How I Trained Claude Code to Edit Like a Real Editor (7 Rules)
A creator who has trained over 100 human video editors lays out the seven rules he taught Claude, and why the assets behind the rules matter more than the prompt.
Posted
1 weeks ago
Duration
Format
Listicle
educational
Views
17.7K
213 likes
57 · 43
Big Idea
The argument in one line.
A good AI video editor comes from ten years of curated rules and assets plus clear creative direction, not from a quick vibe-coded prompt, and seven rules capture how to train one.
Who This Is For
Read if. Skip if.
READ IF YOU ARE…
A creator who films talking-head content for short or long form and wants to hand the edit to an AI instead of a human editor.
Someone vibe coding their own video editor with Claude Code who wants to know what rules and assets the output actually depends on.
A creator who keeps getting AI edits back that feel generic and wants a repeatable way to give direction up front.
SKIP IF…
You want a step-by-step Claude Code build tutorial. This is a rules overview with no code on screen.
You are not editing talking-head or educational content. The rules are framed around that format.
TL;DR
The full version, fast.
Stockton Walbeck trained Claude to edit video the way he trained over 100 human editors, and he boils it to seven rules. Give a creative brief with references, feel, and starting visuals. Sift dead space and retakes with measurable rules. Feed specific inputs and assets, since that is the moat over vibe-coded editors. Keep motion and transitions restrained. Cut between intentional scene layouts. Build a visual once and reuse it. Make the AI quality check its own work and remember every piece of feedback. He is launching his own editor November 18.
Free for members
Chat with this breakdown — free.
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.
Cold open claims the whole video was AI-edited and sets up the seven rules.
00:30 – 04:20
02 · Rule 1: Creative direction
Technical vs creative engine room, and a creative brief built from references, feel, and starting visuals.
04:20 – 06:51
03 · Rule 2: Sift scientifically
Dead space and retake removal measured by frames, decibel floors, and five retake types.
06:51 – 10:31
04 · Rule 3: Good input, good output
Specific inputs and ten years of curated assets are the moat over vibe-coded editors.
10:31 – 12:29
05 · Rule 4: Less is more
Avoid retention maxing. Every element needs intention, with a three-transition cap.
12:29 – 14:13
06 · Rule 5: Intentional scene cuts
Long form and short form compositions, and knowing when to cut between them.
14:13 – 16:03
07 · Rule 6: Make it once, use it 10x
Reuse one or two core visuals across a listicle instead of building one-off graphics.
16:03 – 19:35
08 · Rule 7: Catch it before I do
Self quality checks and a feedback memory that compounds into permanent fixes.
19:35 – 20:46
09 · Launch and close
Pitch for the AI video editor launching November 18 with a ten-day enrollment window.
Atomic Insights
Lines worth screenshotting.
A 10 minute creative brief with references, feel, and starting visuals saves hours of revision feedback later.
Give an AI one or two reference videos and say what you like about them. Fifty references only confuses it.
Dead space can be measured: chop audio into 10 millisecond windows, take the quietest 10% as the room floor, and cut gaps of two to three frames.
Retakes fall into five types an AI must learn: repeats, reopens, replacements, rephrases, and exceptions.
Words ending in S, T, D, K, or P are where AI most often cuts too early or too late.
The same request, write me hooks, gives very different results when you add a hook template doc, an ideal customer profile, and viral examples.
Anyone can vibe code a video editor in about two hours, so the real moat is years of curated assets and rules.
Retention maxing, with transitions and motion on everything, is a big reason edits look like AI. How much animation to use is a dial you control.
A cap of three transitions per video keeps scene changes from feeling like a roller coaster.
Spending five hours on a graphic that shows for three seconds is waste. Build visuals that return 12 to 15 times in a listicle video.
Making the AI catch obvious technical mistakes before you review turns 45 minutes of feedback into five.
An AI that logs every timestamped comment can spot patterns after five or six videos and make permanent fixes, which human editors forget.
