Concept · 2025

Video Editor AI

An editing assistant for travel vlogs. It transcribes raw footage, surfaces audio-led moments worth keeping, and produces a clip-by-clip brief that tells the editor where natural audio beats voice-over.

ConceptAI agentsVideoTranscription

The problem

Editing travel footage is slow.

The bottleneck isn’t the cut — it’s finding which clips are worth keeping. Hours of footage hides minutes of moments that actually carry the story.

The approach

Find the audio-led moments. Hand the editor a brief.

  • Transcribe everything. Every clip is transcribed with timestamps so the editor can search the trip the way they’d search a document.
  • Surface what matters. Banter, crowd noise, reactions, and other audio-led moments get flagged with timestamp ranges and a one-line reason.
  • Produce a brief. Final output is a clip-by-clip editing brief labelling each clip as Use / Optional / B-roll / Skip — with explicit guidance on where natural audio should stay and where voice-over fits.

Status

Concept stage. In active development. No public demo yet.

Stack

  • Python
  • Whisper
  • LLM orchestration
  • Final Cut Pro pipeline
  • FFmpeg
  • PDF brief output