One prompt, one Claude Code project, and two hours forty-two minutes: the intro you just watched was cut entirely by an autonomous Premiere/DaVinci pipeline with zero manual touch-ups.
A single Claude Code prompt can run an entire video edit end to end in Premiere Pro or DaVinci Resolve, but only because a human first locked every stylistic decision, color grade, animation, and sound cue into a preset library the AI just executes.
The creator argues every existing Claude video editor can execute an edit but has no taste, so it still needs hours of manual prompting. His fix is to move all of that taste into a preset library, covering color grading, text animation, SFX, and graphics rules, then chain it into a pipeline: intake, a WhisperX rough cut with a recursive review loop, audio polish, color grade, a three-step graphics plan/build/review loop, SFX, a music bed, human review, and render. He proves it live by giving Claude one prompt, walking away for two hours forty-two minutes, and returning to a finished, postable edit with no touch-ups. The video ends pitching paid access to the finished preset package through his Skool community.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →
Cold open: the video was edited end to end by Claude off one prompt, and he promises to show that exact prompt.
Every Claude video editor he's seen, including his own from months earlier, can execute an edit but has no taste or autonomy: it still needs hours of back-and-forth prompting. He commits to building a version that needs none, and recording the whole unsupervised run on camera.
It's a folder of files, assets, docs, and presets that Claude reads to control Premiere Pro and DaVinci Resolve; the newest frontier model still lacks the taste to produce a postable edit alone.
Every stylistic decision, color grading LUTs, zoom in/outs, pop-ins, rise-ins, backgrounds, graphics rules, SFX presets, has to be pre-built so the AI executes a locked style instead of guessing.
The pipeline runs intake, a WhisperX rough cut with a recursive subagent review loop, audio polish, color grade, a three-stage graphics plan/build/review loop, SFX, and a music bed before human review and render.
The system is organized as a presets folder plus separate 'lanes' per target app: Premiere Pro, DaVinci Resolve, and CapCut, with Premiere and Resolve working best today.
He hands Claude the raw footage path, a specific background music track, and one instruction about pulling real YouTube thumbnails, tells it not to stop until finished, and leaves.
He returns from an errand to a finished edit: two hours and forty-two minutes of unsupervised work, about 740,000 tokens of context used, and no manual intervention.
He pitches his paid Skool community for the finished preset package (or a free DIY path from his other videos and a published Claude artifact), then argues Claude now does roughly 80% of a full edit and genuinely removes hours of editing labor.
The finished system is sold as owned files, not a subscription, but only reliably handles talking-head, educational YouTube content today, not vlogs, visual content, or vertical short-form.
The real lesson isn't that Claude can edit video, it's that unsupervised AI execution only works once every stylistic decision has already been made and locked into a preset library the AI just runs.
“Claude by itself lacks the creativity, the taste, and workflow instructions to be able to do an entire edit end to end and have it come out actually postable.”
“I just got back from the store and I think it should be done now. It did it. It did it. Total two hours, 42 minutes.”
“I would say Claude can do 80% of the work for you now.”
“This actually is fully replacing an entire position and department in my business.”
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
The intro playing right now, the creator says, was cut entirely by an autonomous Claude Code pipeline off a single prompt, and he's about to show the exact prompt and the two-hour-forty-two-minute unsupervised run behind it.
The fixed sequence of stages the autonomous system runs a raw-footage folder through, with a recursive review loop inside the rough cut and a three-step plan/build/review loop for graphics.
Everything the creator says has to be pre-built and locked before an AI can run an edit unsupervised without producing generic-looking output.
“if you want to build a system like this for yourself... it's published in my school”
Two-tier pitch: a free DIY path via his other videos, transcripts, and a published Claude artifact, or the finished preset package through his paid Skool community, framed explicitly as owned files rather than a subscription.
00:00
00:12
00:23
00:36
00:41
00:56
01:0601:16
01:26
01:37
01:47
01:57
02:07
02:1502:28
02:38
02:48
02:58
03:0803:19
03:29
03:39
03:49
03:59
04:1004:20
04:30
04:40
04:5105:01
05:11
05:21
05:31
05:42
05:52
06:02
06:12
06:22
06:33
06:43
06:53
07:0307:14
07:24
07:34
07:44
07:54
08:0508:15
08:25
08:35
08:45
08:56
09:06
09:20
09:26
09:37
09:47
09:5710:07
10:17
10:28
10:38
10:48
10:58
11:08
11:19
11:29
11:39
11:49
11:5912:10
12:20
12:30
12:40
12:51
13:0113:11
13:21
13:31Add Modern Creator as a preferred source and Google shows you more of our breakdowns in Search, Top Stories, and AI Overviews. It only changes what you see, and you can undo it in your Google settings anytime.
19:02A creator wires Claude Code into DaVinci Resolve through a free MCP bridge and runs every stage of an edit live on camera, rough cut, motion graphics, and color grade, with a timer running so you can see exactly how long each pass actually takes.
August 7th
27:04A screen-recorded, un-sped-up walkthrough of five real motion-graphics jobs — rough cut, animated background, number counter, typography titles, Vox-style collage — built entirely by prompting Claude Code and letting it drive CapCut.
August 3rd
20:58A YouTuber hands Claude full control of Premiere Pro — cutting a rough cut and generating an HTML-animated Mars/Earth graphic from prompts alone — then walks through building the same system for free.
July 24th
06:38A YouTube creator maps the three-tool, three-gate content engine, Claude Code, Remotion, and WhisperX, that auto-edits his talking-head videos from raw footage to a finished MP4.
July 9th
06:10A three-step prompt workflow turns Claude Code into a chat-driven motion-design studio that reverse-engineers any video's cuts, motion, and B-roll.
September 2nd
09:15A YouTuber hands Claude Code a boring talking-head recording and gets back motion graphics, camera zooms, and a pitch for a $2,500-per-client AI editing service.
July 27th