Two ways to clip a YouTube video
The manual way: download the video with a third-party tool, open it in an editor, scrub through the hour to find the moment, set in and out points, crop the frame to 9:16, add captions, export, then repeat the export for each platform. It works, and it takes an afternoon.
The AI way: paste the link into a YouTube clip maker, let it transcribe the video and pick the moments, review the clips it returns, adjust a boundary if one is off, and publish. The rest of this article is that second workflow, step by step, with the checks that keep it honest.
Step 1: make sure you may clip it
A YouTube link is an intake method, not a licence. Before clipping anything, answer one question: do you have the right to cut this video and publish the result? Your own channel, a client's channel you manage, and content whose owner has given permission all qualify. A public video by a creator you do not know does not, regardless of what a tool allows you to paste; YouTube's terms restrict downloading and reuse of other people's content, and the platforms you post to enforce copyright claims after the fact.
Creators who want their content clipped increasingly say so explicitly, through clipping campaigns with a written brief. If you are clipping under one of those, the brief tells you what may be cut and where it may be posted; our article on virality clipping explains how those programmes work.
Step 2: paste the link and choose a range
Paste the URL. The tool reads the video's language and duration; check the language, because the transcript and the captions depend on it, and change it if the detection is wrong (a video whose title is in one language and whose speech is in another is common).
If the video is long, select a time range before processing. Only that range is transcribed, clipped and counted against your minutes. On a two-hour episode, clipping the twenty minutes you already know are strong is faster and cheaper than clipping everything and discarding most of it.
Choose a clip-length preference, or let the length follow each moment. Fixed lengths are a habit from the manual workflow; a moment is as long as the thought it contains.
Step 3: let the AI choose the moments
The clip maker transcribes the video with a timestamp on every word, then reads the transcript for passages that stand on their own: a claim and its setup, a story and its ending, an answer and its question. It extends a clip to the next sentence when the thought is not finished and skips passages that only make sense with the whole conversation or that depend on something on screen.
What you get back is a list of candidates, each with its transcript, a cover title, a score and a vertical render. The count follows the source; a dense talk yields more standalone moments than a rambling stream.
Step 4: review the three things that matter
Do not review by watching every clip end to end. Check, in order:
- Boundaries. Read the first and last sentence of each clip in its transcript. The first should be the setup, the last the payoff. If a clip starts on "and that's why" or ends on "but here's the thing", move the boundary; that is a one-click edit, not a re-edit.
- Frame. Watch the fastest exchange. The speaker should stay in the picture; two people in one shot should both be visible, which is what the stacked split screen is for. If the source is a stream with a webcam, the webcam should be readable on top and the game visible below.
- Captions and title. Read the captions in the video's language. Names, brand words and code-switched terms are where automatic transcription slips; fix them in the transcript and re-render. Check that the cover title says what the clip says.
Our guide to what makes a good clip lists the patterns that survive on their own; the same rules apply to any long video.
Step 5: download or publish
Every clip renders as a 1080×1920 MP4 with the captions burned in, so the file you download is the file you post, on YouTube Shorts, TikTok or Instagram Reels alike. If your accounts are connected, publish from the results page with the suggested title and description, or schedule the clips over the week. Trial clips carry a visible watermark; paid plans remove it and count minutes of source video, so the price of clipping an episode is known before you start.
Questions people ask
Do I need to download the video first? No. The link is enough. If the video is not on YouTube, upload the file or use a Google Drive link instead.
Can I clip only part of a long video? Yes, with the range selector. It is also the right way to test a long episode on a trial.
What if the tool finds few clips? Then the video has few standalone moments in that range. Widen the range, or accept that a quiet section is quiet. A tool that pads the list to a fixed count is not doing you a favour.
Which formats come out? Vertical by default; square and landscape are available when a platform needs them. The captions are part of the picture, so they survive every re-upload.
The YouTube clip maker page has the tool itself; the AI clipping guide explains what happens at each step under the hood.