crosscut
Log inOpen the app
#1 AI video clipping tool

AI clips that
actually make sense.

Crosscut understands your entire video before creating a single clip. It finds the moments that belong together and assembles them into a complete story — bringing together the right hook, context, and payoff, wherever they appear.

orUpload a file
COUNTERINTUITIVESTORY_TURNSPECIFIC
How it works

One link in. Clips you'd actually post, out.

  1. 1

    Paste a link

    A YouTube URL, or a file. We pull the video and the transcript with word timings — nothing to upload, nothing to mark.

  2. 2

    The whole episode is read

    Not scanned for loud bits — read. The model builds a shortlist of moments with a real reason to be watched, then writes each clip as a script before it picks a single cut.

  3. 3

    Judged, then cut

    A second pass reads each assembled clip back as a stranger would and scores it. Only what survives is rendered — captions, 9:16 reframing, zoom rhythm and sound.

Why assembly

Every other tool cuts a window. We cut a story.

Trimming what everyone else does

Find a good ninety seconds and lop it out. The hook is wherever the conversation happened to be, the proof is three minutes later and gets left behind.

Assembly what Crosscut does

The claim from 58:00, the setup from 12:00, the number from 1:44:00 — joined into one argument that opens on its strongest line and ends when it lands.

  • Scored, not flattered. Each clip carries a craft score and a travel score, and the reviewer names what's wrong with it.
  • Never misquoted. A cut that removes a qualifier and makes a claim read stronger than it was said is rejected, however well it would play.
  • Yours to fix. Every beat is a word-level edit away: trim by word, strike a stumble, swap a beat, or tell the model what to change.
The editor

When the model is 90% right, finish the last 10% yourself.

YOU ONLY GET REPLACED IF YOU REFUSE

Captions in the real fonts

Eight presets, word-by-word animation, drag them where you like.

Media, text and music tracks

Drop b-roll, a title, a lower third or a track under the voice.

Every shape

9:16, 4:5, 1:1 and 16:9, with the speaker kept in frame.

Ask the model

“Shorter.” “Start on the question.” “Drop the tangent about X.”

Start with the episode you already have.

Every new account starts with free credits. See pricing

Open the app