Yimao Zhou isn’t building another AI tool. He’s building cognitive infrastructure, and MiraclePlus bet on him in 10 minutes.
In the rapidly evolving AI agent landscape, where every startup races to build the next ChatGPT wrapper, most founders struggle with a fundamental problem: they’re building tools when they should be building infrastructure. Yimao Zhou, the 23-year-old founder and CEO of Emagen AI, has taken a different approach—one that replaces incremental productivity gains with a radical thesis about the future of human labor.
Zhou’s background suggests why this methodology comes naturally. After starting to code at age nine and leading teams to victories in national AI competitions by middle school, he made an unusual choice: enrolling in medical school at Shanghai Jiao Tong University—one of China’s top three institutions—before transferring to study cognitive philosophy and philosophy of science. That intersection—understanding both technical systems and human cognition—shaped how he would later approach building Emagen AI.
The Infrastructure Thesis
Zhou’s journey into AI agents began with a recognition shaped by his philosophical training. “Tools are invisible when they work perfectly, but they’re still just tools,” he reflects. “Infrastructure changes what’s possible. That distinction guided my entire approach to understanding what the AI industry was getting wrong.”
At Emagen AI, Zhou faced a challenge common to contrarian founders: potential investors often struggled to understand his thesis until he reframed the problem entirely. Rather than conducting standard pitch meetings focused on features and metrics, Zhou led with a provocative question that would define his company’s direction.
“Everyone asks: how do we make AI assistants better?” Zhou explains. “I think that’s the wrong question. The right question is: how do we make human labor optional?”
Building a Category-Defining Framework
Rather than building another productivity tool or ChatGPT integration, Zhou developed a framework that positioned Emagen AI’s product, Cagen, as something fundamentally different: cognitive infrastructure.
“We started by acknowledging that we weren’t building a better assistant,” Zhou explains. “We were building the layer that makes assistants irrelevant. Think about what electricity did to candles. It didn’t make candles more efficient. It made candles extinct.”
The framework required disciplined positioning. While competitors focused on incremental improvements—making AI 10% faster, 10% smarter—Zhou insisted on infrastructure-level thinking. By categorizing Cagen as cognitive infrastructure rather than an AI tool, Emagen AI could conduct conversations with investors and customers that yielded strategic partnerships rather than feature requests.
What distinguished this approach was the emphasis on long-term vision before tactical execution. “Before we wrote a single line of production code, we aligned on the category we were creating,” Zhou notes. “If different team members thought we were building a tool, we’d make tool-level decisions. We needed everyone to think at the infrastructure level.”
From Thesis to Validation
The thesis-driven model transformed Zhou’s fundraising from a pitch into a philosophy discussion. Each investor conversation became a structured exploration of what AI infrastructure should become. Zhou implemented a framework where the quality of questions asked mattered more than the features demonstrated.
The results surprised even Zhou. When he pitched to Qi Lu—the legendary technologist who served as Executive Vice President at Microsoft and President and COO of Baidu, now founder of MiraclePlus (formerly YC China)—he expected skepticism. Lu had evaluated thousands of startups across two continents.
“He stopped me ten minutes into a thirty-minute meeting,” Zhou recalls. “He said he’d heard enough. He wanted to invest immediately. He told me my framing was correct—cognitive infrastructure, not tools. That was the right abstraction layer.”
MiraclePlus invested $300,000 at a $4.28 million valuation. Zhou became the youngest founder in the F24 cohort.
Translating Technical Vision into Market Position
Zhou’s approach addresses a fundamental challenge in emerging technology sectors: how to build for markets that don’t yet understand they need what you’re creating. His insistence on category creation before product development ensures that Emagen AI isn’t competing on features, but defining what features should exist.
“In AI, everyone talks about capabilities and benchmarks, but those are abstract concepts,” Zhou observes. “What matters is understanding that we’re building infrastructure for a species in transition. Our framework forced us to think at a civilizational scale rather than a product scale.”
The structured approach also enables Emagen AI to pivot efficiently when market conditions change. Because every initiative begins with explicit thesis validation, the team can quickly identify which assumptions no longer hold and adjust course without organizational confusion.
Building What Actually Matters
Beyond the tactical benefits, Zhou’s framework reflects a broader philosophy about how AI companies should approach their markets. By deeply understanding the infrastructure thesis before building, Emagen AI allocates engineering resources to capabilities that will define categories, rather than impressive demos that solve problems customers don’t know they have.
“The AI space is full of technically elegant solutions searching for problems,” Zhou notes. “Our job is to reverse that equation—start with the infrastructure thesis, then engineer what that infrastructure requires. The framework keeps us honest about what we’re actually building.”
Under Zhou’s leadership, Emagen AI has assembled a team from elite universities across China and the United States. The company completed its first major development phase in Q3 2025.
The Compounding Effect of Category Creation
As Emagen AI continues building cognitive infrastructure for the next generation of AI applications, Zhou’s thesis-driven approach offers a template for how AI companies can navigate uncertainty without losing ambition. The methodology demonstrates that category creation isn’t philosophical overhead—it’s a competitive advantage that compounds over time.
We’re still early in understanding what cognitive infrastructure needs to become,” Zhou reflects. “But we’re less early than we were when Qi Lu decided in ten minutes that we were right. That’s the power of thesis-first building—it attracts people who see what you see.”
For Zhou, success in AI isn’t measured by feature comparisons alone, but by whether you’re defining the category or competing within someone else’s. His methodical approach—grounded in philosophical training, sharpened by technical execution, and validated by one of tech’s most respected figures—positions Emagen AI to build infrastructure that doesn’t just impress developers, but changes what’s possible for humanity.
When asked what success looks like in ten years, Zhou doesn’t hesitate: “People stop asking ‘what AI tool do you use?’ and start asking ‘what cognitive infrastructure are you running?’ That’s when we’ve won.”
Photo by Steve Johnson; Unsplash
