Big Idea

The argument in one line.

Remotion turns video editing into a prompting task, but reliable output needs four extras beyond the basic tool: word-level timestamps, a rough-cut pass, a beat detector, and a pre-built asset and template bank.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…

You already have raw footage and want an AI coding tool like Claude to cut, caption, and add motion graphics instead of hand-editing it.
You're comfortable prompting an AI coding assistant and want to extend it into a working video-editing tool.
You want the architecture behind a repeatable AI editing pipeline, not just a walkthrough of exporting one video.

SKIP IF…

You're looking for a traditional timeline-editing tutorial in a tool like Premiere or DaVinci Resolve; this is prompt-driven, not drag-and-drop.
You have no interest in connecting outside tools (Deepgram, ElevenLabs, Brave Image Search) to get full value from the workflow.

TL;DR

The full version, fast.

Remotion lets an AI coding assistant like Claude write and edit video through plain-language prompts: install it, describe a basic scene, and it generates the motion graphics. The creator installs Remotion, builds a simple hello-world video, tours Remotion Studio, and feeds it real recorded footage to cut. He then breaks down what separates a rough AI edit from a polished one: a transcript with word-level timestamps, an automated rough cut, a taste-sensitive beat detector that decides where graphics belong, outside tools like Deepgram and ElevenLabs, a described asset bank instead of AI-generated b-roll, and a personal template library that does most of the design work upfront.

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.

Create a free account →

Chapters

Where the time goes.

00:00 – 00:34

01 · Intro

States the promise: describe what you want and Remotion's AI handles the edit. Previews that the video covers install, basic use, and the upgrades that take output from okay to great.

00:34 – 01:22

02 · Installing Remotion

Pick any AI coding tool (Claude, ChatGPT, Grok, Hermes), create a dedicated project folder, copy the install command from remotion.dev, and have the AI tool run it.

01:22 – 02:50

03 · First Basic Video

Prompts a plain white 'hello world' video, then adds an animated waving-hand emoji with a follow-up prompt. Flags that model quality matters, even on simple prompts.

02:50 – 04:53

04 · Remotion Studio Tour

Tours the built-in Studio: timeline, project list, and a props panel for tweaking exposed text/colors. Says it's not a real editor and recommends prompting Claude or using CapCut for actual timing edits; exporting happens through Studio's Render panel.

04:53 – 08:21

05 · Editing Real Footage

Hands a recorded clip to the AI tool with one prompt: transcribe with Deepgram, cut bad takes and silence. Reviews the rough cut in Studio, notes the default cut isn't aggressive enough, then adds a title-card overlay with one more prompt.

08:21 – 09:57

06 · Word-Level Timestamps

Explains why a transcript needs per-word timing, not just the words themselves: it's what lets the AI find silences, isolate bad takes, and know exactly when a graphic should appear.

09:57 – 11:14

07 · Beat Detection

Introduces the 'beat detector,' the step where AI decides which specific moments deserve a title card, overlay, or emphasis. Calls it the hardest, most taste-sensitive part to automate well.

11:14 – 18:23

08 · Tools, Assets & Templates

Lists the remaining pieces: outside tools (image search, voiceover, reusable skills), a pre-approved asset bank described in a JSON dictionary so AI can match by meaning, and a personal template library that does most of the design work upfront.

Atomic Insights

Lines worth screenshotting.

A transcript with word-level timestamps, not just a block of text, is what lets an AI editor know exactly when to cut a silence, a bad take, or drop in a graphic.
AI-generated images of real people are expensive and still aren't the real person, so for recognizable figures it's cheaper and more accurate to pull an existing photo through image search instead.
Letting AI generate every icon, logo, and piece of b-roll from scratch produces inconsistent results, so a pre-approved asset bank cuts failure points and keeps a video on-brand.
A short filename like broll_002.mp4 doesn't give an AI enough context to match the right clip to the right moment; a written description in a JSON dictionary does.
Finding which moments in a script deserve a title card, overlay, or emphasis is the hardest part to automate well, because it's taste-sensitive, not a mechanical rule.
Treating editing as a rough cut first, graphics second, two-step process keeps bad takes and dead air out before any motion graphics get layered on top.
A library of pre-built templates lets AI do roughly 20% of the design work (fitting content into a template) instead of 100% (inventing a design from nothing), which raises quality, speed, and cost all at once.
Codifying a repeatable editing process as a named, reusable 'skill' is what turns a one-off AI prompt into a workflow that produces consistent quality every time.
The built-in Remotion Studio can preview and render a project, but it isn't a real timeline editor, so actual edits are better made through prompts or a dedicated tool like CapCut.
A single well-structured prompt, backed by a pre-built system of templates and assets, can replace a multi-round back-and-forth editing session entirely.

