Pick how tight you want it. The AI finds the slow, repeated and off-topic sections, gives a reason for each, and cuts only the ones you tick.
Free during launch. No account, no watermark.
The AI finds the sections worth losing, explains each one, and cuts the ones you approve in a single undo.
Choose how much to lose
Light, Medium or Tight, or type the minutes you want to keep. Tick the boxes to add retakes and long silences to the same list.
The AI reads or watches
With a Gemini key saved, Gemini watches a small copy of the video once you say yes. Otherwise Whisper reads the words on your PC and a cloud model picks the sections, after asking before any words are sent.
Review the list
Every section shows its time, its length, a summary and the reason. Play any of them, and untick the ones you want to keep.
Cut, or undo
Cut selected closes the gaps in the same order, and the captions follow. One undo brings everything back.
Does it cut anything without asking?
No. Nothing is cut until you approve the list, whether it came from the panel, the AI chat or Claude Code.
Does my video leave my PC?
Only if you agree. The Gemini route asks first, naming the file, the size of the small copy, Google and the price, and the upload is deleted afterwards. The transcript route keeps the audio on your PC and asks before the words are sent, as text.
Can a local model do it?
No. Judging what to keep needs a cloud model, so with Ollama or LM Studio the chat says so. Silence removal in the Audio tab still works.
Can Claude Code run it?
Yes, over MCP. It needs Trusted local switched on in Settings, and the same list appears in Evaloon for you to approve. No answer in five minutes counts as no.