Takeaway
Direction and assets decide AI edit quality
WHAT TO LEARN
An AI editor performs like a trained human one: it needs a creative brief, measurable rules, specific inputs, and a mandatory self-check before you review anything.
Write a ten minute creative brief with one or two references, the feel you want, and starting visuals so the editor does not guess.
Turn sifting into measurable rules such as frame-length gaps, a decibel floor, and defined retake types instead of asking the AI to cut the bad parts.
Feed specific assets and examples with each request. The same prompt gets much better output when it comes with templates, an audience profile, and proven examples.
Treat restraint as a setting. Cap transitions, tie each effect to a reason, and choose how animated the edit should be per video.
Define a small set of scene layouts and the moments each one is for, so cuts are intentional rather than constant.
Reuse one or two core graphics across a listicle video. A visual used a dozen times justifies the time spent building it.
Make the editor check its own output and log every timestamped comment, so repeated complaints turn into permanent fixes.
Glossary
Terms worth knowing.
Sifting
The first editing pass where dead space, mistakes, and retakes are removed from raw footage before any visuals are added.
Retention maxing
Adding as many transitions, motion effects, and sound effects as possible to hold attention, which tends to look cheap and over the top.
Creative brief
A short starting document giving the editor references, the general feel, and starting visuals so the edit matches your expectations.
Technical engine room
The fixed foundation of rules, assets, and workflows every video needs regardless of taste.
Creative engine room
The subjective part of editing, taste and vibe, which changes from video to video and needs direction.
Composition
A scene layout such as full screen, speaker in a side box with visuals, or small circle speaker over B-roll.
“You, right this second, could vibe code a video editor with Claude Code in about two hours or less.”
provocative claim that sets up the moat argument→ IG reel cold open↗ Tweet quote
16:50
“Catch it before I do.”
the rule in four words→ newsletter pull-quote↗ Tweet quote
The Script
Word for word.
Read-along
Don't just watch it. Burn it in.
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.
17px
Everything you're about to watch was made by my AI video editor. I've been editing for over 10 years. I've hired and trained over a hundred video editors myself.
And now I've trained Claude the exact same way I trained my human editors. The same rules, the same templates, the same workflows, same tutorials. And in this video, I'm going to go over the seven rules I used to train my AI video editor to get exactly what you're watching right now.
My goal here today is that you can take these rules and using the vibe code or train your own AI video editor or just use mine, which I'll... I'll show you a waiting list to join here in a minute. But first, let's start with rule number one, and that is to give the editor creative direction.
So first and foremost, I like to think about video editing split into two types of engines, right? Engine rooms. One engine room is the foundation of what every single video needs.
Okay, it's all the rules. It's the assets. It's the workflows to make a video work.
It doesn't matter what the video is. We'll call this the technical engine room. Does not change.
And then you've got the creative engine room. This is where everything is preferential in nature. This is subjective.
It's relative. Every single person's got a different taste, vibe, and flow they want in their videos. This one does not change video to video.
It's the foundation. But this one... does.
And that's important for one obvious reason. Whether it's humans or whether it's AI, if you don't give your video editor good instruction on what you're looking for and what the end goal is, it's going to just do whatever it pleases. Sometimes that might be good, but oftentimes it's kind of a crapshoot.
And AI, just like humans, does not work well with no direction. Technically, yes, I could just go to AI and say, hey, here's my videos, here's my footage, here's my audio, here's some B -roll, make me a video and make it good, please. You can do that, but it doesn't mean the output is going to match your expectation.
If you have no expectation, which maybe that's some of you, then that approach actually does work. That's great. But because editing and videos is creative in nature, you have to give the editor creative expectation.
And that's precisely what I do for every single video edit that I hand off to either a human or AI. Now, people get really lazy here because yes, it takes some time, takes some effort to give some creative direction. But if you don't give it that 10 minutes of just juice to work with and what your expectations are and what you want the video to look like and feel like, what ends up happening is you'll spend two hours giving feedback later.
