Seven AI Agents, One Sentence: Inside OiiOii's Bet on Animation Without a Crew

OiiOii, an AI animation platform founded by NaoNao in July 2025, opened its beta with 100 spots and received over 100,000 signups, with a large share from Japan and Taiwan rather than Mainland China. The platform turns a single sentence into a finished animated short using seven specialized agents, integrating more than 30 foundation models including Seedance 2.5 and 2.0. NaoNao, who previously ran the major animation business at Bilibili, said the response reframed what the company thought it was building.

OiiOii opened its beta with 100 spots and received more than 100,000 signups. Founder NaoNao had expected them to come almost entirely from Mainland China, the company's home market. A large share arrived from Japan and Taiwan instead, which reframed what the company thought it was building. OiiOii is an AI animation platform that turns a single sentence into a finished animated short, coordinated by seven specialised agents rather than a human chaining prompts together. NaoNao, who studied Chinese painting as a child and later ran the major animation business at Bilibili, spoke to StartupFortune about the gap between how much animation people want and how much the industry can actually produce. What led you to build OiiOii, and what gap in AI animation tools were you trying to close? I have loved anime since I was a child. What drew me in was its emotional purity, the kind of unbreakable bonds between characters that felt harder to find in real life. I studied Chinese painting growing up, and after I left Tencent I actually tried to learn animation software myself. The tools were too difficult, and I set the dream aside. The turning point came at Bilibili, where I ran the major animation business. I could see the scale of demand against how limited actual production capacity was, and that gap convinced me AI could close it. I think often about Pixar's Edwin Catmull: he could not draw, but he found an indirect path to his animation dreams through technology. I recognised that same path in myself. I do not draw by hand, but I understand product and technology well enough to build tools that let other people create. I founded OiiOii in July 2025 with longtime collaborators to close two gaps I kept seeing. Freeform, open-canvas tools ask users to manually chain prompts and link nodes, which locks out anyone who is not already a power user. And raw video generation models, even strong ones, tend to require long queues and repeated regeneration cycles to get a usable result, with no real support for long-form video and persistent problems keeping characters and scenes consistent from shot to shot. Walk us through how a creator actually uses OiiOii, from an idea to a finished animated short. It starts with one sentence. A creator describes their idea in plain language, no prompt engineering required, and that single input kicks off our agent workflow. A Script Writer agent develops the story, our Character Designer and Environment Designer agents build out the visual world, and an Art Director sets the overall look, pulling from a style library of more than 170 aesthetics. A Storyboard Artist then writes shot-level prompts the way a real production would, thinking about what actually makes an effective frame rather than passing the user's words straight to a model. From there, our Animator agent handles the final assembly, integrating every scene into one complete short, while a Music Supervisor agent scores the piece. Storyboarding is really the technical core of the whole system, because it is the single biggest factor in whether the final video is usable. We built a dedicated face ID injection system so a character's likeness holds across every cut, and we maintain independent asset libraries for characters, scenes and props so consistency extends across a full sequence, and even across an entire series, not just one clip. Underneath all of that, we integrate more than 30 foundation models, including full-capacity access to Seedance 2.5 and 2.0 with no queuing, and creators can run multiple projects at once. What surprised you most about the beta response, with over 100,000 signups for just 100 spots? Honestly, what surprised me most was not just the number, it was where it came from. We opened beta with 100 spots expecting our registrants to be almost entirely from Mainland China, since that is our home market. Instead we ended up with more than 100,000 signups, and a large share of them came from Japan and Taiwan as well. That told us the demand for accessible animation tools was not a China-specific thing; it was regional, and much bigger and more pent up than we expected. It also reshaped how we operated in those early months. With that many engaged users watching closely, and clearly not just from where we assumed, we leaned on their feedback in near real time to refine the consistency and usability issues that mattered most, instead of guessing at a roadmap in isolation. It confirmed something I had believed since my time at Bilibili: there is an enormous amount of demand for animation that the industry's existing production capacity simply cannot meet. How does the team of seven AI agents divide the creative work, and where does human control still come in? Each one focuses on a distinct stage of production, and they coordinate automatically off a single prompt, closer to working with a small production crew than operating one model. Human control comes in at the very top and stays there. The creator sets the premise, the tone and the direction, and that creative control stays fully in their hands throughout. The agents execute a production pipeline competently and consistently; they do not make the creative decisions themselves. That is really the whole philosophy behind OiiOii: a single person should be able to guide an entire animation production the way a director would, without needing a full crew or years of technical training to do it. What will it take to reach your goal of empowering 100 million people to create their own animations by 2030? Getting there is a three-part evolution for us. Near term, it is about maturing the product itself, moving the agents past their infant stage into something that is a truly capable creative partner. That means tightening character and scene consistency across longer sequences, growing our style library further, and building out content templates so any creator can produce something with real consumable value, regardless of background. Medium term, the tool becomes the foundation for something bigger than the tool itself. OiiOii the product is a means, not the end goal. What we are building toward is a thriving content ecosystem, a creator community where animation becomes a native language for storytelling across genres, from entertainment and education to finance and culture. That mirrors something I learned at ByteDance: you embed a community's future content DNA into the tool from day one, so that by the time the ecosystem forms, it already has shape. Long term, that ecosystem should produce original IP: proprietary characters, worlds and franchises that belong to OiiOii and its creator community, rather than being borrowed from existing anime or games. If we get all three layers right, reaching 100 million creators by 2030 comes down to how many people we can hand a single prompt and seven agents to, and how much of their own story they end up able to tell with it. Where do you want OiiOii to be a year from now? A year from now, I want OiiOii to have moved past being what I would call an infant agent into something that feels like a truly capable creative partner. Concretely, that means deeper character and scene consistency across longer sequences than we handle well today, a style library that keeps expanding to track what creators actually want to make, and enough content templates that someone with zero animation background can still produce work with real consumable value. We are also continuing to scale outside China. Beta already reached Japan and Taiwan alongside our China launch, and since June 2026 we have been building out our North America push, including hiring team members with US and Japanese backgrounds specifically to lead that expansion. A year from now, I want that global footprint to be much further along, and I want the product itself to have earned it.