What the AI is doing, step by step
| Stage | What happens | What it means for you |
|---|---|---|
| Read | The video's YouTube captions are pulled with timestamps for every word | No captions on YouTube means nothing to read: On Creator and Custom, videos with captions turned off are transcribed by AI. |
| Find | The transcript is scanned for complete thoughts: a claim and its reason, a question and its answer, a story with an ending | Clips start and stop on sentences, not mid-word |
| Rank | Candidates are ordered by how well they stand alone and how strong the opening line is | You review a short list instead of the full video |
| Frame | Faces are tracked so the 9:16 crop follows the speaker; two speakers can be split top and bottom | Less manual keyframing |
| Caption | Words are timed individually and animated in your chosen template | Style is yours; wrong words get fixed in YouTube Studio |
More detail on the finding and ranking step is in how the AI clip finder works, and on framing in face-aware vertical reframing.
Steering the AI with a prompt
Left alone, the AI looks for generally strong moments. When you know what you want, a prompt gets you there faster. Write it the way you'd describe the moment to a friend who watched the video.
Prompts that work
- the story about getting fired from the first job
- where she explains why most people's SIP stops after a year
- the funniest exchange between the two hosts
- the 3 tips for answer writing in UPSC mains
- jab woh rent aur EMI ki baat karte hain (Hinglish works too)
Vague prompts (“something viral”) give generic results. Specific nouns from the video give precise ones.
AI vs doing it by hand
| Task | By hand in a phone editor | With an AI clip maker |
|---|---|---|
| Finding the moment | Scrub and rewatch, 10 to 20 minutes | Read ranked suggestions, 1 to 2 minutes |
| Cropping to 9:16 | Manual crop, re-adjust when the speaker moves | Face tracking or split screen |
| Captions | Auto-captions then fix timing by hand | Word-timed and styled from YouTube's captions |
| Judgement | Fully yours | Still yours: you choose, trim and approve |
Where AI clipping gets it wrong
- Jokes that rely on a look or a pause. The transcript doesn't see a raised eyebrow.
- Visual payoffs: a dish coming out of the oven, a before/after, a goal. If nobody says it, the AI may miss it.
- Context traps: a line that sounds great alone but means the opposite in context. Always watch the full candidate.
- Auto-caption errors on names, brands and numbers. These flow into your captions; fix them in YouTube Studio and paste the link again, or trim around the word.
- Crowded frames. With four people on a panel, face tracking may pick the wrong one; try Wide framing.
Treat the AI's list as a first cut from a fast assistant. The final pick should be the one you'd stake your channel on.
What comes out of the AI clip maker
Each clip exports as a 9:16 MP4 at 1080×1920, or 1440×2560 on Creator, with captions burned in from 21 templates (Basic and Basic Box on Free). You also get a suggested title and post caption; paid plans add the hook title, progress bar and hashtags. An SRT of the full video's transcript is available too. MyShorts doesn't post for you; you download and post.
Where other AI clip makers are stronger
Tools like Opus Clip, Vizard and Klap take file uploads and have bigger feature sets, including publishing and scheduling on some of them. If your footage isn't on YouTube, or you need a team workspace, start there and check their pricing pages (prices in dollars, checked November 2026). MyShorts is narrower: YouTube links in, Shorts out, rupee pricing, Hindi and Hinglish captions that render properly, plus summaries, notes and transcripts from the same link.
Cost
The AI runs on every plan, including Free: 2 clips a month from 1 video, with a small watermark. Paid plans start at ₹299 a month. Details on pricing.
Read the full guide: How an AI clip finder picks moments from a long video →
Last updated: