How the yeti finds the good bits

1. It listens

Your recording is transcribed word by word, with the time of every word, in whatever language you speak.

2. It reads the whole thing

Claude, an AI model, reads the full transcript and looks for moments that work on their own: a strong opening line, a clear point and a satisfying end. It skips intros, housekeeping and half-finished thoughts.

3. It scores every pick

Each candidate is scored out of 10 on five things: the hook (do the first seconds make you stay?), the payoff (does it deliver?), whether it stands alone, its energy, and how on topic it is. The best ones come first, and each comes with a plain sentence saying why it was picked.

4. It fixes missing context

If a clip needs one line of setup to make sense (“She was asked why she left her job”), the yeti suggests a short context card or a setup line from elsewhere in the video.

5. It frames and captions

For video, the frame follows the speaker’s face as it cuts to vertical. Captions are timed to each word in your choice of style, using your brand kit’s colours and logo.

6. You edit, then export

Delete words from the transcript to cut them, ask in plain words (“make it 20 seconds”, “punchier hook”), and export in vertical, square or landscape. You pay one credit per exported clip.