slice icon Context Slice

Slide Deck Conversion

Supported Input

  • PDF files exported from any presentation software (PowerPoint, Google Slides, Keynote, etc.)
  • Best results with standard slide dimensions (16:9 or 4:3)
  • File size limit: ~20MB recommended (larger files may timeout)

Output Format

The conversion produces structured markdown with:

  1. Document title - extracted from title slide or filename
  2. Per-slide sections - numbered with titles
  3. Text content - all bullets, paragraphs, captions preserved verbatim
  4. Visual descriptions - semantic explanations of charts, diagrams, images
  5. Layout notes - only when layout conveys meaning (comparisons, timelines)

Visual Element Handling

Charts and Graphs

Described by what they SHOW, not what they ARE:

  • Good: "Revenue grew 35% year-over-year, from $3.1M to $4.2M"
  • Bad: "Bar chart with two bars"

Diagrams and Flowcharts

Explained by relationships and flow:

  • Good: "User requests flow through the API gateway, which routes to either the auth service or the main application based on endpoint"
  • Bad: "Flowchart with boxes and arrows"

Images and Photos

Described by content and relevance:

  • Good: "Team photo showing the engineering department at the annual offsite"
  • Bad: "Photo of people"

Tables

Reproduced as markdown tables when structured data is clear.

Limitations

  • Animations: Static PDF loses animation sequence context
  • Handwriting: Hand-drawn elements may not be recognized accurately
  • Very large decks: 100+ slides may need to be split
  • Embedded videos: Only thumbnail/placeholder visible

Presenting Results

After conversion:

  1. Report the number of slides processed
  2. Mention key topics or sections identified
  3. Offer to show specific slides or the full document
  4. Note if any slides appeared problematic (mostly images, complex layouts)