And what I'm talking about is called a creative brief. It's a 10 minute starting point with some clear expectations and direction that I promise you, it's going to save you hours later of giving hundreds of points of feedback for your video edit. And I found that my creative briefs are generally made up of three different parts.
References, the feel and starting visuals. References is probably the easiest win here. You're literally just treating AI like a monkey, right?
You're giving it a banana, like good banana. This is good. This is a good example.
Do more of this. No, so good. Don't do this.
It's like you got to train it. And so I will take videos. If I'm getting an Instagram video edited, I will take one to three videos that I like or for YouTube videos, same thing.
And you say, this is what I'm wanting it to feel like. And with references, I think discipline is key here. You can't give AI 50 videos to look through.
It's just too much. It's going to be too confusing. I just like to give it one or two videos as a reference point.
And then I talk about what about that reference I like. I like the way they use B -roll. I like the way their compositions are flowing.
I like their sound design. What exactly are we looking for from the reference? Second is the general feel.
It's kind of the vibe of the video. This is the most kind of ambiguous and amorphous part of the creative brief. It's just a general direction of what you're wanting for the video.
How heavy are the visuals going to be? How fast do you want it to move? If there are specific moments you want the editor to really focus in on, you'd be surprised.
how just a really generic kind of like ambiguous, long -winded rant about what you want it to look like can actually be taken by AI, reverse engineered, and turn that into something a lot more tangible and specific. And then finally, I give it starting visuals. As you've noticed, I'm going over the seven rules for editing.
Beforehand, I knew in my creative expectations in my brain that I wanted to have this nice motion graphic of cards to represent those rules. But again, if I never gave it that expectation and showed it what I wanted for the video, it would have just guessed. Maybe it would have and pillars or fingers.
I don't know. Like that's the whole point. But by me knowing what I wanted in the first place and by giving it that starting visual starting point, I screenshotted, hey, here's a look at these cards that I like how they look.
Can you use that throughout the video? It seems so simple and stupid and yet most people don't do this. And so they just hand off a bunch of footage.
They don't like the output and like, this sucks. It's like AI does not know what is good and what is bad. And so you got to give it the starting point.
And that leads us to rule number two, which is to sift scientifically, not by feeling. Sifting is the first step every video editor takes when they get a video. Before we can touch anything, before we can add visuals and B -roll and all that fun stuff, you have to sift your video and that comes in two parts.
Dead space removal and the retake and mistake removal. And I will tell you right now, this seems like one of the simplest parts about editing videos, but I have trained literally over a hundred humans to learn how to sift, how to take out a dead space. What is a mistake?
What is a retake? And I don't know if it's just because I'm a bad teacher or if there really is like... like hundreds of little nuances of how to do that well.
But there are hundreds of different rules you have to follow to sift a video. And honestly, I didn't even know that until I forced myself to have to train AI to sift for me. For example, dead space is normally measured in frames and DB levels.
Two to three frames for me, for example, is generally what I consider a dead space. And so my AI editor chops the audio into 10 millisecond windows. It takes the quietest 10 % of those windows and it calls that the floor.
And that floor, generally speaking, is... just the sound of the room you're in. But I only figured that out after giving AI hundreds of my videos and Premiere profiles to figure out why am I cutting certain dense spaces and not.
There's also a whole nother messy world of sifting where like, for example, when a word ends in an S sound or a T, a D, a K, a P, those are all vowels and letters that AI has a hard time knowing when that word truly ends. And so it often will cut off words too soon or not soon enough. And then we got retakes, right?
If you look at any video that I film, there are dozens of reasons why I might do a retake. I might start a sentence and just not like the tone I set it in. And so I just completely reset back to zero.
I might finish my whole sentence and just not like the pacing of how I said it. And so I do it again. But the most common one is I just start saying a sentence and I flub a word.
