← The Lab

No Final Cut at all: Ben Vallack had Claude Code build his edit suite, then directed it with voice notes

Based on the creator’s video · not yet tested by us

Instead of adopting someone else's AI editor, a YouTuber who has edited in Final Cut Pro for years asked Claude Code to build the few tools his own process needs: a transcript cutter, a searchable B-roll library, a phone app for fine-tuning cuts, and a player where he leaves spoken notes on the timeline.

Ben Vallack posted this video this morning (Oct 20) and says the whole thing was edited in a Claude Code session, without opening Final Cut Pro. He did the last 5% of his previous video in Final Cut, which he says came out first in the top-10 comparison YouTube shows him. He's a software developer, but his claim is that you don't need to be one: describe your editing process to the agent and have it build the tools for each step. He says he built the whole stack in less time than it took to edit his previous video.

The parts, as he describes them: (1) a transcript view of the A-roll where deleting words makes the cuts, plus an automatic first pass that removes flubs and retakes; (2) a B-roll library with every clip transcribed, so the agent knows what's in each shot and can place one that matches the line being spoken (his example is a clip of him moving from floor-sitting to standing, timed to the words about it); (3) a small web app on his phone that steps through each cut with waveforms so he can drag to trim and set crossfades; (4) a review player where he watches the edit on a phone or TV and records voice notes at points on the timeline, so the agent gets the timecode with each note ('we've seen this B-roll before', 'this is too long'); (5) scripts for titles and sound effects, plus exposure correction and LUTs.

The setup idea is worth copying even if you keep your NLE. Everything runs on one always-on desktop with the real footage. Your phone, laptop or smart glasses are just screens and microphones into it, connected privately over Tailscale, with no cloud subscription. He talks to the session by voice and uses a screen only when he needs to see a picture. As he puts it, it's like supervising an editor rather than dragging clips yourself.

Our take: the timestamped voice-note player is the piece most people are missing. It turns 'watch it and send notes' into input the agent can act on, much like the click-to-comment editors in huashu-motion and BaoCut, but built around your own footage. If you'd rather not build tools, those two and our editors roundup are ready-made versions. He's clear this replaces his YouTube workflow, not high-end finishing. Based on the creator's video, not tested by us. Credit Ben Vallack.

More from the Lab

Topics: Audio, voice & music