Web App
Getting creators accurate chapters without scrubbing their own timeline.
Role
Product Design
UI Design
Full-Stack Development
Code
Problem
Chapters help viewers jump to what they came for and help videos get found. Making them is tedious: scrub your own video, note timestamps by hand, write a title for each one. YouTube generates chapters automatically, but creators can't rely on the result; the titles are often generic or wrong, and there is no way to correct them. So most creators face a bad choice: spend the time, or ship chapters that don't represent their video.
Solution
Upload the audio and get back a timestamped transcript and a set of draft chapters. Review them, retime them, retitle them, then copy the finished list in exactly the format YouTube expects. The AI does the tedious part; the creator makes every call that reaches the audience.
Key decisions
Nothing publishes without review. Generated chapters are a draft, not an output. The editor shows the transcript, timestamps, and titles side by side, because a creator will only trust chapters they were able to change.
Meet the creator where the work happens. The end of the workflow is YouTube's description box, so the export is one click and already formatted. No account, no library, no place the chapters live except where they are needed.
Understand in any language, publish in one. The tool transcribes eighteen languages but writes chapters in English, a deliberate call for reach on a platform where English titles travel furthest.
Outcome
Built end to end and released as open source, so any creator can run it with their own accounts. It was my first full pass at the loop I now default to: find the tedious step, let AI draft it, and keep the human in charge of what ships.