Open Design Motion is the new motion-video feature in Open Design, the free, open-source AI design workspace: describe a video in one prompt and your AI agent builds an editable animation you can export as an MP4.
📺 Watch: Open Design Motion: FREE AI Video Editing With ONE Prompt!
🔥 Get the Agent OS as a free bonus: AI Profit Boardroom members get the full Agent OS zip, prompt libraries, daily tutorials and weekly live coaching calls. → Get inside
No timeline. No keyframes. No editing software you have to learn first. You type what you want, the agent writes the animation, and you get a video you can preview, tweak and post.
I plugged it into my own setup this week and made a trailer from a single prompt. Below I'll show you what it is, how it works, how to run it for free, and how it stacks up against Claude Motion.
What is Open Design Motion?
Open Design is a project from nexu-io/open-design. The repo calls it "The 1st Collaborative AI Design Agent Workspace." It's local-first and open source under Apache-2.0 (some bundled skills and templates keep their own licences), and it sits at roughly 102K GitHub stars and 11.7K forks.
You already could use it to build prototypes, websites, apps, dashboards, presentations and images with your team and your AI agents. Now it does video too.
That's the v0.25.0 release, called "OpenDesign Motion," which shipped on 15 October 2026. Per the release notes, it's pitched as "The Open-Source Alternative to Claude Motion." It came together from 53 pull requests by 16 contributors over 21 days.
The idea is simple. You turn ideas into videos. You get editable animations, not a locked file. And you export an MP4 when you're happy.
If you want to learn how to plug tools like this into a real workflow that saves you hours every week, that's exactly what we do inside the AI Profit Boardroom for $69/mo.
How it works: HyperFrames turns code into video
Here's the part most people miss. Open Design Motion doesn't generate video the way an image model generates pixels.
It uses HeyGen's open-source HyperFrames. Your agent writes HTML, CSS and GSAP animation code. Then that code gets rendered to an MP4 through headless Chrome and FFmpeg.
Why does that matter? Because the video is built from code, every part of it stays editable. Want a different colour, a new headline or slower timing? You change it and re-render. You're not rolling the dice on a new generation every time.
Per the release notes, a video request gives you two things: the editable source files and the MP4. Interactive animations stay as runnable HTML. And when you export, it checks the MP4 actually plays. If the export fails, it keeps your source so you don't lose the work.
📺 Watch: Open Design DESTROYS Claude Design?
My one-prompt demo
I hooked Open Design into my Agentic Operating System and gave it one prompt. Something like: make a 20 second paper cutout story, a trailer for Open Design.
That was it. One prompt.
It came back with a motion video I could preview right there. I could see the code behind it, export it, or share it. No back-and-forth to get a first version.
Honestly? The demos look better than what my human team makes. That's not a knock on my team. It's just how far this stuff has come.
If you want to see how I wire tools like this into one system, my Agent OS guide walks through the whole setup.
Run it for free
You don't need a paid model to use this. I ran it on free models inside OpenCode. The ones I tried were LongCat, Nemotron 3.5 Super VL, GLM 5.3 Flash and DeepSeek V4.1 Flash.
If you want the full walkthrough on that model, I covered running Nemotron 3.5 Super VL in OpenCode separately.
I also ran it with Hermes Agent, and it worked well there too. Per the repo, Open Design works with 19 coding agents, including Claude Code, Codex CLI, DeepSeek Harness, Cursor, GitHub Copilot, OpenCode, OpenClaw, Antigravity, Cline, Kimi CLI, Pi Agent and Hermes Agent. If you're on Hermes, my notes on Hermes Agent v0.22 are worth a read before you start.
You can also bring your own key. Per the repo, it supports Anthropic, OpenAI, Azure, Google, Ollama, LM Studio, vLLM and any OpenAI-compatible endpoint. There are optional OpenDesign Cloud models too.
Does it run locally or is it a website?
A viewer asked me this, and it's a fair question. The answer: it runs locally. Open Design is local-first. You install it as a desktop app, run it in Docker, or build it from source, and it all runs on your machine.
There's also an optional hosted mode if you'd rather not run models yourself. When you set it up, you pick from three options: Local AI, API key, or OpenDesign Hosted. So you get the choice. Local if you want control and free models. Hosted if you want convenience.
Want help picking the right setup for your business? Grab a free AI strategy session and we'll map it out with you.
