08:38Jay E | RoboNuggets · Demo
Someone Rebuilt Adobe for Free, and Claude Can Run All of It
ArtCraft cloned seven Adobe apps from scratch in Rust, open-sourced the code, and built them so Claude Code can drive every one.
October 8thThe argument in one line.
Claude Motion builds video as editable code instead of baked pixels, so one word, color, or object can be fixed or swapped without regenerating the whole clip.
Read if. Skip if.
READ IF YOU ARE…
- You already use Claude on a Team or Enterprise plan and want to produce animated promos, explainers, or ad creative without hiring an editor.
- You make short-form or ad content regularly and are tired of regenerating an entire AI video clip just to fix one typo or swap one object.
- You're curious how to connect Claude to cheap third-party image and video models to get realistic shots inside an AI-built animation.
SKIP IF…
- You're not on a Claude Team or Enterprise plan with no near-term plan to upgrade; Motion isn't available on Free, Pro, or Max yet.
- You want fully photorealistic AI video from a single prompt, not animated text, shape, and chart-driven motion graphics.
The full version, fast.
Claude Motion is Anthropic's new beta feature, currently on Team and Enterprise plans, that turns a plain-English brief into an animated video built from code instead of baked pixels. Because every word, color, shape, and timing value is written as code, a typo or wrong price can be fixed in the Tweaks panel, or by clicking an object and leaving a comment, without regenerating the whole clip the way video models like Kling or Veo require. The workflow: align with Claude on a detailed prompt first, paste it into Motion on Opus 5.5 extra-high effort, then edit live. Connecting Claude to a cheap model marketplace like kie.ai adds photorealistic shots behind the animated text, and HyperFrames or a self-built rig cover anyone without Team access.
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 →Where the time goes.
00:00 – 00:50
01 · Cold open: no editing, no Higgsfield
Jay opens by stating every animation in the video was built with Claude Motion, Anthropic's new plain-English-to-code video feature, then introduces himself and the promise of the tutorial.
00:50 – 02:03
02 · What Claude Motion actually is
Defines Motion as a beta Claude desktop feature, reachable via Artifacts, the dashboard/motion buttons, or /motion in any chat, and contrasts it with prompt-based AI video models like Kling or Veo: Motion writes the video as code, so individual elements can change without a full regeneration. Notes it's limited to Team and Enterprise plans for now.
02:03 – 06:20
03 · How to use it: building and editing a 20-second ad
Walks a real example, a coffee-brand ad: align with Claude on a detailed prompt first, paste it into Motion on Opus 5.5 extra-high effort, wait one to four minutes. Explains the bakery-cake-versus-recipe analogy for why code-based editing beats pixel regeneration, then demos the Tweaks panel for word, color, scene-length, and speed edits, plus comment-and-click editing for non-text elements like swapping a cup for iced coffee. Covers sharing, duplicating, exporting to MP4, and how usage limits work.
06:20 – 11:33
04 · How to master it: adding real image and video generation
Shows connecting Claude to kie.ai via the free kie-mcp connector to generate photorealistic shots, GPT Image 2 at about five cents an image, that Motion then animates words and timing around, demoed with a real product photo and an iced-coffee pour clip. Covers the reverse technique, Motion builds a storyboard first and a video model like Seedance renders it in a chosen style such as hand-painted anime. Includes a mid-video pitch for Jay's RoboNuggets community, then two non-Team alternatives: HyperFrames and a self-built local rig, with the tradeoff that Motion renders in Anthropic's cloud while local alternatives are CPU-bound.
11:33 – 12:08
05 · Wrap-up
Recaps Claude Motion in one line, asks for a subscribe, and points to the free PDF prompt guide in the description.
Lines worth screenshotting.
- Claude Motion generates video as code instead of baked pixels, so you can edit one word or one element without regenerating the whole clip.
- Aligning with Claude on a detailed text prompt before opening Motion produces a result far closer to the brief than winging it.
