JayDang.

Profile

On January 3, 2023, Jay Dang co-founded a company. On January 5, 2023, he launched it. The gap between those two dates says more about how he operates than any pitch deck ever could.

FlowGPT - the platform he built with co-founders Lifan Wang and Alex Xi - went live two days after it existed as a legal entity. Within a week, it had 14,000 users. Within months, millions. Jay was a freshman at UC Berkeley when it happened. He was also, by then, a dropout.

The Setup

Jay Dang - Jiacheng Dang by birth - grew up coding. Middle school for the hobby, high school for the ambition. By the time most teenagers were picking college applications, he was publishing a paper on AI-assisted glaucoma detection, inspired by a family member's diagnosis. He didn't pursue the research further. The paper was a proof of curiosity, not a career plan.

He landed at UC Berkeley for Computer Science. One semester in, he had already completed internships at four different startups. Then, before the semester ended, he tried to start his own company: an AI-driven A/B testing platform for advertising. He assembled a team, landed clients from listed companies, and watched it fall apart over a co-founder dispute.

The lesson wasn't to be more careful. The lesson was to move faster.

December 2022 Changed Everything

ChatGPT launched in December 2022 and Jay pivoted his entire worldview within weeks. Consumer-facing generative AI was the obvious next frontier - but nobody had built the place where people could share, discover, and build on top of AI prompts. The gap was enormous. He could see it.

He pitched the idea to Lifan Wang - an MBA graduate and former Amazon engineering manager - at UC Berkeley's SkyDeck accelerator. Wang signed on as COO. Alex Xi came in as CTO. The three of them incorporated on January 3, 2023.

Two days later, they shipped.

What FlowGPT Actually Did

The original pitch called it "Product Hunt for prompts." After Jay and Lifan ran a thousand user interviews - a number that reads like a typo but isn't - the positioning shifted to "GitHub for prompts." The distinction matters: GitHub isn't a showcase, it's infrastructure. A place where builders build, share, fork, and improve.

FlowGPT became that. A marketplace where anyone could publish an AI application, and anyone else could use it, remix it, or build on top of it. The philosophy was deliberate: no algorithmic gatekeeping, creator ownership, maximal openness.

That openness attracted users. It also attracted attention from critics. In February 2024, TechCrunch ran a piece titled "FlowGPT is the Wild West of GenAI Apps," documenting the platform's content moderation gaps. Jay responded publicly, describing the platform as "ethical and rule-abiding" while acknowledging that scale creates challenges. The tension between accessibility and safety is a problem every open platform eventually has to solve. FlowGPT was no different.

The $10 Million Vote

On February 26, 2024 - the same day TechCrunch ran their critical piece - FlowGPT announced it had raised $10 million in a pre-Series A funding round led by Goodwater Capital, described as the world's largest venture capital fund exclusively focused on consumer technology, with DCM Ventures participating. DCM had already been backing FlowGPT through its stealth phase.

The timing was either terrible or perfectly calibrated. Either way, the capital arrived.

The money went toward expanding the engineering and research team, building out mobile product capabilities combining LLMs with text-to-speech, image generation, and video tools, and hiring across product and operations. The company, with its 24-person team operating out of San Francisco, was moving from prompt marketplace to full-stack AI content platform.

Kaon: The Next Chapter

FlowGPT didn't stay FlowGPT. The company rebranded as Kaon - short for nothing in particular, expansive in intent. Where FlowGPT was a marketplace for AI prompts and applications, Kaon positioned itself as an "AI-native content lab," building multiple consumer products under a single corporate entity.

The flagship product from this era is Emochi - an AI-powered interactive storytelling app focused on anime, manga, games, and entertainment. Users can create and explore AI-generated interactive stories, build customizable AI characters, and access AI image generation. By April 2026, Emochi had ranked #12 globally with 12.3 million monthly active users.

That number - 12.3 million - is not nothing. Most funded startups never see a fraction of it. Jay Dang reached it before turning 23.

How He Runs It

Lifan Wang, FlowGPT's COO and Jay's co-founder, gave a direct answer when asked what he had learned from building with Jay: "Execution is crucial. That is the most important thing I learned from Jay, my co-founder."

That word - execution - keeps appearing in every account of how the company operates. Launch first, iterate constantly, talk to users obsessively. Jay and Lifan completed over a thousand user interviews in the early months. The product roadmap wasn't built from vision documents; it was built from feedback loops.

Jay's advice to other founders tracks with this: when a project shows rapid growth with minimal costs, commit fully. The timing of the AI wave is irreversible. The window for building consumer AI products was open and closing. He saw it, jumped through it, and has been running since.

What Makes Jay Dang Unusual

There are a lot of young AI founders. Most of them have a thesis about where the market is going. Jay Dang had a thesis and a shipped product within 48 hours - before competitors had time to copy the concept, let alone execute it. The speed of that first move set the entire competitive dynamic.

He also did something that gets underreported: he talked to users. A thousand conversations, not to validate assumptions, but to discard them. The "Product Hunt for prompts" framing got replaced. The "GitHub for prompts" framing got refined. The product shifted based on what people actually did, not what Jay assumed they'd do.

The glaucoma paper from high school is worth noting, not because it launched a research career, but because it didn't. Jay explored a problem, published something real, and moved on when the opportunity cost shifted. That pattern - engage fully, ship, and pivot when the data says to - repeats throughout his career.

He is also, reportedly, one of the youngest Chinese entrepreneurs to secure venture financing in the AI industry. The record is worth naming because the structural barriers to that recognition are real, and clearing them at 20 while building a company with millions of users is not a routine outcome.

The Bigger Picture

The company Jay Dang is building is not the company he started. FlowGPT was a marketplace for AI prompts. Kaon is a content platform for AI-native entertainment. The arc from "GitHub for prompts" to a 12-million-user storytelling app reflects something about where consumer AI is actually going: not productivity tools for power users, but entertainment products for everyone.

Emochi's success in anime and interactive fiction is a specific bet on that thesis. The demographic consuming that content is global, young, and deeply comfortable with AI-generated media. Jay Dang - himself young, technical, and without nostalgia for the pre-AI world - is building for that audience.

Whether Kaon becomes a defining consumer AI company or a stepping stone to something else isn't yet settled. But the trajectory from two-day launch to 12 million monthly users in three years suggests that Jay Dang's operating rhythm - ship fast, listen harder, pivot when needed - is the kind of engine that tends to compound.

He dropped out of Berkeley to pursue an opportunity he could see clearly. Three years later, the bet looks like it's paying off. The question now is what he builds next - and how fast.

Key Achievements