📺 Watch: Open Design DESTROYS Claude Design?
Templates and modes
You don't have to start from a blank prompt. Per the release notes, Motion ships with five bundled examples: Motion Future, Product Promo, Showreel, UI Morph and Visual Lesson.
You open Motion Design from the Home screen or from Community. You can also pull in motion skills from the @ picker while you're working.
Beyond video, the templates I saw cover a lot of ground:
- Reels, like a 20-second high-energy reel where you choose the BPM
- Slides and documents
- Images and websites
- HyperFrames animations
- Live dashboards
Per the repo, the wider workspace has 100+ skills, 151 design systems and 15 deck templates. It exports HTML, PDF, PPTX, ZIP and Markdown, plus live artifacts.
One small thing I liked: it scans your machine and picks up the agents and CLIs you already have installed. Then you can switch model or provider without digging through settings.
Auto-post your videos with Blotato
Making the video is half the job. Getting it out is the other half.
I pair Open Design with Blotato. Blotato auto-posts the videos and reads the analytics back. So the loop looks like this:
- Write the script
- Edit the video in Open Design
- Post it with Blotato
Then you check what worked and make more of that. That's the whole content machine, mostly run by agents.
If you want a list of the other tools I lean on in this stack, check out my roundup of the best Hermes Agent tools.
Iterate and edit
The first version is rarely the final one. That's fine, because iterating is easy.
You just talk to it. "Here are my brand colours." "Make the intro faster." "Swap the headline." The agent updates the code and you re-render.
You can also go back and edit previous work. Because every video is built from source files, nothing is a dead end. That's the big difference between this and tools that hand you a flat clip and wish you luck.
Open Design Motion vs Claude Motion
Open Design positions this release as the open-source alternative to Claude Motion. I've used both, and here's my honest take.
I prefer Open Design. It's cleaner, it's easier to organise, and everything sits in one place: video, slides, websites, dashboards. I'm not going to make up specs for Claude Motion I haven't checked, so anything I haven't compared is marked below.
| Feature | Open Design Motion | Claude Motion |
|---|---|---|
| Open source | Yes, Apache-2.0 (per the repo) | Not compared |
| Runs locally | Yes: desktop app, Docker or source | Not compared |
| Free models | Yes, e.g. in OpenCode | Not compared |
| Works with many agents | 19 coding agents, incl. Hermes Agent | Not compared |
| Editable source | Yes, HTML/CSS/GSAP plus MP4 | Not compared |
| Organisation | Cleaner, easier to organise (my view) | Less tidy for me (my view) |
| All-in-one workspace | Video, slides, sites, dashboards in one place | Not compared |
If you care about benchmarks for the models you'd run behind it, I keep my own tests in Goldie Bench.
How to install Open Design
You've got three ways in, per the repo:
- Desktop app: available for macOS (Apple silicon and Intel) and Windows. This is the easiest route for most people.
- Docker: run it in a container and it's served at 127.0.0.1:7456 in your browser.
- From source: you'll need Node 24 and pnpm. Follow the steps in the repo.
There's no official Linux build yet, so Linux users will want Docker or source for now.
Once it's running, pick Local AI, API key or OpenDesign Hosted, let it scan for your agents, then open Motion Design from Home and try one of the five bundled examples. Then write your own prompt.
Want more step-by-step builds like this, plus a group of people actually using these tools every day? Join the AI Profit Boardroom for $69/mo.
FAQ
What is Open Design Motion?
It's the motion-video feature in Open Design, added in v0.25.0. You describe a video in a prompt, your AI agent writes HTML, CSS and GSAP code, and HyperFrames renders it to an MP4 you can edit and export.
Is Open Design Motion free?
Yes. Open Design is open source under Apache-2.0, and you can run it with free models in OpenCode. Paid API keys and the hosted option are there if you want them, but you don't need them.
Does it run locally?
Yes. It's local-first. Install the desktop app on macOS or Windows, run it in Docker, or build from source. There's an optional OpenDesign Hosted mode if you'd rather not run anything yourself.
Can I use it with Hermes Agent?
Yes. Hermes Agent is one of the 19 coding agents Open Design supports, and I've run Open Design Motion with it myself. It's a solid combo.
If you want a plan for using Open Design Motion and AI agents to grow your business, book a free AI strategy session and we'll build it with you.