Takeaway

The four pieces a reliable AI edit needs

AI EDITING SYSTEM

Remotion turns editing into prompting, but consistently good output needs four extra pieces: word-level timestamps, a rough cut, a taste-sensitive beat detector, and a described bank of assets and templates.

02Installing Remotion

Remotion installs through whatever AI coding tool you already use (Claude, ChatGPT, Grok, or similar) by pasting its install command and letting the assistant run it inside a dedicated project folder.
You don't need to learn Remotion's own syntax to get started; the AI coding assistant handles the install and the first prompts for you.

03First Basic Video

A first Remotion video can come from a single plain-language prompt describing the background, the text, and how it should animate in and out.
Model choice matters even for simple prompts; cheaper models produce noticeably rougher results than frontier models on the same prompt.
Follow-up prompts, like adding an animated emoji, can be layered onto an existing project instead of starting the whole thing over.

04Remotion Studio Tour

Remotion Studio previews a project and lets you tweak exposed properties like text and color, but it's not built for dragging clips or adjusting timing by hand.
Rendering out of Remotion happens through Studio's Render panel, which exports the final video file to your machine.
For real timeline edits, prompting the AI tool directly, or using a conventional editor like CapCut, beats fighting Studio's limited editing UI.

05Editing Real Footage

A recorded clip gets handed to the AI tool with a single prompt: transcribe it, then cut bad takes and silence, including dead space at the start and end.
A default AI rough cut tends to be too conservative, leaving extra dead space around cuts, which means the underlying detection logic needs tuning to get aggressive enough.
Turning a one-off editing prompt into a repeatable, reliable process means codifying it as a named 'skill' rather than re-explaining the same instructions every time.

06Word-Level Timestamps

A standard transcript only tells you what was said, not exactly when each word starts and ends, which isn't precise enough to place cuts or graphics correctly.
Word-level timing is what makes it possible to find silences, isolate bad takes, and know the exact moment a motion graphic should appear.
This precision is the specific reason the creator chooses a transcription tool with word-level timestamps over a plain speech-to-text tool.

07Beat Detection

A 'beat detector' is the step that decides which moments in a transcript deserve a title card, overlay, or visual emphasis, rather than treating every line the same.
Deciding which points deserve emphasis is described as the hardest part to automate well, because it depends on taste rather than a fixed rule.
Building a beat detector that consistently makes good calls takes a long, iterative tuning process, not a one-shot prompt.

08Tools, Assets & Templates

Remotion alone can't generate images, voiceover, or complex effects; those come from connecting outside tools like an image search API, a voiceover API, and reusable 'skills.'
For real, recognizable people, pulling an existing photo through image search beats generating a fake likeness: it's cheaper and avoids an inaccurate AI-rendered face.
A pre-approved bank of b-roll, icons, logos, and music, each described in a short JSON entry, lets the AI match assets by meaning instead of guessing from a short filename.
A personal template library is the highest-leverage asset: letting AI fit content into existing, pre-approved designs is faster, cheaper, and more consistent than asking it to invent a design from scratch every time.

Glossary

Terms worth knowing.

