AutoClip
PROJECT / 08Turn long footage into short clips worth reviewing.
Three things to know
- My responsibility
- Product design & AI-assisted development
- Key decision
- Surface candidate clips before a person decides what to cut.
- Delivery scope
- Implemented as a local tool and exploration of video workflows.
A pipeline connecting transcription, candidate selection, human review and exports in different aspect ratios, exploring how to find useful short clips in longer material.
Find candidates before making the cut.
Local Whisper transcription feeds a staged AI selection process. Coherence, the opening and platform format inform suggestions, with human confirmation before output.
Scene changes, audio levels and brightness help inform selection. FFmpeg handles cuts, captions and aspect ratios. Scores support editorial choices; they do not predict reach.
Manage the job as well as the video.
Long jobs need progress, durable records and retry paths. A separate worker, SQLite and WebSocket updates keep processing independent of an open browser page.
AutoClip focuses on automated selection pipelines. EazyCut focuses on subtitle correction and an editable workspace. They explore different parts of the video workflow.
A local product exploration.
The main processing pipeline has implementation and test records. Transcription and media processing run locally, while AI analysis uses an external model service. Mobile apps and social publishing are not currently provided.
PROJECT NOTES
Project notes
These dates describe updates to this portfolio record, not product releases.
- Case record updated
- Recorded product status
- Local tool · implemented
- Documented work
- Implemented as a local tool and exploration of video workflows.
- Next evidence to add
- A publishable input-to-output example with source permission.
LET’S MAKE IT USEFUL
Have an AI product idea for a real workflow?
Give your next step some shape.
Your text stays on this page. It is not saved or sent automatically.
Next project