- Claude Motion is currently limited to Team and Enterprise plans, though wider plan availability is a reasonable bet given Anthropic's release pattern.
- The Tweaks panel edits every word, color, and timing value in a scene live as you type, with no re-render needed.
- Clicking an object directly and leaving a comment lets Claude edit a single visual element, like swapping a coffee cup for an iced coffee, without touching anything else in the scene.
- Claude Motion can only draw what it can code: text, shapes, charts, and your own images. It can't generate new photorealistic images or video on its own.
- Connecting Claude to a cheap model marketplace like kie.ai, GPT Image 2 runs about five cents an image, lets Claude generate realistic photos or clips that Motion then animates words and timing around.
- You can reverse the workflow: have Claude Motion build a storyboard first, then pass those frames to a video model like Seedance to render in a specific visual style.
- There's no extra charge for using Motion beyond your plan's normal usage limits, but longer and more complex animations consume more of that shared budget per generation.
- If you're not on Team or Enterprise, HyperFrames, a free open-source HeyGen plugin, and a self-built local editing rig are both viable Claude-driven alternatives to Motion.
- Claude Motion renders in Anthropic's cloud, so it's typically faster than local alternatives like HyperFrames or a self-built rig, which are bottlenecked by your own computer's CPU.
Code-based video beats prompt-and-pray AI video
CODE OVER PIXELS
Claude Motion treats video as editable code rather than finished pixels, which changes how you brief, fix, and layer AI-generated footage into a finished animation.
- Writing video as code instead of pixels means one wrong word, color, or object can be fixed directly, without regenerating the entire clip and paying for it again.
- Aligning on a detailed written plan with Claude before opening Motion, including the brief, audience, and platform, produces output far closer to what you actually wanted.
- The best results come from picking the strongest model and effort setting available, like Opus on extra-high, rather than defaults, especially for a complex multi-scene animation.
- A live editing panel that lists every word, color, and timing value lets you iterate on copy and pacing as fast as you can type, with no render wait.
- Clicking directly on a visual element and describing the change in a comment is how you edit anything that isn't text, like swapping a prop or background detail.
- A tool that only draws text, shapes, charts, and your own images still can't generate new photorealistic footage on its own; that gap has to be filled by a separate model.
- Connecting a code-based animation tool to a cheap third-party image or video model turns flat motion graphics into something with real photographic depth, for cents per asset.
- Reversing the usual order, building the storyboard first and asking a video model to match its style, locks a specific visual look before you spend on full generation.
- Usage limits scale with animation complexity even without a separate line-item charge, so planning before generating protects a shared budget as much as it protects quality.
Terms worth knowing.
- Claude Motion
- A beta Claude feature that turns a plain-English description into an animated video, writing every word, shape, color, and timing value as editable code rather than rendering fixed pixels.
- kie.ai
- A third-party marketplace that offers access to image and video generation models, such as GPT Image 2, at a fraction of the cost of a dedicated monthly AI video subscription.
- MCP (connector)
- A connector that plugs an outside application or service into Claude so it can take real actions, like generating an image through a paid API, directly from a chat.
- HyperFrames
- A free, open-source Claude plugin built by the HeyGen team that teaches Claude to make and edit videos, including a built-in browser studio for scrubbing and adjusting the result.
- Tweaks panel
- Claude Motion's live editing panel that lists every word in a scene, letting you change copy, color, and timing and see the video update as you type.
- Seedance
- A video generation model used here to turn Claude Motion's storyboard frames into fully rendered, consistently styled moving footage.
Things they pointed at.
06:48linkClaude Motion prompt guide (PDF)
07:23toolGPT Image 2
07:52toolkie-mcp connector
10:00toolHyperFrames
10:44toolSeedance
Lines you could clip.
03:05
“AI video models, they're sort of like a bakery that sells you a finished cake.”
03:20
“What Claude Motion gives you is the recipe.”