I just like maybe blanked out or lost my train of thought mid -sentence. There could be an awkward pause in there somewhere. These are all mistakes and retakes.
And what's why? Wild to me is only after training AI to do this, I found that it's actually better than my human editors at doing this. I found that there are five types of things I had to train AI on to get it to recognize what is what.
There's repeats, reopens, replacements, rephrases, and exceptions. Each one of those has dozens of rules for recognizing what is a retake, what's a mistake, what's a dead space. And so now I can take out my phone if I'm filming a short form or I can film on my camera for long form.
I just take my raw out of camera footage. I drop it into my AI video editor. No instructions, no context.
It understands everything. Sifts it all the way down. And I've got a beautiful sifted output ready to storyboard, add visuals to, and so on.
Then we've got rule number three, which is that good input equals good output. This, in my personal opinion, might be the most important rule in the entire deck. You should all know by now that AI's output is only as good as your input.
That goes for everything, right? AI video editing is no exception. Perfect example.
Okay, let's say that if you asked AI, hey, write me a good hook. That request is a complete crapshoot as to whether or not it's going to stumble, like trip into writing a good hook or not. Chances are low.
You're probably instead going to get a bunch of AI. sounding hooks that are not rooted in any copywriting principles. But let's say instead I said, here's a Google doc with 100 of the best hook templates with examples.
Here's another document outlining who my ideal customer profile is. So you know who I'm speaking to and everything about them. And lastly, here's five Instagram videos that all went viral on similar topics to mine.
Look at those, get some inspiration, and then write me 10 hooks for my next video. The only difference between those two, they had the same request, write me hooks. But the one difference is that I gave it three distinct types of input and assets.
The other one was generic poo -poo, right? We all know which output is going to be better. Yes, AI right now has an okay starting point that hundreds of creators right now are vibe coding off of, right?
They're making their own AI video editors. It's built off of that same AI slop. that again, sometimes is surprisingly good, but it's super hit or miss.
It's really inconsistent. It's not going to give you the best of the best stuff, especially if your goal is to make talking head content like I do for both short form and long form videos. And Claude right now is just scanning the entire internet with a huge broad net to just get general good editing, motion, graphic, visual templates and understanding.
But to really level up your video edits, you have to have specific inputs. This to me is going to be the real moat of what separates exceptional AI. video editors from vibe -coded slop ones.
That moat. is the difficulty level or the barrier to entry to build an AI video editor. You, right this second, could vibe code a video editor with clod code in about two hours or less.
I promise it's not that hard. The moment to do that is very small. But this again is precisely what thousands of creators are already doing right this second.
I've bought a lot of these. And what I found is that they're all just a bunch of washed down skills and prompts that have generalized direction. And you can tell the people who made these spent maybe two hours on it and it shows any video edit.
But what you cannot do is build over 10 years. of editing assets to feed to AI. Assets like, for example, composition rules and templates.
I have 127 of them. Some for long form, some for short form, some for both. Infographic and framework templates.
You're watching a framework right now that I used, these seven different cards. That's one of over 93 different framework graphics mapped to five different types of frameworks. You got icon and PNGs, 120 files, 36 icons, 20 logos, 25 device mock -ups.
Then you've got sound effects and sound libraries. Teaching AI exactly when, to attach a sound effect to a motion and when not to.
Then you've got motion mechanics. I found there's about 12 of them. These are how things enter on the screen, how they leave the screen, how they animate on the screen.
You've got sifting rules, storyboarding processes, quality checking rules. The list just keeps going on. You have to have good assets.
And this to me is why I believe my short form and long form videos look so good. It's dipping from assets that I hand curated and crafted over 10 years. But it's also why I don't have to spend eight rounds of revisions on one video and millions of tokens on my cloud code.
subscription to get to this point, which by the way, while we're at it is exactly why I'm launching my own AI video editor on November 18th. You can click the link in the description to join the waiting list for that. Or if it's already launched by the time you're watching, you can go and get it now.
