Long video in, short clips out

Tradywatched73minutesandfound32momentsworthcutting.

Give it one podcast, interview, stream or webinar. It comes back with vertical clips and word by word captions, and it tells you why it picked each one.

One episode, as Trady read it
73 minutes watched, 32 clips found, median 1 minute 4 seconds

Real clips, not a mockup.

Every clip below came out of Trady, in the language it was spoken in. Play one and judge the captions and the framing for yourself.

hot takeKalau nggak pakai AI, saya pecat
storyZona blunder: kehilangan uang dalam semalam
storyDari modal 200 juta ke miliaran, lalu berhenti
hot takeNaikkan penghasilan, bukan tanya makro
quotableKekayaan sejati itu membosankan
informationalGen Z minum lebih sedikit, dan klub merasakannya
hot takePrioritaskan penghasilan, bukan investasi kecil
storyKenyataan pahit mencapai 1 miliar pertama

It reads the whole transcript, then argues its case.

Every candidate is scored 0–99 and tagged by what kind of moment it is. You see the shortlist with timecodes before anything renders.

28:54The pitch meeting that went sideways94
44:52The email that leaked everywhere92
1:36The five year detour that became the company91
1:00:00The demo that sold the company90
35:44Two years of building the wrong thing89

Captions land on the word, not near it.

Transcription is word-timed, so each caption appears as it is said — the same treatment this page's own headline wears.

A real clip, cut and captioned by Trady. The lit word is the one being said.

Wide in, phone-shaped out.

The frame is recut to a vertical crop that follows the speaker instead of the middle of the room, cut at the boundaries Trady found rather than a round number of seconds.

Watched
73 min
Clips found
32
Median clip
1:04
Shortest kept
60s
How it works

Three steps, and one of them is waiting.

00:01Hand it one long video

Paste a URL or upload a file — a podcast, interview, stream or webinar.

~ the length of a coffeeTrady watches all of it

It transcribes with word-level timing, scores every candidate moment, and renders the ones worth keeping.

your callReview, then post

Every clip shows its score, category and timecode. Download MP4s with SRT and VTT, or publish straight to your channels.

Questions

The things people ask first.

What does Trady do?

Trady takes one long video and returns short vertical clips with word-by-word captions. It reads the transcript, picks the moments worth cutting, and renders each one ready to post.

What can I feed it?

A podcast, interview, stream, webinar or talk — as a file you upload or a video URL you paste. It works on the audio and the transcript, so a talking-head recording is its best case.

Does it understand Indonesian?

Yes. Transcription and moment-finding both run on Indonesian audio, and the clip titles come back in the language the video is spoken in.

How long does a video take?

Roughly the length of a coffee for an hour of source. Transcription and scoring take under a minute; most of the wait is rendering each clip with its captions burned in.

How is Trady priced?

Per minute of source video, not per clip. An episode costs the same whether Trady finds thirty moments in it or four, and a run that finds nothing worth clipping is never billed.

What do I get back?

Each clip exports as a vertical 9:16 MP4 with captions burned in, plus matching SRT and VTT subtitle files, and every clip carries its score, category and timecode.

Point it at your last episode.

You see every moment it picked, with its score and its timecode, before you post anything.

Create an account