Podcasters, streamers, and interviewers sit on hours of long-form video while short-form platforms reward posting every day. The clips are in the archive. The problem is getting them out: finding the hooks, reframing to vertical, captioning, and scheduling does not scale past a few clips a week when it is done by hand.
The pipeline
Viral Cut automates the whole path from a long video to published shorts:
- ClipAnything scans speech, scenes, and emotion to surface the moments most likely to land, across podcasts, interviews, gaming, and vlogs.
- Every candidate clip gets a 0–100 Viral Score, so the strongest hooks get posted first.
- ReframeAnything tracks the speaker and keeps them centred in 9:16, 1:1, or 16:9 with no manual keyframing.
- Animated captions in word-by-word karaoke, bold, box, and simple styles.
- AI B-roll cut on the beat from your transcript, plus generated opening hooks.
- A calendar to schedule clips and auto-publish to a connected TikTok account.
Built for teams and agencies
Brand templates lock the logo, fonts, and colours so every clip stays on-brand, whoever exports it. That matters most for agencies running several creators at once.
How it is built
Viral Cut runs on Next.js with an FFmpeg-based video processing pipeline, speech, scene, and emotion models, YouTube import with transcription, credit-based billing through Stripe, the TikTok publishing API, and a cloud render queue. The models pick the moments, but most of the engineering is in the pipeline around them: queueing renders, retrying failures, and keeping costs predictable. That is the same approach we take on AI automation projects for clients.
Try it
Viral Cut is live at viralcut.click. Drop in a long video and get ranked, captioned clips back.