Again, doesn't matter to me which route you take. You can either go vibe code and build your own version of it, which again, you got to have really good assets to do that. Or you can just get mine, which is already wired up with all those assets.
And you're taking this ginormous shortcut to just get to exceptional far faster. But next we've got rule number four, which is less is more. This one is a bit relative, Maybe it's a little bit subjective in nature, but I find that it's generally speaking pretty universal.
One of the biggest complaints I see on AI edits right now is that people say this looks like AI. And a big part of why a video looks like AI is because it's just a little bit over the top. It's like retention maxing, right?
Everything's moving. There's so much happening all around me all the time. It looks a little cheap, maybe a little out of control.
But what's funny about this is that lever of how retention maxing or not you want your video to be is entirely in your control. How much AI chooses to animate things or make visuals for or add sound design to and so on is a choice. It is a lever, a spectrum that you can choose.
And this concept of less is more as it applies to videos. And as I've learned as a video editor is intentionality. Everything you put in a video has to have a reason, a purpose and an intention.
One of the biggest mistakes I made as a beginning and up and coming video editor is I tried so hard to make the video interesting that I engaged in what I call retention maxing, right? Which again is where you add as many transitions as you can.
You're slapping motion onto everything possible. You're using over the top sound designing and sound effects. And again, some of you, truth be told, might prefer that look.
Others hate it. And again, lucky for you, it's a choice. When I upload my videos to my AI video editor, again, sometimes I'm like, hey, let's dial it back this time.
And let's just not have as much stuff on the screen. Maybe just focus on the captions and maybe the occasional B -roll. But this less is more rule applies to other things like transitions and tile animations, right?
I've got a rule, for example, where I don't want to have any more than three transitions in a video. When I transitioning from scene to scene if i'm swiping and zooming in really aggressively between every single scene like you're watching right now do you see how this starts to feel like a roller coaster it's like really nauseating and it's not an enjoyable user experience sometimes i want to swipe other times a zoom sometimes more subtle like this or a film burn like this everyone's going to have their own taste and flair but again less is more rule number five is to have intentional scene cutting in the world of video editing there are what i call personally scenes you can also call them compositions, layouts, a couple different words for it.
Long form videos have different compositions than short form videos do. For example, here's a look at all the different kinds of compositions that I like to cut between when it comes to long form. Sometimes I like to be full screen like I am now.
And then other times I want to push myself off the side inside of a box with some visuals on the right hand side. Other times I want to be right here in the middle and then compare two options on both sides of me. Or we might just go full screen to a motion graphic or B -roll and I'm not in there at all.
Maybe I want to be a small screen circle on the left or right -hand side. You get the idea, right?
There's dozens of different kinds of compositions and scenes that after you've sifted and transcribed your whole video, you got to try to reverse engineer and figure out when should I cut to different scenes and compositions. And then short form is a whole different animal. It's got far less scenes, but still many to choose from.
And there is a skill and an intelligence behind knowing when to cut between different scenes and how. Again, editing, especially with AI is all about discipline and fitting it within rules and bounds. AI has to have a good starting point, foundation, assets, and inputs, but also has to know when is too much, too much, which is again, why making a good AI video editor is so hard.
I look around at all these influencers vibe coding their AI video editors. And I think number one, first of all, I think it's fine. There's nothing wrong with wanting to make your own version of something and then cash in on this big hype train that's happening right now.
But two, again, I realized there are literally thousands of rules, guidelines, assets, inputs, AI needs to do it well. And the cool part about my editor launching on November 18th is I'm always going to be pushing. weekly and monthly updates, both to the engine that's technical in nature and the creative room where you get all the best assets, templates, sound effects, and so on.
So the tool just gets better every single month. Next, we've got rule number six, which is to make it once and use it 10 times. This rule is where we're entering now the world of efficiency.