04:40
“Type Don Roast over the current name, and change the 19 to 22, and the video now updates as you type.”
11:10
“That is Cloud Motion, where you describe a video in plain English and Cloud creates it for you.”
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
Every single animation you're about to see in this video was not edited manually and not generated with Higgs Field. It's all done through Entropic's newest release called Claude Motion, where you describe a video in plain English and Claude builds it as an animation like the examples you're seeing here. And unlike other AI video tools, Claude does all of this using code, which means that you can give comments to the video and Claude can just tweak it without having to charge you for every video generated.
So today I'll show you what Cloud Motion is, how to use it, and how to master it, so that by the end you can automate your own animated promos, explainers, and any video material that you need for your business. And if you're new, my name's Jay. I spent over a decade working with brands you probably know, have been in AI since my master's in data science, and now I'm leading our AI business in one of the largest AI communities globally.
So Cloud Motion is a new feature in the Cloud desktop app where you describe a video in plain English and Cloud creates it for you. It was released in beta on the 8th of October. And to find it, you just click artifacts in the sidebar and at the top, you'll see two new buttons, dashboard and motion.
Or in any chat, you can actually just type slash motion or pick motion from the output menu. When it opens, your chat would be on the left. And the video that Claude is about to create is on the right, where you can also give comments to edit it.
And the big difference of Claude Motion from AI video models like Kling or Veo is how much control you get. So with a video model, you generate the clip from a prompt. And if you don't like it, you generate it again because you can't actually reach in and fix just one part of it.
But Claude Motion writes the video. as code, which means that you can change exactly the part you want, like one word or one element, and everything else stays as it was. So right now, just to note, Cloud Motion is limited to the team and enterprise plans, but knowing Entropic, I'd expect it to be available on more plans soon, so it's a good time to get ahead of it.
And in case you're not yet on team now... Then you can stick around because near the end, I'll also show you an alternative on how to make these videos that you can edit with Claude on any of their plans. Now let's talk about how to use Claude Motion.
Let's say we want a 20 -second ad for a small batch coffee roaster. Now the first thing I would recommend is to align first with Claude on a plan because the last thing you want is a video that is off -brief. So in a normal chat, you just give it your rough idea and you can also add here reference images, your brand's websites, or even a video whose style you want to copy.
Then just do a bit of back and forth on it until it hands you a detailed prompt with the look, the scenes, and their timing. Then we go into motion. So you can click artifacts and then motion.
Or again, you can just type slash motion in that same session. Paste that prompt in. Then for the model and effort, if you want the best results, I would recommend Opus 5 .5 on extra high.
And once you send that prompt, the right sidebar will now show the scenes for you. And Chloe is now going to start creating your video. Now, this usually takes around one to four minutes, depending on the complexity of your video.
But because we agreed with Claude on the plan first, it would now come back much closer to what we had in mind. Now, while Claude is building here, it's actually good to understand what's going on behind the scenes and to also compare it versus a normal AI video model. Because one analogy that you can think about is AI video models, they're sort of like a .
So every frame that they generate, it is baked in as pixels. So for example, if there's a typo on the screen, you can't actually fix just that one bit and you end up ordering a whole new cake or in this case, generating an entirely new video, which can cost a lot. Now what Claude Motion gives you is the recipe.
So every word, every color, movement and timing is actually written down as code, which dictates what you see in the final video. So if you want to fix the typo, you can just have Claude change that one line and the rest of it comes out identical. However, because it is a recipe, it can only use ingredients that it can draw, which means text, shapes, charts, and your own images.
And it can't generate images or videos the same way you would expect other generative AI models. But of course, there's a workaround here. And near the end, I'll also show how you do that without having to have another monthly subscription to another tool like Higgs Field.
