Zhu Haoyu, After Hosting China's Largest Hackathon | elselier

At the start of every new cycle, VCs do the same thing: find the right people.

What do the right people look like?

There was a time you could trust schools or majors—Ivies and overseas experience, or years of polished résumés. But when skills, cognition, and IQ have become standardized commodities that AI can provide on demand, the "perfect résumé" no longer distinguishes the extraordinary from the ordinary.

The outlier is what's scarce in this era.

"Looking for outliers" has long been a Silicon Valley VC mantra, but in the past three years it's been elevated to unprecedented heights—and VCs have grown more aggressive in their search. "I ask about childhood a lot. I want to know: Were you an outsider? Were you the misunderstood kid? Outliers aren't born at the center of the bell curve." — Shaun Maguire, Sequoia Capital

"Venture capital is one hundred percent an outlier's game. If you're investing in 'reasonable' ideas, you're destined to fail. You have to find people who are radical, non-consensus, and driven by the irrational." — Marc Andreessen, a16z

"True outliers aren't people solving problems—they're people overturning reality. What true outliers build initially looks like a mistake to everyone else, including AI." — Brian Singerman, Founders Fund

Several new fund leaders in China have recently touched on similar concepts in various ways—"divergent thinking, childhood hardship," "different from the mainstream but with their own obsessions," and so on.

Of course, judging people is the extreme tail of probability in VC investing. Harder to pin down, harder to claim anyone has mastered.

We're launching a new column—「elselier」—to try to distill some of the unique qualities in founders.

"Else-lier" is a made-up word. Using "else" in place of "out" is partly to signal this column carries a certain elsewhere subjectivity; partly to suggest that this group isn't just exceeding (out), but leaping outside traditional quadrants. That's the meaning we're trying to build with an alternative (else) framework.

We've drawn on some VC perspectives, but we don't fully align with them. Below are our annotations—you're welcome to have your own:

Hustle: Street smarts, resourcefulness, getting shit done

Obsession: Fixation, singular focus

Narrative: Storytelling, reality distortion field

Ego: Self, the non-negotiable

Sensitivity: Attuned to the direction of the current, momentum

Taste: Aesthetic judgment, knowing what's good

These terms aren't purely complimentary—they carry something of both light and shadow.

Coincidentally, the first letters of these six words spell: HONEST. We'll have founders who enter this column self-assess across these six metrics, and we'll also score them in the article. The gap between self-assessment and external assessment might bring the judgment closer to honesty.

If you're that elselier, come find us 👇

[Scan QR code to fill out questionnaire]

Below is the first installment of this column:

Wu Ruirui @Ruiruiwow__

1. Zhu Haoyu, Ryan Zhu, 19, from Hangzhou, ENFP/ENTJ.

2. After disappearing from China for five months, Ryan has finally resurfaced with his new venture, Photon AI. The project wraps Agents into iMessage, WhatsApp, Slack, and X accounts, and teaches Agents how to interact with humans—so users can call up Agents as naturally as chatting with a real person.

3. The more Agents explode, the more they need underlying infrastructure. Ryan sees himself building interaction infra: in a future where humans and Agents coexist in a 70-30 split, users will need easy, human-feeling ways to interact with Agents, plus a personal assistant that can complete tasks across apps. Embedding as an IM account is the most logical form factor.

4. There are companies in Silicon Valley doing similar work, most recently Interaction, founded by two Gen-Z guys. It raised $15 million led by General Catalyst, with investors including founders and executives from PayPal, Stripe, Dropbox, Google, and OpenAI, at a $100 million post-money valuation. In Silicon Valley, more and more Agents are being deployed through social accounts—series.so, text.ai, and others.

5. Ryan's difference from them: they're building products; he's building infra, aiming to help Agent companies create products like Poke. Even in Silicon Valley this is a very early-stage赛道—Agent infra's flourishing will necessarily lag behind Agents themselves, and Poke is already one of the most prominent companies in the space.

6. Photon AI charges $500 per account per month, has ten customers, and has received investment from Cory Levy (founder of Z Fellow) and Kartik Hosanagar (Wharton professor).

7. Ryan is the founder of AdventureX, billed as "China's largest hackathon." But with some drama: after July, he became mired in controversy over leaking participant information and alleged fraud. "Exposing" him briefly became a traffic hack on Xiaohongshu.

8. His response to AdventureX 2025: a globally top-three hackathon, absolutely not a failure; 80% of the problems were intentionally designed to filter for true idealists; even if it had "bad effects" on him, so what—life has infinite bullets.

9. This seems at odds with what many see as the facts. He appears to possess a reality distortion field that can spin everything into his narrative.

10. What infuriates him is that one of his attackers was a hackathon judge—he "felt betrayed." He analyzed several of his attackers: not young anymore (around 25), main gig is Xiaohongshu influencer, probably no girlfriend—and found peace: "Their lives are pretty failures, so I understand them lashing out."

11. After summer he went to school in the US, though he wasn't really attending classes. He and his cofounder (Penn M&T) escaped campus and moved to San Francisco to build. Chinese founders and investors in Silicon Valley typically cluster in Palo Alto and Chinese enclaves; he believed this made real integration into American society impossible. So he threw himself in with locals, even though his English wasn't strong.

12. In elementary school, Ryan was the model Chinese good student. Then one day in seventh grade, lying in bed during an online class, he suddenly felt "too tired, never want to listen to another lecture." To kill time, he taught himself Python. From then on he was disenchanted with exam-oriented education: What can't be self-taught?

13. His mother, who works in a traditional industry, couldn't accept her son's sudden aversion to school and panicked to the point of hiring a "master" to perform an exorcism. The "master" placed a purple-gold pagoda on his desk to suppress evil spirits, with limited effect. A year later, he shot to fame with AdventureX.

14. But some darkly comic things happened this year. Before the event, someone had warned him not to block the large stone at Hangzhou's Hubin Center (the AdventureX venue) that the "master" had blessed, but he put up a display board there anyway. Soon after, a supplier got in a car accident. Everything fell apart.

15. Zhu Haoyu's emergence is richly symbolic. Chinese VC's generational anxiety peaked this year—virtually every fund is desperately hunting for young people. But his standing out from the crowd, I think, has to do with his high-fidelity replication of the Silicon Valley hackathon ethos—he identified the moment when China's AI industry most needed Silicon Valley spirit, and with boldness and street smarts almost unbefitting an 18-year-old, played the missionary.

16. But here's the thing: last week, Zhu Haoyu met with first- and second-tier dollar investors, 25 firms in total. But running a hackathon and being a founder are different. One partner who had attended AdventureX early on never replied to his WeChat; one investor who had posted "encouraging young people to think highly of themselves" told him directly: This is less cool than AdventureX.

17. But then again, Zhu Haoyu is still only 19.

「elsewhere」chaotic take:

He rated himself too modestly on some dimensions—Hustle I'd score at least a 4. But this method of maxing some out and giving himself 2s on others is very Ryan indeed.

Cover image: Provided by interviewee