Remotion: A code-based video creation framework that generates and edits video through plain-language prompts to an AI coding assistant, instead of a manual timeline.
Remotion Studio: Remotion's built-in preview and render interface. Useful for previewing a project and exporting video, but not a full drag-and-drop editing tool.
Word-level timestamps: A transcript format that marks the exact start and end time of every individual word, not just full sentences, so edits can be placed precisely.
Rough cut: A first editing pass that removes dead air and obviously bad takes from raw footage before any stylistic edits are made.
Beat detector: The step in an AI editing pipeline that decides which specific moments in a script are worth a title card, graphic, or emphasis.
Asset dictionary: A JSON file that stores a short written description next to each piece of b-roll, icon, or music track so an AI can match assets by meaning instead of a short filename.
Skill (AI workflow): A saved, reusable instruction set that lets an AI coding tool repeat a specific multi-step task, like a full video edit, from a single prompt.

Resources

Things they pointed at.

00:33toolClaude

00:33toolChatGPT

00:33toolGrok

00:33toolHermes

00:56toolRemotion.dev ↗

05:43toolDeepgram

12:31toolElevenLabs

11:42toolBrave Image Search

11:22toolHiggsfield

11:40toolNano Banana 2

11:40toolGPT Image 2.5

17:27productApplied AI Mastermind ↗

Quotables

Lines you could clip.

00:20

“honestly, using it feels like a cheat code once you know how it works”

tight hook line that sets the thesis fast→ TikTok hook↗ Tweet quote

10:30

“this is actually one of the hardest bits to train AI for because it's very, you know, taste sensitive”

names the real difficulty, relatable to anyone building with AI→ IG reel cold open↗ Tweet quote

16:44

“we are doing 80% of the work first and AI is just doing the 20% of the work to make it fit the video”

clean, quotable rule of thumb→ newsletter pull-quote↗ Tweet quote

14:45

“a filename is too short... what you need instead is something I call a dictionary”

names a concrete fix with a memorable label→ TikTok hook↗ 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.

metaphoranalogy

Remotion is a super powerful tool that allows you to use AI to automatically edit videos and create motion graphics. All you have to do is describe what you want and it will do it for you. And honestly, using it feels like a cheat code once you know how it works.

So in this video, I'm going to break down all of Remotion for you for absolute beginners. I'm going to show you how to install it, how to use it, and then if you stick around to later on in the video, I'm going to show you the upgrades which allow you to turn your motion graphics in Remotion from okay to really great. Okay, so the first thing you're going to need is an AI tool of some kind.

It doesn't really matter which one you use. I'm going to use Claude, but you can use Claude, you can use ChatGPT, you can use Grot, you can use Hermes. It doesn't really matter.

The next thing you're going to want to do is make sure that you're working within a Remotion folder. So you want to create a folder on your computer, just like you create any other folder. And within Claude or whatever tool you're using, just make sure you're working within that folder here.

Then once we are within that folder, there'll be nothing in that folder to start with. Then we need to install Remotion. To do that, you come to Remotion .dev and you see a website that looks like this.

If we click this button here, it's going to copy this to the clipboard. Then we can come to Claude and we can say, please can you install this tool locally using the command and then just paste in what we copied and go ahead and run that. Now I've already done this, so it's going to come back and say Remotion's already there.

But you run this command, it might ask you for a few things, but after you've done that, you will have Remotion installed. It's that easy. Cool.

So it's telling me I've already got it installed. Let's just get rid of this. And in fact, let's create a new chat.

So let's create a really basic video to start with. I'm going to paste in this prompt here. Create me a re -motion project where there is a white background and the words hello world come onto the screen.

And then they are then replaced by the words. This is a re -motion test. Export this video for me.

So it's going to take maybe a minute or two minutes for this to generate. And we're going to see what happens. whilst it's generating, let me just say here, that you want to be using the best model that you can.

Now, this is a very, very simple project, so I'm only using Sonnet 5, but you will notice a difference between using an OK model like Sonnet 5 or using something like Fable 5 .1, which is one of the best ones at the moment, or GPT -6 Astra. Later on, I'm going to show you some little tricks that you can do to get a great result, even when you're not using one of those really expensive models, though.

Okay, so that is completed. Let's have a little look at it. Okay.

Hello world. And then it says, this is a re -motion test. There we go.