But now let's go back to Cloud Motion because once it's finished, you can see we can now play the animation here on the second panel. And in the bottom right, you can see there's a button called Tweaks. Now, let's say yourself or the client tells you the name of the coffee brand is actually Dawn Roast.
and a bag is $22 now instead of $19. Well, what you can do is open tweaks where every word in the current scene is listed, type Don Roast over the current name, and change the 19 to 22, and the video now updates as you type. You can also change each scene's length, the speed, and also the number of scenes.
So that is the granularity that you can actually tweak using Cloud Motion. Now, that tweak feature that covers the words as well as the timing. But for anything without words on it, like let's say you want to change this element, the coffee cup, what you need to do is to click the little comment button at the top, then click the cup right there in the preview, and type something, a comment like, make it an iced coffee with, let's say, ice cubes.
Claude then now works on that one piece only, and when it's done, the cup is iced while the beans, the roast, and the name stay exactly as they were. So the implication of this is that you can fix one object in a scene you already like without changing the rest of it because those already look okay. Now, when you're happy with a video, you can actually click share and send it to your team.
And that way they too can duplicate this project and have their cloud apply edits to it. The same way that you would with a Google doc or a Microsoft SharePoint document, for example. And to get the actual video file, you can just click Export, choose MP4, then pick the resolution, frame rate, and quality, and it saves straight to your computer.
Or you can actually just ask Claude to do that, and that would result to the same thing. And with that, you just created a video without ever opening an editing app. Now, in terms of how much it costs, there's actually no extra charge for motion.
So if your plan already includes it, then you already have it. But of course, every animation does count towards your plan's usage limits. And obviously, the longer and more complex the animation, the more of your limit it uses compared to a normal prompt.
Now, Atropic hasn't said how many videos that works out to. So do keep an eye on your usage while you're testing. And as a general advice, at least for videos, it is much better to agree on a plan with Claude first before you just let it go forth and create that complex video for you.
So now let's take it to another level because everything we've made so far is just text, shapes, and charts because that's what Claude Motion can draw on its own. Now, where it gets really interesting is when you give Claude image and video generation, which basically means that you can connect it to AI models that can make pictures and realistic footage.
Then what happens is that Claude can make the real looking shots while motion still handles the words, the timing and the movement on top. And just to give you a couple of examples, here are some demos of what that can give you. And as we dive into this more complex topic, what I did is just create a PDF guide where all of the prompts and the resources that I'm about to talk about, you can just grab that, send that to Claude so that you can immediately teach Claude all the best practices that we are about to cover.
So the link for that is just in the description if that is something that you need. Now, when connecting Claude to image and video models, a lot of people immediately think that you need Higgs field for this, which is not surprising because Higgs fields marketing is sensational, but there are actually tons of other providers that offer the same models for cheap.
For example, I use Key .ai. We're not sponsored by them, but we just use them quite a lot, especially in our work. And to give an example, they have GPD Image 2, which is one of the best image models out there for about five US cents an image, which is much cheaper versus having to subscribe on any monthly plan from Higgsfield.
To connect it, you can just use a free community connector called the Key MCP. And an MCP, that's basically just a connector that plugs an outside application into Cloud. And to do that, you can get started using this prompt on.
screen and just send that to Claude, and it sets everything up for you, including instructions on how to connect your account using an API key, which is sort of like a password that lets Claude access your account in Key .ai. So once it's connected, what you can now do is to have Claude generate the images or the video clips first, and then ask it to build the animation around them.
For our sample ad, you can now see that's a real looking photo of the rosary behind the opening scene. And we can also have a short clip of beans pouring behind the price. And all the texts that you're seeing right now are still animated by motion on top.
So we can still freely change them. So now our video gets quality shots and because the words still come from Motion, you can change any of them in Tweaks without having to remake the footage again. Or another technique is that you can do the reverse.
So you can have Cloud Motion make the scenes first as a sort of storyboard. And then you can pass those scenes to a video model like Seedance, for example. And the model will now reference those storyboard scenes and create them in the exact style that you like.
