I Built an Open-Source AI Video Clipping Pipeline

I Built an Open-Source AI Video Clipping Pipeline I wanted a tool that could take a long video and turn it into social-ready clips without uploading the entire video to some SaaS platform. So I built AutoClip. The idea is simple: YouTube URL / Video File ↓ Transcription ↓ LLM Clip Selection ↓ Vertical Reframing ↓ Speaker Tracking ↓ Animated Captions ↓ MP4 Export ↓ Buffer Scheduling The interesting part is that the video processing happens locally. AutoClip uses an LLM to identify moments worth clipping, but only the transcript text is sent to the LLM. The actual video stays on the user's machine. It supports multiple caption styles: - Word Pop - Karaoke Fill - Clean Lower Third - Boxed The final clips are rendered as 9:16 MP4s ready for social platforms. Publishing Architecture I also wanted publishing to follow the same privacy model. Instead of building a cloud scheduler, AutoClip uses a BYOK approach. The stack is: Your PC ↓ AutoClip processes video ↓ Cloudflare R2 ↓ Buffer ↓ Social Platform The R2 bucket acts as temporary media storage so Buffer can access the video. Once the job is handed off, the user's computer doesn't need to stay online. No central media storage. No SaaS layer. No subscription. No watermark. No telemetry. Just your infrastructure and your API keys. Why I Built It There are plenty of AI video tools already. I wanted one where I could actually understand and control what happens to my content. The goal was: Your machine edits. Your Buffer publishes. Your content stays yours. AutoClip is open source. GitHub: AutoClip The local-first AI clipper that finishes the job. Drop in a long video → get ranked, caption-burned, speaker-tracked 9:16 clips out - and schedule them to post while your PC is off. Paste a YouTube link or drop a file. AutoClip transcribes it, uses an LLM to find the moments actually worth clipping, reframes them to vertical while tracking whoever is speaking, burns in animated captions, exports platform-ready MP4s - then hands the finished clips to Buffer to publish on schedule. No accounts with us. No uploads to our servers. No watermarks. No subscription. Your machine edits. Your Buffer publishes. Your R2 bridges the media. Nobody owns your content but you. What it does - Ingest anything - a YouTube URL or a local file, hours long or minutes. - Find the real hooks - an LLM reads the transcript and ranks the moments worth clipping, down to the word. Timing… I'd love feedback from anyone building content automation, video tooling, or local-first AI applications. Top comments (0)