Manus AI is having a moment. The all-in-one AI agent from China has been building buzz for months, thanks to its ambitious approach to task automation. Manus has positioned itself as an autonomous online coworker complete with invisible mouse and keyboard and capable of breaking down requests into steps and carrying them out across multiple tools. That toolkit now includes image generation.
Of course, there’s a lot of competition among AI image generators. ChatGPT’s own recent upgrade, in particular, has stood out for its high-quality creation. But, Manus is pitching its visuals as a part of a larger workflow, one that can incorporate those images into bigger projects.
Introducing Manus image generation.Manus doesn’t just generate images. It understands your intent, plans a solution, and knows how to effectively use image generation along with other tools to accomplish your task. pic.twitter.com/pWbaVOy7XbMay 15, 2025
Sandwich Shop
I decided to take Manus for a spin and see if this new visual muscle was just a gimmick. I gave it a couple of complex, multi-step prompts, including images, to see just how well it did at understanding my requests.
To start, I went for a batch of related business-related images by telling Manus I was setting up a food truck and I wanted it to come up with the name and to show me the truck, the menu, and even come up with a shirt for me to wear. After clarifying what kind of food (sandwiches) and who I would be marketing to (hungry people). Manus went to work and produced the images above. Admittedly, it took about seven minutes, but that’s because the AI threw in some marketing guidelines, sandwich recipe ideas, and even a business card to get me started.
To round out the project, I asked for asked Manus to show someone working at the sandwich truck. Manus asked who should do it, and I uploaded a headshot of myself. The AI did a pretty good job of putting the shirt on a human body and showing what a popular truck restaurant line would look like.
Manus the fate of AI
What impressed me most wasn’t just the aesthetic quality of the images, but how seamlessly image generation was integrated into larger task workflows. Manus never treated these prompts as endpoints. Everything felt modular and adaptive. Want to use the logo on merch? It mocks that up too.
That kind of cohesion puts Manus in a different category than most visual AIs. Midjourney, Firefly, ChatGPT, and others are all great for making artwork, but you usually need multiple prompts to get it to put the images in a different context. Manus generates images to complete a task, not just as a random piece of art.
Of course, there are still kinks. Not every human looks quite human, and it takes significantly longer, even than ChatGPT’s image creator, because of how it fits the images into the larger workflow. Still, for something this early in its visual evolution, it’s remarkably polished.
What Manus is building hints at the next phase of AI use. Models won’t just get smarter, they’ll be more integrated. Manus isn’t trying to win an art image generation contest. It’s trying to get you to your goal, whether it’s launching a business or building an imaginary universe full of glowing jellyfish.