And efficiency, again, is important for both humans and AI. It's important for AI because if you don't have an efficient editor, your tokens are going to get used like crazy. That gets expensive.
Efficiency is also important because I don't want to wait 10 hours to get a short form video. back. And so a big common issue that I found with editors is that they would spend sometimes five to six hours on a single visual, like one graphic or one motion graphic, et cetera.
And that visual was only on the screen for about three seconds of the whole video. And so if I'm filming a 15 minute video, that is a massive waste of time. But what you'll find with almost every video is that there are some visuals that you're going to be using throughout the entire video.
For example, this YouTube video right here, I was going over the top 12 prompts I use with AI. And to visualize those 12 prompts, I built this cool little carousel looking motion graphic. This graphic took a long time to build, but the investment of that time was only worth it because we used it 12 to 15 times throughout the rest of that 18 minute video.
Again, I make educational content and educational content. I use a lot of listicle. type of videos.
This video right here is listicle. I've got seven rules. This one was also listicle.
It's 12 prompts. And at each prompt, I was revealing that motion graphic. Once again, it'd come back on the screen.
And so I built it one time and used it 12 to 15 times, making that video really come to life. And it made that time investment worth it. Again, when it comes to like educational talking head content like this, they're generally speaking is going to be one or two core visuals.
And you can reuse those or make slight variations to them all throughout the whole video. This is I know this rule might sound kind of boring, like, oh, like be resourceful, be efficient.
But the reality is, again, we're all trying to save money on edits. We're trying to save time. So if speed and money is important to you, you got to have a resourceful, efficient AI video editing machine.
And lastly, we've got rule number seven, which is to catch it before I do. I have probably reviewed, I don't know, in the last 10 years, over a thousand videos, and I've given feedback on all of those. And I would say my biggest pet peeve is that I absolutely despise giving feedback on a really obvious technical mistake in a video that my editor should have caught himself the first time.
Because one of the biggest time suckers when it comes to getting video edits back is revisions. And so if my video editor hands me over this kind of piece of junk, not quality checked video, I'm going to be spending 45 minutes leaving hundreds of comments throughout the whole video that otherwise would have taken me five minutes had they quality checked it.
And this rule, rule number seven, is to me the biggest advancement when it comes to AI and the biggest reasons to use AI to edit your video. videos and no longer humans. Because again, human editors oftentimes just kind of get lazy.
They do the first revision. They don't want to go back and rewatch the whole thing and catch spelling mistakes and these little technical glitches and export errors. That takes a lot of time.
I don't want to do that. I'll make him give the feedback for me. With AI, I make a non -negotiable rule.
It has to follow. You have to go through and catch all these mistakes before I do. So I'm not wasting time on a revision.
Every revision is going to take anywhere from 10 minutes to two hours, depending on how much stuff you're giving back. And I don't want to waste my time giving feedback on those things. So the goal is to save the amount of time you're spending in the revision room.
Because remember, revisions are always going to be part of the game with any video editor. There is no such thing as a perfect edit on version one. It's almost never happened in my lifetime.
But where AI is so powerful, okay, is it will take a video and each and every time I leave my feedback with comments on that video, it is going to keep track of every single comment down to the timestamp, what my... preferences are, my non -negotiables, the things that I don't like, the things that I do like. And it becomes this compounding effect where all of this feedback starts to become one cohesive understanding of how Stockton likes his video edits done.
You know, I think back to the hundreds of videos that I've had my human editors do for me. And collectively, there are tens of thousands of comments, revision requests, and instructions that I've given to my editors that are now lost forever. I can't get those back.
It's such a shame, right? I'm never going to be able to go back in time, retrieve all those, and then train my AI on that exact taste.
But now with AI, it doesn't forget anything. It has perfect memory. It understands all your comments and it becomes this self -improving model.