It's not blowing anyone's socks off, but there we have it. Now let's make a little change to this. Like I said before, you can prompt anything that you like and it will do that.

So I'm going to paste in again, something fairly basic. I'm going to say on the hello world screen, add a big emoji of a waving hand and animate it in. Let's give it a second to do this.

Okay. That is finished generating. Let's have a look to see what it's done.

Hello world with a waving emoji above it. Yeah, good stuff. Now, as I said, we can make any changes that we like to this.

We can change the colors, we can change the text, we can add new screens, we can add in fancier animations, literally anything that you want. But there's another side of Remotion which is really important to know about, and that is the studio. So if I say, please open this project in the studio, it's going to go ahead and open up in the studio.

And if you've ever used anything like CapCut before or DaVinci Resolve or maybe Premiere Pro, you're going to sort of recognize the interface that we're going to see. So this here is the Remotion Studio. Now we've got the timeline along the bottom and we can drag this playhead and we can go through the Remotion project.

You can see your different Remotion projects over here on the left -hand side. You can also open the properties tab on the right -hand side. And if we select something at the bottom, such as let's go like the hello text, we can see that we can actually move this around on the screen.

So we can do a little bit of editing within the studio. And when we don't have any of the elements selected, we do have this panel on the right -hand side where it breaks it down into props, which is where we can change aspects of it. So I can change the hello world to your what's up.

Okay, we'll spell that wrong, but there you go. You can see we can change the text and we can change the emoji. We can change the colors as well here.

So we can change the accent color to like a pink color. And then if we go through to the next page. That color has now changed.

Let's change it again to maybe like a yellow or a green. There we go. Now, you can make edits within the Hyperframe Studio.

I really would not recommend it, though. It's just not very good. There are much better alternatives if you want to be...

tweaking your edits that you make with AI. Hyperframes is better, or even using something like CapCut alongside of Remotion is actually much better if you want to make tweaks. This is not a full editor, the studio, where you can drag around things and change the timing of things.

Although it looks like that, it's terrible for that functionality. So if you do want to make any tweaks to it, then I'd highly recommend doing that with inside of Claude or whatever tool you're using. rather than inside the studio.

The studio can be good for previewing what you're creating before you export it. And talking about exporting it, we can just click on render down here, set the settings that you want, and then go render video. And you can see your render happening over here on the right hand side, and it's rendered.

You can click on this button here to see exactly where it has saved it on your PC. Let's have a look at a little bit more of an interesting use case. So editing a video that I've recorded.

So I'm going to go and find that video which I've recorded. I'm going to drag it into Claude. and I'm going to paste in this prompt here.

I'm going to say, take this video, transcribe it with Deepgram, and then make cuts to the video in re -motion to remove the bad takes and silences for me, include any of the start and the end, and then show it to me in studio. Okay, so let me show you this video whilst it's doing that. It's a really short video, and it's just me.

I've got some silence at the end, wiping my nose, and I'm talking. i mess up a little bit as well so i have to do like a retake and then they've got some silence at the end as well so essentially we're doing like a rough cut here we are passing in a something that we have filmed and getting it to take out of those blank spaces and those bad takes Now in that prompt, I mentioned a tool called Deepgram.

Deepgram is a tool that we use to transcribe audio. I'm not affiliated with them in any way, but they're very, very good. When you sign up, they give you $200 worth of credit.

And in like a year of using it, I've used $5. So I'd highly recommend using Deepgram. And we're going to talk about...

why transcription is so important in a later stage. Okay, that's completed. And we are now inside of the studio.

And we can see it's taken away that blank space at the beginning of me wiping my nose. And it's got me there. And then it's also cut out a bad take here.

Now, I'm using like a vanilla re -motion setup. I'm not using using my normal setup. I'm gonna show you a few things later about how you can improve it.

But one thing, for instance, here, which isn't great, is that it's not as aggressive as it should be in terms of the cuts. So if we look at this cut between, you know, where it's cutting out a bad take. You need...

You need to have a good bank of assets. It's not perfect, right? It has found that bad take and it has cut it out.

