Applied AI / Personal build
KLIPPY: long video to usable clips.
A content creator needed captioned clips from longer videos. I built a tool around that specific workflow with Claude Code and Codex.
Build note based on my published project description. No public app signup is currently offered here.
The job to be done
Take a longer video, find a useful excerpt, edit it, and export a captioned clip. Keep the selection and editing process understandable to the person making the content.
The workflow I built
- 01 / TranscribeUpload a video and create a transcript with Whisper.
- 02 / Select & editReview clip suggestions or select transcript text to create a clip.
- 03 / ExportAdd captions and export the selected video.
The feature set I described also includes deleting video through its transcript and removing filler words.
What this project demonstrates
A useful AI project starts with a task specific enough to finish. For this build, transcription and editing belong in one workflow, with the creator deciding what makes a good clip.
Claude Code and Codex helped me build the application. Whisper handles transcription in the workflow. Those are two different uses of AI, and both should be evaluated against the work they actually improve.
What Iād measure in a production rollout
- Time from uploading a video to an approved clip.
- Transcript and caption corrections required.
- How often a suggested clip is useful without reselection.
- Export reliability across the video formats people use.
These are evaluation criteria, not measured client results.
Working through something similar?
Tell me about the process, the tools, and the part that keeps getting in the way.
Talk Through Your Setup