Hey Reader,

Every day, someone on my feed posts the same lie.

"Claude edited this video in just two minutes, I just dragged it in!!!"

This is complete bullshit.

I know, because I spent the last 2 months and $70k+ in AI costs solving this.

Most of that went into building my own editor, more on that at the end.

But the basic version takes just 3 tools and about half an hour of setup.

Here's how you can test it yourself for free.

How it actually works.

The 3 tools you need are (1) Scribe by ElevenLabs, (2) FFmpeg, and either (3) Claude or Codex.

Scribe writes down what you said in the video.

Claude or Codex decides what to cut or rearrange.

FFmpeg does the cutting without you even seeing it.

You can just ask Claude/Codex to install all of them for you.

Why Scribe by ElevenLabs.

Listen, I've tested them all. Yes, there are free versions. But they suck.

To cut a video cleanly, the AI needs to know where every word starts and ends.

And only Scribe gives you that, down to the millisecond, to a degree that is worth your time.

It's like karaoke lyrics that light up word by word, so Claude knows exactly where each word begins.

I tested Whisper by OpenAI, Grok Voice, and a bunch of others. None of them did as good a job.

What's FFmpeg, and why is it everywhere?

If you've watched any videos on the topic, you'll have heard about FFmpeg.

It's essentially a free video editor with no buttons, that your AI can control.

Like the engine in your car, because lots of editing tools run on it without you ever seeing it.

You never touch it yourself, since Claude or Codex already knows how to use it really well.

It can even add simple subtitles, and fancy subtitles are overrated anyway.

Claude or Codex?

Use the one you already pay for. It really doesn't matter.

Astra in Codex and Opus 5.5 in Claude are both incredible writers.

And that is exactly the skill you want the model to have for editing. I know, it sounds weird.

but the way that this works is that AI looks at the words and tries to make sense of them.

That's a writing skill, not a video editing skill.

Just keep everything in a single folder, so it remembers how you like your cuts.

The fancy stuff (and why you should skip it).

There are two free tools that can add moving graphics, text, and sound effects to your videos:

1. Remotion, which is a little bit older but more stable
2. Hyperframes, which is newer, faster, but also a little less reliable

But here's the thing: while they both look really cool, they will eat a ton of your time and will absolutely not add money to your bottom line.

The thing that people really care about is what you teach or show them, not how fancy your edits are.

I focus on one thing and one thing only, and that is reducing my time-to-post.

I call this TTP, because I learned that the faster you post, the more ideas I get out there.

Why I built my own.

But once this worked, I hit the next wall.

A chat window is a messy place to edit video, because Claude gets distracted, burns through your subscription, and leaves files lying around that you then have to clean up yourself.

So I built The Attention Machine, which runs the whole job from the edit to the hook to the post.

Try the setup this week, then hit reply and tell me what broke.

Talk soon,
Rob

P.S. If you're an online coach and you want to put the machine to the test on a live call, book a demo here.​