However, it's not as aggressive as we would like. We can see down here that there's like a little bit too much dead space at the bottom. And the way you make sure this never happens again is you create a skill around it.

But let's add some sort of motion graphics, some sort of, you know, something within this video. So I'm going to go over to Claude and I am going to say, now add in at the start a title card that says the title of the video while I am talking in the first few seconds. Make it have a white background, purple text.

and nice animations in and out. That's now done. We are here within Remotion Studio.

Let's have a look. When it comes to getting good results with Remotion, you need four things. There we go.

We can see the overlay. Now, it's not the words that I would have chosen. However, we're not using the best model out there.

We didn't give it any extra guidance to say what the title should be. And I think the actual title of the file is called the full Remotion thing. So it's no great surprise that it's chosen that title.

Again, you can go back and forth, make whatever changes you want, make it look better, add in whatever you want to make it however you want it to look. Now, those are the very basics of Remotion. You can simply prompt it to make whatever changes or create whatever motion graphics or overlays that you want it to create.

But what we've created so far isn't amazing, right? What I'm going to show you are the three or four different things that you're going to need to do in order to create much, much better looking motion graphics and video edits using Remotion. Once you put all of these together, then you create a system which is really super powerful.

To show you what's capable when you put together all of the things that I'm going to show you, I'm going to edit this video just in one single prompt. I've got a skill called full edit re -motion. And that's all I'm going to say.

I'm not going to say anything else. And it's going to go, it's going to edit that video. in just one single prompt.

And here you go, this is the output from that single prompt. And as you can see, it is much better than what we had created before. And we didn't have to do any back and forth in terms of you know, describing what we wanted.

It just did it for us in one single prompt. Okay, so the first thing you need to understand in order to create much better motion graphics is the idea of a transcript with word level timestamps. When you normally get a transcript, you just get all of the words that are spoken within a video.

However, what we need is we need a transcript, but also we need to know when every single word is said. This is because when we make edits, like if we want to add in subtitles or if we want to add in motion graphics, we need to know exactly when that particular thing is mentioned. And unless we know when each word is said, we don't know when it's mentioned.

So these word level timestamps are super important when we want to find the silences, find the bad takes, when we want to know when motion graphics should appear, when we want to know when anything should happen. And that's the reason that I'm using Deepgram, is that Deepgram allows us to get this transcript with word level timestamps.

That's a really, really important thing. Then step number two of any process when it comes to editing a video, if you are giving it an existing video, if you're creating something from scratch, not from an existing video, then you don't need to worry too much about this. The next thing we should do is run a rough cut, right?

This is where we take out those bad takes and those silences. using the transcription. That's all the data that we're using here is the data from the transcription with those word level timestamps.

If you've recorded anything before and you're anything like me, then you probably have got like a lot of blank spaces or a lot of stumbles. A lot of times you mess up and this just gets rid of it in one take, right? You just say, get rid of the bad takes.

And it uses that transcript to identify the bad takes, identify the silences and just cut them out immediately. The next thing is really important to have when you're editing anything with Remotion is an idea of a beat detector. What is a beat detector?

Well, it sounds like some sort of musical thing, right? What it is, is that when there is a narration or a script. we need to identify okay which part of the scripts should there should be a motion graphic for so sometimes it's pretty easy like if i'm mentioning a list of items then those items should probably appear on the screen or if i'm mentioning a really important point then maybe i want that point you know like emphasized along the bottom of the screen i call this the beat detection part because what you're doing is you're trying to find out which parts of the video are worth having extra edits or extra emphasis in right you don't want the word and emphasized usually right but you may want bullet points emphasized and this beat detection stage is where we go through the transcript or ai goes through the transcript and determines which parts should have edits now this is actually one of the hardest bits to train ai for because uh it's very you know taste sensitive right like it's kind of hard for it to know i spent a long long time building my beat detector

You'll have to go through and just play around with it and see what works and what doesn't work. If you want my version and the version which allows you to just do a full edit, then you can join the Applied AI Mastermind. Add all of my assets.

