Back to Blog
Release Notes
PDF Image Extraction
April 1, 2026
New Features
- 📄 PDF Image Extraction: Attach a PDF to any presentation and the AI will automatically extract all embedded images — logos, photos, diagrams, charts, and illustrations
- 🔍 Smart Classification: An AI sub-agent reviews every extracted image, classifies it by type (logo, photo, diagram, icon, etc.), and filters out backgrounds and decorative clutter
- ✨ Transparency Preserved: Extracted images keep their original alpha channel — transparent logos and icons come out exactly as they were in the PDF
- 🧹 Automatic Deduplication: Repeated images (slide backgrounds, recurring icons) are detected and kept only once — no duplicates in your asset library
How It Works
- Upload a PDF when creating a presentation or attach it in the chat — the AI detects PDF materials automatically
- The material extractor agent parses every page, pulls out embedded images, and uploads full-size originals plus small thumbnails to storage
- The AI sees the thumbnails, reviews each image, and classifies what's valuable — logos, team photos, partner logos, product shots
- Only the valuable assets are passed to the main AI agent, which uses them directly in your slides — with full-size quality
Why It Matters
- PDFs often contain unique assets that aren't available anywhere online — transparent company logos, internal team photos, proprietary diagrams
- Previously, you had to manually export images from your PDF and re-upload them — now it happens automatically in seconds
- Works with any PDF: pitch decks, brand guidelines, reports, marketing materials — anything with embedded images