Building AutoClip: An Open-Source AI Video Clipping Pipeline
Building a local-first AI video clipping pipeline that finds the best moments, creates vertical clips, adds animated captions, and schedules them through Buffer.
Building AutoClip: An Open-Source AI Video Clipping Pipeline
Long videos are easy to record. Finding the parts worth posting is the hard part.
Introduction
Short-form content is everywhere.
YouTube Shorts. Instagram Reels. TikTok. LinkedIn clips.
The problem isn’t creating the long-form content.
The problem is turning one long video into multiple short clips without spending hours scrubbing through a timeline, finding good moments, reframing the video, adding captions, exporting everything, and then scheduling each post.
There are plenty of AI tools that already solve parts of this problem.
But most of them follow the same model.
Upload your video.
Let their servers process it.
Download the results.
Pay for another subscription.
I wanted something different.
So I built AutoClip.
AutoClip is an open-source AI video clipping pipeline that takes a YouTube URL or local video, identifies moments worth clipping, converts them into vertical videos, tracks the speaker, adds animated captions, and prepares them for social publishing.
The main goal was simple:
Your machine edits. Your Buffer publishes. Your content stays yours.
Why I Built AutoClip
I wanted the convenience of AI-powered clipping without handing an entire video library over to another SaaS platform.
A typical workflow for creating short-form content looks something like this:
- Find the original video.
- Watch through it.
- Identify interesting moments.
- Cut the clips.
- Convert them to 9:16.
- Make sure the speaker stays in frame.
- Add captions.
- Export the videos.
- Upload them somewhere.
- Schedule each post.
Doing that manually for every video gets old very quickly.
So I wanted to automate the boring parts while keeping control over the actual media.
What AutoClip Does
The pipeline takes either a YouTube URL or a local video file.
From there, AutoClip handles the rest.
```text YouTube URL / Video File ↓ Transcription ↓ LLM Clip Selection ↓ Vertical Reframing ↓ Speaker Tracking ↓ Animated Captions ↓ MP4 Export ↓ Buffer Scheduling