Everything that I'm showing you here is going to be inside of the Mastermind, including that skill which can edit videos just in one single prompt. That's a really important part, being able to detect which parts of the video should have motion graphics or overlays or something happening on screen. So the next part of creating better motion graphics with Remotion.

is to give it more tools. So Remotion is kind of smart. However, it can't do certain things without other tools.

For example, if we wanted to create a video about, I don't know, Napoleon, right? AI by itself couldn't create an image of Napoleon. However, if we connected it to like an AI image model, like Nano Banana 2, like GPT image 2 .5, then it could create that image.

So we can connect it to tools like that to create AI images and AI videos. But if we actually wanted Napoleon, then we probably wouldn't want AI to generate a photo of it. Napoleon's a bad idea, actually, because he is a historical figure.

But let's say we want to create something to do with Jeff Bezos, right? We want to create a video which is talking about Amazon, and we want an image of Jeff Bezos. it wouldn't be that good to go and create the image.

It would also be relatively expensive as well. What we want to do is actually just get that image from Google Images. And the way to do that is with a tool called Brave Image Search, which allows you to go out there, essentially get Google Images and bring them into your Remotion project.

Then there are tools out there as well, like 11 Labs you can connect to, so it can create voiceover for you. So if you've got a video that you want to create, but you don't want to speak on it personally, then you just get 11 Labs 2, but you've got to connect 11 Labs to Remotion so that when you're editing something, it can go and create that voiceover and then bring it into your Remotion project.

And then another tool you want to give it are skills. I talked a little bit about skills earlier. So my beat detector skill is a skill, but there are other skills out there as well.

There are skills like rotoscoping skills, which will like do this, right? So text appears behind me and between me and the background, that's rotoscoping. Or allow you to have like a little bit more smoother animations.

There are skills out there for those things as well. So they're kind of like plugins to enhance Remotion itself. Then the next thing is something which is really skipped over by a lot of people.

And that is... that you need to give Remotion assets. And that is because even if you are connecting to Google Images via Brave Search, even if you are connecting to Higgs field or to Kai AI or something that can create you AI images and AI videos, you don't want to be using these all the time.

Because if you are creating a motion graphic or you're creating a video which you want to have lots of icons in it, lots of logos in it, then you don't want to be using AI to generate all of those because besides anything else, it'd be expensive to create 200 of these for every video. Also, AI messes up sometimes, right?

And if we want to create a pipeline which does things correct pretty much the first time, then what we want to do is actually give it a bank of assets, assets such as B -roll that you've pre -approved, right? So if we're talking about AI, I talk about AI a lot. and I want to talk about data centers, then I've got a bank of videos of data centers.

So AI knows, ah, when Matt is talking about data centers, we've already got a video which Matt has pre -approved, which is good. for data centers, let's use that. And the same with icons and logos.

AI is pretty bad at creating icons and logos. So I've given it a load of icons, a load of logos, which it can use when I'm talking about something and it goes, oh, okay, Matt's talking about music or images or about databases. We can plug in an icon that we've already got within a database.

So having this bank of assets really does make a big improvement to your emotion gain because essentially you're relying upon AI less, which means there's less failure points and it can create a better outcome first time. Also, sound effects and music are a really important thing to have because AI is pretty bad at creating these.

Now, an important thing to note here is how does it know which thing to select, right? I've got a bank of files on my computer and they've all got file names, but a file name is pretty short. I can't include...

that much information just within the file name right if i've got a bit of b -roll like a 10 second video which shows several different things i can't describe everything in that title you know i can't describe what it shows or when it should be used or the colors of it or anything like that so what you need instead is something which i call a dictionary what this does is essentially um pre -defines every single asset right every bit of B -roll, every icon, every bit of music, every sound effect, it will describe that sound effect or that bit of music.

So it can say, okay, this bit of music here is inspirational and it's got an orchestra in it and it's got strings in it. So that rather than AI trying to figure out what it is just by a short file name, it can look at this file here, which has descriptions of bits of B -roll or bits of music or anything, and it can match it up via that description.

