)
  • Error handling examples
  • Best practices

Why pptxgenjs (Method B)?

• Better code block formatting (monospace, syntax highlighting)
• Custom diagram positioning
• Fine-grained control over layout

Time: ~8-10 minutes

5. Architecture Diagram + Presentation

Scenario: Add custom architecture diagrams to existing template

Input:

• Existing presentation template
• Architecture components (boxes, arrows, labels)

User Request:

"Create an Azure architecture diagram and insert it into slide 5 of this template"

Workflow:

PLAN → User confirms: Insert diagram at slide 5
    ↓
DIAGRAM → create_pptx.js (generate diagram with shapes)
        → Output: diagrams.pptx
    ↓
MERGE → insert_diagram_slides.py
      → Insert diagrams.pptx[0] at position 5
      → Select layout: "Title and Content"
    ↓
DONE → Combined PPTX

Diagram Creation (pptxgenjs):

// Azure components
slide.addShape(pptx.ShapeType.rect, {
  x: 1,
  y: 2,
  w: 2,
  h: 1,
  fill: "0078D4", // Azure blue
  line: { color: "FFFFFF" },
});

// Arrows
slide.addShape(pptx.ShapeType.rightArrow, {
  x: 3.5,
  y: 2.3,
  w: 1,
  h: 0.4,
  fill: "505050",
});

Time: ~5 minutes

6. Slide Count Reduction (Summarization)

Scenario: Reduce 60-slide presentation to 20 slides for shorter session

Input:

• Original PPTX (60 slides)
• Target: 20 slides (1/3 compression)

User Request:

"Reduce this 60-slide presentation to 20 slides for a 15-minute talk"

Workflow:

EXTRACT → reconstruct_analyzer.py
    ↓
SUMMARIZE → summarize_content.py
          → AI combines related slides
          → Keeps key visuals
          → Merges bullet points
    ↓
BUILD → create_from_template.py
    ↓
REVIEW → Check for information loss
    ↓
DONE → 20-slide condensed version

Summarization Strategy:

• Merge similar content slides
• Keep all section dividers
• Preserve key images
• Combine related bullet points
• Maintain logical flow

Time: ~10 minutes

7. Multilingual Presentations

Scenario: Create same presentation in 3 languages (EN, JA, ZH)

Input:

• Original content.json (English)

User Request:

"Create this presentation in English, Japanese, and Chinese"

Workflow:

EXTRACT → content.json (English)
    ↓
TRANSLATE → Localizer (EN → JA) → content_ja.json
    ↓
TRANSLATE → Localizer (EN → ZH) → content_zh.json
    ↓
BUILD → create_from_template.py × 3
      → output_ppt/{base}_en.pptx
      → output_ppt/{base}_ja.pptx
      → output_ppt/{base}_zh.pptx
    ↓
DONE → 3 language versions

Font Handling:

• English: Arial, Calibri
• Japanese: Meiryo, Yu Gothic
• Chinese: Microsoft YaHei, SimHei

Time: ~20 minutes for 3 languages

Quick Reference

Tips & Best Practices

For Blog Conversions

• Prefer standard detail (option 2) for balance
• Always include images from the article
• Use method D (first template) for professional look

For Translations

• Use method A (template inheritance) to maintain design
• Review speaker notes carefully
• Check for technical term consistency

For Technical Content

• Use pptxgenjs (method B) for code-heavy slides
• Limit code blocks to 10 lines per slide
• Include comments in code for clarity

For Business Reports

• Use formal templates
• Include agenda and summary slides
• Add speaker notes for presenter guidance
• Keep bullet points to 5-8 items max

For Diagrams

• Use pptxgenjs for custom shapes
• Match template colors (read from master)
• Position diagrams at "right" or "bottom" for text space