Transforms video content (screen recordings, demos, bug reports) into structured multimodal data — keyframes synced with audio transcription — for AI analysis.