The things people ask before their first episode. Anything missing — support replies to every message.
A Mac with Apple Silicon (M1 or newer) running a recent macOS. Everything — transcription, face tracking, rendering — runs locally, so a faster chip means faster clips, but there is no cloud dependency and no GPU server bill.
It's built for long-form talk content: podcasts, comedy shows, interviews, streams, panels. One universal pipeline handles them all — there are no content-type modes to configure. Gameplay-style layouts with a facecam get detected and framed as stacked automatically.
A reasoning pass reads the transcript like an editor: it looks for a hook that grabs a cold viewer in three seconds, a payoff that completes, and self-containment — no dangling references a stranger wouldn't get. Clips are cut to sentence boundaries and run their natural length rather than being chopped to a fixed duration.
It combines three signals: face identity (recognizing the same person across shots), voice-to-face matching (who on screen is producing the audio), and speaker turns from diarization. The crop follows whoever is speaking, with broadcast-style holds and cuts — and a quality gate verifies a face is in frame during speech before a clip can be published.
TikTok and YouTube natively from inside the app — you connect accounts on each platform's own sign-in page, and Cutfarmer uploads the clip with a title, description, and hashtags you can edit first. Instagram Reels publishing runs through the same pipeline. You can also just export files and post anywhere.
No. Processing is entirely local. The only things that leave your machine are the finished clips you choose to publish, sent directly to the platform you connected. There are no Cutfarmer servers and no account system — see the Privacy Policy.
Revoke Cutfarmer in the platform's own settings (TikTok: Settings → Security & permissions → Manage app permissions; Google: myaccount.google.com → Security → Third-party apps) — access stops immediately. You can also delete the local configuration file, which removes the stored tokens entirely.
Yes — that's the point of the app half. Per-clip timeline, framing inspector with manual subject override, tap-a-word caption editing, trim handles, and a natural-language command bar. Your corrections are remembered and improve future runs.
Ingest (transcription + analysis) typically runs several times faster than real time on Apple Silicon. Finding clips takes a couple of minutes per episode; each clip renders in roughly real time or faster depending on length and machine.
Cutfarmer is in closed beta and free for beta testers. Pricing will be announced before general availability — beta users get first access and a founder discount.
Request beta access with your email. Slots open in small batches; you'll get a download link the moment yours does.
A small independent team that clips real shows with it every day — see About. The roadmap is driven by our own daily publishing workflow.