For example, these ones are in a soft hand -painted anime style, a bit like a Studio Ghibli film. And Seedance turned each scene that Cloud Motion did into a moving shot. And that way you're able to use Cloud Motion to unlock more video ideas and options for you.
And by the way, if you want to learn how to build and sell AI systems that businesses actually pay for, then that's pretty much all we do over at the RoboNuggets community, where not only do you get access to the Cloud Living Masterclass, which we update every week and takes you from zero to mastery with the latest on AI, but you also get access to our Agents as a Service course, which walks you through how to actually get paid for all these AI skills that you are learning.
You also get to be part of a genuinely great community of AI builders. In fact, you can see just some of the recent wins our members are getting from the program right here.
So if you want to start earning from AI, then check that just in the pinned comment below. Now back to the video. However, if you're not on the team or enterprise plan yet, and you don't have CloudMotion, don't worry because you of course have some alternatives.
And in fact, you actually have two. The first one is Hyperframes. It's a popular option because it's a free and open source tool by the team from Haygen.
And to use it, you just send Claude the prompt on screen, which installs the HyperFrames plugin for you, which is basically an add on that teaches Claude how to make videos with it. And from then on, when you ask Claude to make or edit videos using HyperFrames, it can actually drive that application for you. it has a built -in studio that can open in your browser, which is another editor where you can scrub through the video, change things yourself, pretty much similar to Cloud Motion.
Now, the second alternative is that you can actually build your own rig, which essentially is just your own little editing app that runs on your computer. You can actually get started with a prompt on screen. And if you use Opus 5 .5, you can get pretty far with a self -built tool.
Personally, this is what I do most of the time. And I use Rubric Editor, which is part of the tools that we build for our community. And on top of the motion graphics, it also lets me edit around raw footage of a person talking.
So it gives me the exact cuts for each video, edits out the pauses, apart from just the normal animations and image and video generations that we illustrated a while ago. This, in my view, where you have a custom application that is customized for your needs, I think is the most effective way for you to edit videos and animations with Cloud.
Just note that versus Cloud Motion, these tools that I mentioned, Hyperframes and your own local rig, there is one clear difference, and it's that Cloud Motion runs in Antropix Cloud. So when it's editing and rendering footage, it is usually much faster because it's essentially using Antropix computers. If you're using a local rig or Hyperframes, that runs on your computer.
So depending on your CPU, that can render much slower. So there you go. That is Cloud Motion, where you describe a video in plain English and Cloud creates it for you.
usual thanks for watching till the end and if that was useful consider tapping subscribe below because that just helps us a lot to put out more educational stuff like this which helps you use ai better for your business and like i mentioned you can just grab that pdf guide down below if you need all of the resources that i mentioned in this one i'll see you next time cheers
The Hook
The bait, then the rug-pull.
Every animation in this video, Jay says, came from one plain-English description and Anthropic's new Claude Motion feature, not a timeline, not Higgsfield. The pitch: because Motion writes video as code instead of baking pixels, you can hand it a comment and have it fix one word or one object without touching anything else.
Named ideas worth stealing.
02:51list
The Claude Motion workflow
- Align with Claude on a detailed written plan first, brief, audience, and any reference images or sites
- Paste the finished prompt into Motion (Artifacts to Motion, or /motion) on the strongest model and effort you have
- Use the Tweaks panel for word, color, and timing edits as you type
- Click an object and leave a comment to edit anything without words, like a prop or background detail
The four-step loop Jay demos end to end on a coffee-brand ad, from written brief to exported MP4.
Steal forany tutorial that needs a clean before/after software demo structure
How they asked for the click.
VERBAL ASK
09:05product
“if you want to learn how to build and sell AI systems that businesses actually pay for, then that's pretty much all we do over at the RoboNuggets community”
A soft mid-video pitch for his own paid community, framed as a bonus for sticking around, then a clean return to the tutorial. A second, lighter subscribe ask closes the video.
MENTIONED ON CAMERA
FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
AFFILIATECommission earned if you click.
Visual structure at a glance.