This is really crucial when you are using assets and linking assets with your emotion to create great projects. And I've saved the best to last. One of the most important things to have and probably my most valuable asset that I use the most in my business when creating videos is my template library.

As this sounds, this is a bank of templates which I can use time and time again. And this is working off the similar sort of thing that I've just talked about where AI is pretty good at a lot of things. However, if we want to make something good, make it to the quality that we like make it in line with like our brand and our colors and everything like that um then what we want to do is create a bunch of templates essentially we are doing 80 of the work first and ai is just doing the 20 of the work to make it fit the video this here is my bank of uh templates that i've got and if you've seen any of my other videos then you probably recognize some of these because i use these a lot right so when i am getting ai to edit a video it is looking at these templates using the same dictionary method that i mentioned before it's seeing okay what is the best template for any specific beat that it also has found before and it's selecting that template and then it's going to change things right so if we select this template here with these two icons rotoscope behind me then it might say okay actually we only need one icon or we need to change the icons within

those two rectangles so that it better represents what Matt is talking about. But we are doing 80 % of the work upfront, so the AI only has to do the easy bit of the work, which dramatically improves the quality of your outputs, the speed to get a good output, and also makes it cheaper because AI has to do less of the thinking.

So if you're going to take away anything from this video, I would highly, highly recommend using templates, creating your own templates. If you want access to all of my templates, then you can join the mastermind and you can get access to all of them there. Use templates and assets.

And using these templates and these assets that I've got is exactly how it was able to create that edit for me in just one single prompt earlier, because I already knew the different assets that I had available and the different templates which I like. And all it had to do was figure out which ones to use and what to swap in and out.

So there we go. Here we have covered absolutely everything to do with Remotion from installing it to running a basic project and creating a basic video with Remotion to the most important things to connect up to Remotion to get it to work properly. And I'll show you what you can do when you connect all of those things and how you can build something which will edit an entire video for you just within one single prompt.

If you take everything from this video, you will be able to build your own video pipeline where you can get AI to completely edit videos for you in the way that you've seen in this video. However, if you want to save yourself 30 or 40 hours, then you can just use my process and you can get that down in the mastermind, which is linked down below.

That's it from this video. I hope you have found it useful. If you have found it useful, then go ahead and watch this video next.

As if you'd like this one, I think you'll like this one too. That's it for me. I'll see you in that next one.

Bye -bye.

The Hook

The bait, then the rug-pull.

Matt Penny opens with the pitch: Remotion lets you describe a video edit in plain language and have AI build it, no manual timeline required. He walks through installing it and building a first basic clip, then spends the second half on the four things that separate an amateur AI edit from a reliable, polished one.

Frameworks

Named ideas worth stealing.

07:10list

The Four Things Remotion Needs For Better Output

Word-level timestamps
A rough cut
A beat detector
Tools, assets & templates

The four-part system the creator says separates a mediocre one-shot AI edit from a reliably polished one: precise word timing to place every edit, an automatic rough-cut pass, a taste-sensitive beat detector that decides where graphics belong, and external tools plus a described asset/template bank.

Steal forany AI-assisted content pipeline, not just Remotion — the same four pieces apply to building other automated editing or content-generation workflows

11:40concept

Search, Don't Generate

For real or recognizable people, pull an existing photo via image search rather than generating a fake likeness with AI. Cheaper, more accurate, and avoids rendering an inaccurate AI face for a real person.

Steal forany AI content pipeline that needs real people, logos, or branded assets

CTA Breakdown

How they asked for the click.

VERBAL ASK

18:03product

“If you want to save yourself 30 or 40 hours, then you can just use my process and you can get that down in the mastermind.”

A single soft closing pitch after delivering the full framework for free in the video; links the Applied AI Mastermind in the description rather than interrupting the tutorial. A lighter version of the same pitch also appears mid-video around the beat-detection section.

MENTIONED ON CAMERA

00:56toolRemotion.dev ↗

17:27productApplied AI Mastermind ↗

FROM THE DESCRIPTION

PRIMARY CTAWhere the creator wants you to go next.