Skip to main content

Overview

ThinkEx supports a variety of media types beyond plain text notes, allowing you to create a rich, multimedia knowledge workspace. Each media type is represented as a card on your canvas with specialized features and interactions.

PDF Support

Adding PDFs

Upload PDF documents to your workspace:
  1. Click the New button or right-click on canvas
  2. Select PDF/Image from the menu
  3. Choose one or more PDF files (or drag and drop)
  4. PDFs appear as cards on your canvas
ThinkEx supports multiple PDF uploads at once, making it easy to add entire document collections.

Automatic OCR Processing

When you upload a PDF, ThinkEx automatically:
  • Extracts text content using advanced OCR technology
  • Processes each page individually for accurate results
  • Preserves formatting including headers, footers, and structure
  • Detects tables and converts them to structured data
  • Identifies images within the PDF for later access
OCR processing happens in the background. Large PDFs may take a few moments to fully process, but you can start viewing immediately.

PDF Features

Page Navigation

Flip through pages using arrow buttons or page number input

AI Reading

The AI can read specific pages or page ranges from your PDFs

Text Search

Find content across all pages using workspace search

Active Page Context

The AI knows which page you’re viewing for contextual help

OCR Content Structure

Extracted PDF content includes:
  • Headers and footers - Preserved separately from main content
  • Body text - Main content converted to clean markdown
  • Tables - Structured data with rows and columns
  • Images - Placeholder references (accessible via AI tools)
  • Hyperlinks - Preserved links from original document

PDF Limitations

Password-protected PDFs cannot be uploaded. Please remove password protection before uploading.
File size limits:
  • Maximum file size varies by plan
  • Consider splitting very large documents
OCR accuracy:
  • Best with clear, typed text
  • Handwritten notes may have reduced accuracy
  • Complex layouts may require manual review

YouTube Support

Adding YouTube Videos

Embed YouTube videos directly in your workspace:
  1. Click NewYouTube
  2. Paste any YouTube video URL
  3. Video appears as an interactive card
YouTube playlists are also supported! Paste a playlist URL to access the entire series.

Video Features

  • Embedded player - Watch videos without leaving ThinkEx
  • Progress tracking - Resume where you left off
  • Thumbnail preview - See video thumbnail when not playing
  • Title and metadata - Automatic video information retrieval

Video Layout

YouTube cards:
  • Maintain 16:9 aspect ratio - Proper video display
  • Minimum 200px height - Required by YouTube IFrame API
  • Expandable - Resize for focused viewing
  • Compact mode - Shows thumbnail when minimized

YouTube Limitations

YouTube features depend on video availability and YouTube’s API. Some videos may have embedding restrictions.
Current constraints:
  • Requires internet connection for playback
  • Video analysis uses AI tokens (direct video understanding via Gemini)
  • Video length affects processing time

Image Support

Adding Images

Upload images to your workspace:
  1. Use the PDF/Image upload dialog
  2. Select image files (JPG, PNG, GIF, etc.)
  3. Images appear as visual cards

Image Features

  • Full image display - High-quality rendering on canvas
  • Alt text - Accessibility and context information
  • Captions - Add descriptive text to images
  • AI vision - The AI can analyze image content when selected
Images from PDFs can be extracted and analyzed by the AI using the processFiles tool with PDF image references.

Audio Support

Audio Recordings

ThinkEx includes audio recording and transcription:
  • Voice recordings - Record directly in the app
  • Audio uploads - Import existing audio files
  • Automatic transcription - Speech-to-text processing
  • Speaker detection - Identifies different speakers
  • Timestamped segments - Navigate by timestamp

Audio Features

Transcription

Automatic speech-to-text conversion with high accuracy

Speaker Labels

Identifies and labels different speakers in recordings

AI Summary

Generates summary of audio content automatically

Timestamps

Navigate to specific moments with time-coded segments

Audio Processing

When you upload audio, ThinkEx:
  1. Uploads to storage - Securely stores the audio file
  2. Processes audio - Extracts speech and identifies speakers
  3. Generates transcript - Creates full text version
  4. Summarizes content - AI creates concise summary
  5. Detects emotion - Optional sentiment analysis per segment
Audio processing happens asynchronously. You’ll see processing status updates as the system works through your audio.

Audio Data Structure

Transcripts include:
  • Speaker identification - Who said what
  • Timestamps - When each segment occurred
  • Full text - Complete transcript
  • Language detection - Identifies spoken language
  • Translation (optional) - Translates non-English segments
  • Emotion tags - Sentiment for each speaker segment

Other Card Types

Notes

Rich text notes with markdown support:
  • Block-based editor - Flexible content blocks
  • Math support - LaTeX equations inline and display mode
  • Images - Embed images directly in notes
  • Lists and tables - Structured content

Flashcards

Study tools with spaced repetition:
  • Multiple cards per deck - Organize related concepts
  • Front and back - Question and answer format
  • Rich content - Images, math, formatting on both sides
  • AI generation - Create from selected content

Quizzes

Interactive assessments:
  • Multiple choice - 4 options per question
  • True/False - Binary questions
  • Explanations - Learn why answers are correct
  • Progress tracking - See which questions you’ve answered
  • Score calculation - Automatic grading

Media Best Practices

File Organization

1

Use descriptive names

Rename files to reflect content (e.g., “Chapter 3 - Supply Chain” not “document1.pdf”)
2

Organize with folders

Group related media files together in folders
3

Add context

Use subtitles and descriptions to add context

Working with Large Files

  • Upload during off-hours - Better performance with large files
  • Use specific pages - Reference specific PDF pages for AI assistance
  • Split if needed - Break very large PDFs into chapters
  • Monitor processing - Wait for OCR completion before intensive AI work

Accessibility

  • Add alt text to images for screen readers
  • Include captions for context and searchability
  • Use transcripts for audio and video content
  • Describe visuals in notes when sharing with others

Media and AI Context

How AI Accesses Media

When media is selected as context:
  • PDFs - AI can read specific pages or search across document
  • Images - AI can analyze visual content (vision capability)
  • Audio - AI reads transcript and speaker segments
  • Videos - AI can search web for video content/transcript
For PDFs, tell the AI which pages to read (e.g., “read pages 10-15”) for faster, more focused responses.

Processing Requirements

  • PDFs must complete OCR before full AI access
  • Audio must be transcribed before AI can read it
  • Images work immediately with AI vision
  • Videos link to external content (no transcript yet)

Troubleshooting

PDF Issues

Problem: PDF upload fails
  • Check file size limits
  • Ensure PDF isn’t password-protected
  • Try re-exporting PDF from source
Problem: OCR isn’t working
  • Wait for background processing to complete
  • Check OCR status indicator on card
  • Try manually triggering processFiles tool

Video Issues

Problem: YouTube video won’t play
  • Verify video isn’t age-restricted
  • Check if video allows embedding
  • Ensure stable internet connection

Audio Issues

Problem: Transcription is inaccurate
  • Check audio quality (clear recording helps)
  • Ensure proper language detection
  • Review and edit transcript manually if needed

Next Steps

Visual Canvas

Learn how to arrange media cards on your canvas

AI Context

Discover how to give AI access to your media files

Build docs developers (and LLMs) love