How to Repurpose One Long Video into 10 Shorts with AI

Use AI to find the most clippable moments in a long video and turn one upload into a week of shorts across platforms.

10 minIntermediate

One long video usually contains five to ten self-contained moments that work as standalone shorts. Finding them by hand is slow. AI clip tools and a transcript-driven workflow let you pull a week of vertical content from a single recording in under an hour.

What you need

A long-form video (a talk, podcast, or tutorial) with clear speech
A transcript, or a tool that generates one
An AI chat tool plus a clipping tool such as a captioning app

Step 1: Get an accurate transcript with timestamps

Everything downstream depends on the transcript. If your video tool does not export one, generate it locally with an open-source model so you keep the timestamps that tell you where each clip starts.

zsh - transcribe

transcribe with whisper, keeping timestamps

$whisper episode.mp4 --model small --output_format srt

[00:00.000 --> 00:04.200] Welcome back to the show.

Writing episode.srt

Step 2: Ask AI to find the clippable moments

Paste the timestamped transcript and ask the AI to flag self-contained moments that would hook a cold viewer: a strong opinion, a surprising number, a quick how-to, or a story with a punchline. Have it return the start and end timestamps.

find-clips.txt

Below is a timestamped transcript. Identify 10 segments (20-50s
each) that work as standalone vertical shorts. For each, return:

- start and end timestamp
- a 6-word title for the clip
- the hook line to overlay on the first second

Prefer moments with a clear claim, number, or mini-story.
<paste SRT transcript>

ChatGPT - Clip list

You

Identify 10 standalone short segments with timestamps and overlay hooks from this transcript...

Agent

1. 04:12-04:46 'The pricing mistake everyone makes' | hook: I lost $10k on this one setting.

The model returns ten timestamped segments ready to cut.

Step 3: Cut and caption each segment

Use the timestamps to trim each clip, crop to 9:16, and add burned-in captions. Burned-in captions matter because most people watch shorts on mute, and they noticeably lift watch time.

zsh - cut a clip

cut the 04:12-04:46 segment with ffmpeg

$ffmpeg -i episode.mp4 -ss 00:04:12 -to 00:04:46 -c copy clip1.mp4

frame= 1020 ... clip1.mp4 written

Overlay the AI hook in the first second

Put the suggested hook line as on-screen text for the opening second. A muted viewer reads the promise before deciding whether to keep watching, which is exactly when most swipe away.

Step 4: Stagger the posts

Do not dump all ten clips at once. Schedule them across the week and cross-post to Shorts, Reels and TikTok. Each clip can also link back to the full video, sending search and feed traffic to your long-form upload.

Result

One recording becomes ten captioned vertical shorts and a week of posts. Your long video keeps earning reach, and you spend the saved hours making the next one.