Artificial Intelligence (AI) is taking businesses by storm, rapidly transforming how we work and think. While some are hesitant about the technology, the World Economic Forum (WEF) reports that 86% of employers expect AI to have a transformative impact on their businesses by 2030.
Amid this shift, Henry (Lifan) Wang, an MBA graduate from the University of California, Berkeley’s Haas School of Business, saw an opportunity to help shape the future of the AI space. Believing the technology is here to stay, he feels there will be long-term opportunities for AI innovators, and has set about building products to help consumers get the best out of the technology.
Here, Lifan shares how his time at business school inspired his entrepreneurial journey in the ever-evolving world of AI.
Tell us about your businesses?
I’ve co-founded several tools within the AI space, which I’m really proud of.
Under Kaon, which builds consumer AI products that turn ideas into interactive experiences, I currently have two products: FlowGPT and Emochi.
In 2023, I launched FlowGPT, a community platform where people create, share, and fork prompt-based AI agents (e.g., résumé generators, business-plan helpers). Discovery and remixing are built in, and within 10 months, I’m proud to say we reached 5 million monthly users. Our unique selling point is that we’re a creator ecosystem for AI agents with rapid iteration, social discovery, and forking—like GitHub, but for AI experiences.
A year later, my co-founder and I started Emochi, an AI-personalized content platform with over 1 million interactive AI stories. Users co-create in real time—the AI adapts plots and characters to each person’s inputs and preferences. We’ve already surpassed 10 million downloads worldwide. At its core, we turn passive content into dynamic, personal, and replayable stories powered by AI.
What inspired you to start Kaon?
There were several points of inspiration.
The seed was planted during my MBA at Berkeley Haas—particularly during Professor Pieter Abbeel’s 'Business of AI' class—when we discussed GANs (Generative Adversarial Networks) and the first AI-generated artwork that sold for over $400K.
The real “go” moment came when GPT-3.5 launched. My co-founder and I noticed people eagerly sharing prompts and agents across Discord, LinkedIn, and other communities. It felt like the early open-source days—think Ruby on Rails—before platforms like GitHub supercharged collaboration. We believed prompts were becoming the interface between humans and AI, and there was room for a “GitHub for AI.”
Tell us about your business today?
We have a small but agile team with fewer than 10 full-time employees. It’s a collaborative, fast-paced environment where everyone wears multiple hats and contributes directly to product evolution.
What part did your b-school experience play in developing your business idea?
As I mentioned, my MBA played a huge role in inspiring my business—it gave me both the spark and the sandbox.
The spark came from the Business of AI course. Seeing GANs and the first AI artwork sale reframed what “software” could be. The sandbox was the Haas/Berkeley ecosystem, which allowed us to pressure-test the “GitHub for AI prompts” idea with real users, refine the business model, and ultimately launch FlowGPT and later Emochi.
In what ways do you draw upon the skillset you developed at b-school?
The skills I learnt during my MBA are the backbone of how I operate the business today. Every week, I use what I learned at Haas—structured problem-solving, data-driven experimentation, unit economics, pricing, storytelling for fundraising, and people leadership.
Tell us about your typical working day?
Monday through Thursday, I usually work from 11:00 am to 1:00 am, with quick breaks for lunch and dinner.
From 1:00–3:00 am, I reserve quiet time for reflection and personal learning—journaling, reading research, and refining product roadmaps.
Fridays wrap up earlier for recovery, and weekends are mostly for social time and recharging so I can show up energized on Monday.
How has your b-school network helped you develop Kaon?
The Berkeley Haas community has been a real force multiplier. I met angel investor Brad Bao (a Haas alum) at an event, and his early belief in our vision accelerated everything.
What have been some of the biggest obstacles you’ve encountered?
The toughest stretch was the first month after launching Emochi. Our growth, monetization, and retention all came in our plan (below). Rather than pivot, we treated it like a diagnosis problem and rebuilt our way forward.
- Bet on model quality. We invested in model selection, tuning, and personalization because response quality is the biggest driver of engagement. Better relevance leads to longer sessions, stronger retention, and higher monetization.
- Make experimentation a habit. We adopted disciplined A/B testing with clear hypotheses, success metrics, and shutdown plans if something hurt the core experience. This allowed us to scale what worked and retire what didn’t.
- Run on unit economics. From day one, we tracked payback windows and monitored ROAS/ROI by channel, creative, geography, and cohort (CTR, CVR, CAC). That rigor allowed us to acquire users up to 10× cheaper than many competitors.
Where would you like Kaon to be in five years’ time?
In five years, I want us to be the breakout consumer platform of the AI era—the app with the longest time spent across all AI platforms. I truly believe we’re building the next big thing in consumer AI.
What is one surprising thing you’ve learned from starting your venture?
Small, “boring” systems beat big ideas. Tight onboarding, clean metrics, and a weekly kill-or-scale cadence have moved our numbers more than any single breakthrough.
What advice would you give to someone thinking about starting their own business?
Start narrow. Ship fast. Do it now.