ClipFlow
Fully autonomous short-form video pipeline
- Status
- Shipped
- Role
- Architect and developer
- Source
- Private
Overview
ClipFlow takes raw fitness and lifestyle footage and returns production-ready reels without a human in the loop. Ingestion is triggered by a Google Drive drop; publishing happens on the other end automatically.
In between, the system runs AI video analysis to find the usable moments, sources trending music, handles the audio engineering, and generates dynamic captions timed to the cut.
The problem
Short-form content is a volume game, and the volume is bounded by editor hours. A creator posting daily spends more time trimming, captioning, and finding audio than filming.
The work is also almost entirely mechanical. The judgement calls are which moments matter and which audio fits — everything around them is repetition that a pipeline can absorb.
What was built
- Google Drive-triggered ingestion, so uploading a file is the entire creator-facing workflow.
- AI video analysis that scores and selects segments, feeding the cut list for the render step.
- Trending music sourcing from a local scrape catalog, with audio engineering applied to match levels across clips.
- Dynamic caption generation aligned to the final cut rather than to the source footage.
- A management dashboard for reviewing queued reels, and hands-free automatic publishing to Instagram and TikTok.
Tech stack
- Next.js
- React
- Fireworks AI
- LATE API
- Maton API
- TikTok API
Next.js
React