AI Video Cut

The core job is turning one long file into many short ones. AI Video Cut reads the transcription, looks for the moments most likely to hold attention, and cuts them into standalone clips. You get vertical cuts sized for social feeds instead of a single trimmed file. Simple as that.

Rather than describing every edit from scratch, you start from a prompt library covering trailers, viral clips, funny moments, and product reviews. If none fit, a custom prompt lets you write your own instructions for how the AI should cut the video. The library is the fast path; the custom option is for footage that doesn't match a template. Pick a mode and go.

The tool shows you the transcript and lets you cut by editing text. Delete a sentence to remove that part of the video, and clean up captions in the same pass. It works like a lightweight AI video editor that speaks your words. It's a practical way to kill filler words and false starts without zooming into a waveform, and it keeps the audio and video in sync automatically. Why type when you can edit the words instead?

Instead of a fixed center crop, the editor detects speaker faces and keeps them framed as people move. That matters on 9:16 output, where a careless crop chops off heads or leaves the subject off screen. Miss the face and the clip flops. The framing shifts to follow whoever is talking. No manual keyframes needed.

A range of caption styles lets you match the look of your channel, and burned-in captions keep viewers watching with the sound off, which is how most people scroll. You pick a style before generating, so captions land on the clip rather than needing a second app. That saves a step.

Prompts work with videos in any language, so non-English footage still gets clipped. The transcript accuracy is the part that decides whether this works: off captions mean off cuts. From our reading of the help docs, accuracy holds up on clear speech. Heavy accents or overlapping talk are the usual weak spots for any transcriber.

You can export in 9:16, 4:3, 1:1, or the original size, so the same clip fits Shorts, a square feed post, or a widescreen embed. Length is set by phrase count, with options like 7, 15, or 25 phrases, which roughly maps to short, medium, and longer clips. Regeneration lets you re-cut the same upload with new prompts without uploading again. No second upload needed.