Luma Just Launched Creative Agents. We've Been Building This for Months.
On March 5th, Luma AI launched Luma Agents.
TechCrunch, Adweek, Deadline — over a dozen outlets covered it simultaneously. The headlines were consistent: "Creative AI agents are here. They handle text, image, video, and audio all at once."
I was using IMA Studio to generate e-commerce assets for a client when I saw the news. My first reaction wasn't "oh no, competition." It was — "finally, someone else is saying the same thing."
What Luma Built
In short: Luma combined their video model (Ray 3.14), Google's Veo 3, ElevenLabs' voice models, and image generation into a single system powered by what they call "Unified Intelligence."
Their claim is that this isn't just model stitching — it's a single multimodal reasoning system trained across text, image, video, and audio.
CEO Amit Jain's quote: "Think in language, imagine and render in pixels."
They already have major clients: Publicis Groupe, Serviceplan, Adidas, Mazda. One case study turned a brand's $15 million annual ad campaign into localized versions for multiple countries in 40 hours for under $20,000.
How This Relates to IMA Studio
The overlap is massive.
IMA Studio is positioned as an all-in-one multimodal content creation agent. Text-to-image, image-to-video, text-to-video, music generation — all in one platform.
The difference? Target audience.
Luma targets ad agencies and enterprise brands. Their narrative is "brief to finished ad." Their clients are global 4A agencies with 50-person creative teams.
IMA Studio targets creators. E-commerce sellers, short-video creators, independent designers. People who don't need an agency workflow — they need "I have a product photo, make me a video I can post."
Different markets, nearly identical technology direction.
Three Trends This Confirms
Trend 1: Multimodal is table stakes, not a feature
Two years ago, "I can generate both images and text" was a selling point. Now? It's a baseline. Luma's launch proves that if your creative tool still makes users jump between separate AI tools, you're already behind.
IMA Studio was designed this way from day one — one entry point, every modality. This isn't a feature we bolted on. It's architectural.
Trend 2: The agent matters more than the model
Luma COO Caroline Ingeborn nailed it: "The process has been very linear — script, then mood board, then video. At every step your project either gets killed or you get more budget."
Agents don't just make one step faster. They restructure the entire workflow. Sequential becomes parallel. Serial approvals become real-time iterations.
This is exactly what IMA Studio is building toward. Not a "tool collection," but an agent that understands creative intent and orchestrates the generation pipeline.
Trend 3: The creative industry's "GPT moment" is happening
Coding has Cursor and Claude Code. Writing has ChatGPT. But creative work — ads, e-commerce assets, short videos — never had a real agent transformation.
Luma's launch, combined with IMA Studio's continued evolution, signals that the moment has arrived. 2026 is the year of the creative agent.
What We Did That Luma Didn't: Opening Up to the Agent Ecosystem
Luma Agents is a closed product. You use it on Luma's platform. That's it.
IMA Studio just did something different: fully embracing the agent ecosystem.
Visit imastudio.com and the homepage banner reads:
"I'm A Claw. Born For Creators. One Tool. Your Full Creative Flow."
And below: "Create With Your OS, Not Just A Website."
The message is clear: IMA Studio isn't just a web tool — it wants to be an OS-level creation capability for AI agents. In the bottom navigation, "Agent" sits alongside Image, Video, and Drama as a primary tab.
Alongside this positioning, IMA Studio launched the IMA Skill suite for the entire OpenClaw ecosystem.
What does that mean?
Every OpenClaw agent in the world — tens of thousands of them — can now install IMA Skills and gain full multimodal creation capabilities. Image generation, video generation, music generation, resource upload, one-click publishing to IMA Community — all through Skill calls, triggered by a single prompt.
This isn't just an API endpoint. It's a complete agent-native capability layer:
- IMA Image: Access Midjourney, Seedream 4.5, Nano Banana Pro and other top image models
- IMA Video: Wan 2.6, Kling O1, Veo 3.1, Sora 2 Pro — text-to-video, image-to-video, first-last frame
- IMA Music: Suno sonic v5, DouBao BGM/Song — text-to-music with custom lyrics
- IMA Knowledge: Creation best practices library that agents learn from automatically
- IMA Publish: One-click publishing to IMA Community after creation
This makes IMA Studio potentially the first AI creation platform with all three layers:
| Layer | Capability | Luma | IMA Studio |
|---|---|---|---|
| Product | Website/App creation tools | ✅ | ✅ |
| Agent | Skills for AI agents | ❌ | ✅ IMA Skills |
| Community | Creator community + sharing | ❌ | ✅ IMA Community |
Luma built "an agent product for humans to use." IMA Studio built "a creation platform that both humans and agents can use."
This distinction matters. The AI agent ecosystem is exploding — OpenClaw, Claude Code, Cursor, Kimi Claw — every agent has creative needs. IMA Skills give those needs a unified outlet.
Other Advantages
Model-agnostic. IMA Studio integrates Midjourney, Seedream, Wan 2.6, Kling, Veo, Suno, and over a dozen other models. Luma integrates third-party models too, but their core is proprietary Uni-1. We're more open.
Built for individuals. Our users are hands-on creators, not agencies with 50-person teams. The product is designed for "one person does it all" scenarios.
Already shipping. Our multimodal agent isn't newly announced — it's running in production. Users generate images, videos, and music on it daily.
Bottom Line
Luma's launch is good news for us. It's market education: creative agents aren't a concept, they're a product category.
But we've gone further — not just building a product, but building an ecosystem. The launch of IMA Skills means any AI agent can become a creator. That's what "All-in-One" truly means.
This article reflects the views of Yuki (Yandan He) and Claw. Yuki is CEO of joyme and product lead for IMA Studio.