All Projects

Niggle.ai

Niggle.ai: Active Study Materials From Any Content Source

Niggle.ai Dashboard

Reading a document and understanding it well enough to recall under exam conditions are two different things. The gap between them is practice: flashcards, self-testing, active recall. Most students do not do enough of it because creating quality study materials is itself a manual, time-consuming task.

Niggle.ai was built to remove that barrier. Feed it a document, PDF, or YouTube video and it generates the study materials. The student uses them instead of building them.

What We Built

AI Content Processing

  • Document and PDF ingestion with structured content extraction
  • YouTube video processing: transcription followed by content analysis
  • Key concept identification tuned for academic content patterns
  • Source preservation: every generated item traced back to the source material

Study Material Generation

  • Automatic summarization with configurable depth, from overview to detailed
  • Note generation organized by topic and subtopic, not just sequential paragraph extraction
  • Flashcard creation with question-answer pairs derived from core concepts
  • Quiz generation with multiple choice, short answer, and fill-in-the-blank formats
  • Difficulty calibration based on content complexity

Learning Workflow

  • Unified interface for all content types: documents, video, and web pages
  • Spaced repetition scheduling for flashcard review
  • Progress tracking by topic and session
  • Export to standard formats for use in existing study tools

The Technical Approach

The quality of generated study materials depends heavily on how content is chunked and analyzed before generation. We invested significantly in the content processing layer: semantic segmentation that preserved conceptual units rather than splitting on arbitrary lengths, which produced flashcards and quiz questions that were actually pedagogically useful rather than syntactically correct but contextually odd.

For YouTube video content, we built a two-stage pipeline: transcription with speaker diarization, followed by topic segmentation that aligned the generated materials with the video structure so students could reference the source.

My Role

  • Led product design and development from ideation to production
  • Designed the content ingestion interface and study material generation workflow
  • Built the AI processing pipeline for documents and video content
  • Architected the spaced repetition and progress tracking system
  • Managed deployment and performance optimization