Unlike a human, it doesn't forget the feedback you gave 10, 15, 100 edits back. It looks at everything collectively and it knows where it went wrong. And what I found with my AI video editor is that after the fifth or sixth video, and after leaving a ton of feedback on it, right?
It was able to look at that and say, okay, I'm getting some pattern recognition going on here. You keep leaving feedback about how bad the sound design is. You keep leaving feedback on these types of visuals.
You keep pointing to the fact that the sifting is not catching your retakes very well. And then it takes those like three macro problems. It looks into why it's getting it wrong.
And then it makes a permanent fix. And then all my videos after that, don't make that mistake again. It's a beautiful thing.
But again, I'd say this mostly applies to the technical side of things. For me, the goal is to get video edits to a point where from the technical engine room side of stuff, it is 100 % sound every single time with no mistakes.
And when that happens, you can spend 100 % of all of your time on the creative stuff, which again is where you are needed anyways. You have to give it feedback. You have to give it that direction.
And you're going to be making some tweaks here and there according to your preferences, tastes, and likes. And... With those seven rules out of the way, this is exactly why I've spent so much time building an AI video editor myself.
I've taken hundreds of templates, hundreds of assets, rules, principles, tutorials that I used to train my humans on. I've put it into a nice UI that makes it really easy to just take a talking head video. If you are a creator who wants to make talking head content for short or long form content to just drag and drop it, it grabs all your footage, syncs it all up, sifts it all together, transcribes it, storyboards every single thing you're saying verbally, matching it to a visual.
quality checks the whole thing, gives it back to you. You then review it. There's always going to be some things you like or don't like.
And I am now at a point where I don't really have more than two or three rounds of revisions on my video edits until they're perfect. This is a tremendous time and money saver when it comes to getting video content out. Yes, you still have to have something like a Cloud Code subscription.
I'll make some future videos showing you exactly how much usage and time these edits cost. I'll do that a little bit later. But again, if you click the link in the description below, you can join the waiting list for when this goes live on November 18th.
I'm going to close the doors 10 days after, probably reopen them down the road. But whether you grab my AI video editor or you want to make your own, hopefully this video is helpful for you to apply these seven rules that I use to train AI video edits. Heck, transcribe this video, give it to your AI video editor if you like.
But that's it. Rock and roll. Edited entirely by AI.
We'll see you in the next one.
The Hook
The bait, then the rug-pull.
The video you are watching was edited by the AI it describes. Walbeck, who has trained over 100 human editors, says he taught Claude the same rules, templates, and workflows, and walks through the seven that matter.
Frameworks
Named ideas worth stealing.
00:30list
The 7 Rules of Training an AI Video Editor
Give creative direction
Sift scientifically
Good input equals good output
Less is more
Intentional scene cutting
Make it once, use it 10 times
Catch it before I do
The spine of the video, shown as a seven-card motion graphic.
02:15list
Creative brief
References
The feel
Starting visuals
Three parts given to any editor, human or AI, before an edit starts.
06:20list
Five retake types
Repeats
Reopens
Replacements
Rephrases
Exceptions
Categories an AI needs rules for to tell a retake from a keeper.
00:50model
Two engine rooms
Technical engine room is fixed rules and assets. Creative engine room is subjective taste that changes per video.
CTA Breakdown
How they asked for the click.
VERBAL ASK
10:06product
“Click the link in the description to join the waiting list for my AI video editor launching November 18.”
Dropped mid-video after the asset moat argument, then repeated at the close with a ten-day enrollment window.
Add 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.
Add to Preferred SourcesOpens your Google source preferences with us pre-loaded. Tick the box and you're done.
A working video editor walks through the five-step workflow he uses to have Claude edit his reels, ads and VSLs, and why the creative rules matter more than the model.
Two podcasters test whether the polished AI-edited videos going around YouTube can really be reproduced with one prompt, and explain why templates, separate tracks and real data decide the answer.
A full workflow for making an animated video entirely in code, from an empty folder to upload, built around a reusable style skill that gets better with every video.