IF THE UNIVERSE IS EXPANDING at the speed of light, what exists beyond the edge it hasn’t reached yet? During junior high back in China, Reni Cao brought the question that had been bothering him for a while to his physics teacher.
His teacher couldn’t answer but told him to stop asking distracting questions and finish his homework.
About two decades later, on the first day GPT-4 became available, Cao asked the chatbot the same question. It wrote him a detailed philosophical treatise, citing Wittgenstein’s idea that the limits of language are the limits of our world — that perhaps the question itself was constrained by three-dimensional thinking, by assuming all things must have an inside and an outside.
“I thought: if I’d had someone or something that engaged with my curiosity like this when I was growing up, I would be more enlightened,” Cao says, sitting in a sunny San Francisco office where he now runs a startup. “I immediately thought: how do I give this ability to kids? Or just to my past self?”
That impulse led Cao to quit his product marketing job and co-found Dex, a toy-like device that looks like a magnifying glass with a camera on one end and a touchscreen on the other. Kids photograph objects around them, and AI supplies the names and translates them into eight major languages, plus 34 dialects. The device was launched in August 2025 and is now on the market.
What may be more interesting than the product is the philosophy behind it: Cao is building for children like his younger self, the ones teachers called “distracted,” the ones who couldn’t stay in their lane. In an era that demands specialization, he’s betting on staying curious.
The unmarketable candidate
As an undergraduate studying financial engineering in China, Cao arrived in the U.S. as an exchange student and took courses covering everything from finance to abstract algebra, computer science and electrical engineering.
“My advisor criticized me for being too spread out,” he says. “They’d say, ‘What are you going to do with all this? You need to become a specialist.’”
He did the opposite.
When it came time for graduate school, he switched fields entirely, earning a master’s in information systems from Carnegie Mellon. Staring at his resume before graduation, he wondered what job he even qualified for, feeling that he wasn’t truly a computer science person or a finance person — that he didn’t quite fit neatly into any category.
Then he discovered product management, a field that in some respects sits within disciplines and a role that required exactly his odd combination: analytical thinking, technical fluency and business acumen. He applied to LinkedIn, which received over 1,300 applications that year for its associate product manager program. They hired six people. Cao was one of them.
He points to a classical Chinese idiom that captures his belief — often rendered as “a gentleman is not a tool” — a Confucian idea that a person should not be limited to a single, narrow function.
“I’ve always believed generalists will outperform specialists in contemporary society,” Cao says. “I think it will be even more true in the AI era.”
It’s a contrarian position in Silicon Valley, where depth of expertise often trumps breadth. But for Cao, his scattered background wasn’t a bug but preparation for building something nobody else would think to create.
The education nobody asked for
Becoming a father intensified Cao’s questions about education. His daughter, now five, is growing up bilingual in Chinese and English. When he talks about her language development, he doesn’t frame it as a typical heritage language preservation story — the anxious immigrant parent watching their child lose connection to their roots.
“She hasn’t encountered difficulties,” Cao says. “But I always feel like her potential hasn’t been fully realized yet.”
As she grew up around him, Cao watched her vocabulary explode and saw how quickly children acquire language compared to adults. What struck him wasn’t what she was losing, but what the current education system wasn’t designed to unlock.
He believes that the next decade will bring an education paradigm that emerges outside the traditional classroom, one meant for kids like his childhood self, the ones overflowing with questions.
His co-founder Charlie Zhang had a different but related motivation. Zhang watched his younger son experience speech delays. The boy was reluctant to speak, falling behind developmental milestones. But when they gave him an early Dex prototype, something shifted. The child’s first word was “blueberry,” which was spoken not to his parents but to the device.
Another user review describes a child who started with Spanish, discovered the device supported other languages, and began experiencing the same content in multiple languages like someone watching a film with different subtitle tracks. That resonates with Cao’s broader philosophy.
“There’s this concept in Wittgenstein called language games — the idea that language fundamentally is a kind of game,” Cao says.
Dex, he explains, isn’t just a translation app but more like a toy. It first captures an object the child points to and identifies the object by name. The child is then prompted to repeat the word out loud. Based on that interaction, Dex responds with more context like a short story connected to the object.
“Like a magic spell. You say these words to the device, and it renders interesting content and experiences,” Cao says. “For kids, that’s how their world works.”
The non-consensus bet
Quitting stable tech employment to build an educational toy was, by conventional wisdom, insane. Cao had spent years in product marketing at established companies, but he soon realized that the mindset and skillset needed to be a founder were entirely different.
In his view, large tech companies are structured to minimize risk above all else. Their revenue machines run so reliably that innovation becomes secondary, and progress requires navigating a maze of approvals from vice presidents and directors. A single objection from a senior stakeholder can halt a project entirely. Startups operate on opposite principles.
“You can have 10 investors in a room, nine think you’re a complete idiot, and one believes in you. You win,” Cao says. “In fact, consensus among investors usually means the idea isn’t bold; it’s something mediocre and already competitive, leaving little space to build a great company.”
“You can have 10 investors in a room, nine think you’re a complete idiot, and one believes in you. You win.”
Reni Cao, Dex co-founder
He and his co-founders faced many rejections. The criticisms were predictable: the market was too niche, the hardware was too risky, the approach was too unconventional. Venture capitalists suggested pivoting the messaging.
“They’d tell us to call it ‘Children’s AI’ or ‘Children’s ChatGPT.’ We refused. This had to be conviction-driven, not consensus-driven,” he says. “We need to stay true to ourselves.”
That conviction ultimately paid off: the company recently raised $4.8 million in funding from ClayVC, EmbeddingVC, Parable, UpscaleX and several angel investors. According to Cao, thousands of families have already purchased the device since its launch.
The message to his daughter
Cao could have stayed in big tech, collected a comfortable salary, maybe interviewed at AI companies for a significant compensation package. Instead, he’s taken “meager pay, or no pay at all for long stretches,” building something venture capitalists struggle to categorize — too technologically ambitious for traditional toy makers, too focused on children for typical tech companies.
It’s an echo of that junior high classroom in China, where a boy’s question about the universe’s edge was dismissed as a distraction. Cao never stopped asking questions. Now he’s built a toy that tells his daughter, and every child who picks it up, that their curiosity isn’t something to suppress but to follow.
There is a method to the financial sacrifice, a deeper message embedded in the choice itself.
“This is a once-in-a-lifetime chance to build a generational product,” Cao says. “It’s a message to my daughter. She’ll see that her dad could have taken safe jobs and earned money but chose to do something we felt was meaningful. Life isn’t only about climbing or earning. There are other possibilities.”
This story originally appeared in Peninsula Press, a publication of the Stanford Journalism Program.