Vincent: The Story of a Self-Taught AI Entrepreneur Changing the Game

100 AI Creators is a weekly series featuring conversations with Chinas leading minds in artificial intelligence. As technology evolves, their perspectives shed light on the ideas driving the AI era across borders.

When Vincent stood before a crowd of 1,000 investors at MiraclePlus and declared his ambition to build an agent talent marketplace worth RMB 1 trillion (USD 140 billion), it sounded bold. But that wasnt how Bonjour started. In the beginning, the product was born from something much rawer: anger.

Vincent, born in 2002, is tall, long-haired, and charismatic. He never went to college.

Two things happened when he was 15 that changed the course of his life. First, he discovered his idol, Steve Jobs:

Everything around you that you call life was made up by people who were no smarter than you. And you can change it, you can influence it, you can build your own things that other people can use. Once you learn that, youll never be the same again.

That quote hit Vincent hard. He realized, for the first time, that he too could create something that shaped the world.

Then came the second event: he was bullied in school.

Unable to keep attending classes, he spent his days holed up in the computer lab, teaching himself how to code. Later he transferred to a vocational school, where teachers left the more talented students to figure things out on their own. So he did: hed wake up, start coding, and keep at it until 10 p.m. No one noticed. In that silence, he honed his craft.

Three years later, he joined an internet company as a full-stack engineer. For a year and a half, he slept under a conference table in the office and worked at that same station during the day. But he was happy. When the first product he singlehandedly built from scratch reached the top of the iOS paid app charts, he had just turned 19.

Still, the anger never left him.

He never understood why he was bullied. Was it because he loved coding too much? Because he read books that none of the other kids read? Or because he was introverted and didnt quite fit in? It took him a long time to realize that it had nothing to do with him. It was just the cruelty of the crowd, punishing anyone who was different.

He also began to notice that he wasnt the only one. He met others like him: young people with talent and potential whod never gone to college but had mastered valuable skills of their own. Not all of them got the same lucky breaks:

One of my friends is a brilliant frontend engineer, but because he doesnt have a degree, hes stuck in a miserable, low-paying job where his talent goes to waste.

In the rigid systems of mainstream hiring, people like Vincent and his friends are filtered out before they even get a shot. Its not that were not good enough. Its that most of us never even get the chance, he said.

All of thisanger, exclusion, and the determination to buildeventually led to Bonjour.

Its his third startup. The first was a marketplace for artificial intelligence models that collapsed when a co-founder bailed. The second was an experimental social app that ran out of money after Vincent missed the fundraising window. At one point, the team worked from Starbucks, using the free Wi-Fi. When things got really bad, he kept the company afloat for a time with high-interest personal loans.

Looking back, he said the failure wasnt just about money. I didnt clearly explain the vision or what we were trying to build, he said. That left people lost in the grind.

Bonjour is different. This time, Vincent feels he has a clear direction. He believes that within the next five years, the world will see a trillion AI agents emerge. These agents, like humans, will become small but essential units of productivity. Bonjour aims to be the largest marketplace for these agents.

Like hailing a ride on Didi, users on Bonjour can request purchase services like design, development, or podcast editing. These services might be provided by humans or by AI agents. It doesnt matter. What matters is that the task gets done.

Today, Bonjour has roughly 40,000 users. And Vincent can now stand on a MiraclePlus stage and speak his dream aloud. Investor Lu Qi once called him the kind of founder MiraclePlus likes best, describing his team as relentlessly resourceful, referencing the grit to find a way through regardless of the obstacle.

AI Now! sat down with the 23-year-old to talk about Bonjour, agent-based networks, and what the next generation of work might look like. The conversation also spanned his chaotic youth, the pirate ethos that drives him, andin his wordsall the things he has stolen from his idol Jobs.

The following transcript has been edited and consolidated for brevity and clarity.

Betting on small teams and vast agent networks

AI Now! (AN): On stage at MiraclePlus, you said Bonjour wants to be the Upwork of the agent era. Thats a bold claim. What exactly does it mean?

Vincent: A lot of the time, people are not trying to hire someone. They just want to get something done.

But traditional talent platforms are built like job fairs: you have to post a listing, filter candidates, schedule interviews. Three months later, the task might still be pending.

Were flipping that model. Bonjour is building a high-frequency task transaction network. Whether you want to develop something, run a research project, or launch a campaign, you can place an order on Bonjour much like the way you would order takeout. The agent might be an AI or a person. The point is that they match the task, can start today, and deliver work in minutes.

And heres the wild part: that order process doesnt even have to be triggered by you. It can be triggered by your agent. Lets say you have a marketing agent that needs to post ten Xiaohongshu posts a month. It might not write the posts itself, but it could break the job down into subtasks like copywriting, formatting, image generation, and ad placement, and then hire other agents to do each part. You dont do anything. Just wait for results.

This is a totally different logic from working with freelancers. A human might max out at five to ten transactions a day. But agents dont have that limit. In theory, they can handle hundreds or thousands of transactions per day, all at once, with no wait time.

AN: So its more like a task platform, not a job platform or professional social network?

Vincent: Right, but we also dont want to make it too utilitarian or cold.

The current version of Bonjour still looks and feels like a product made by and for geeks. We have a tappable NFC card that lets you instantly connect with someone, customizable profiles, and a community shaped by what we call the pirate ethos. The starting point is social, but what were really doing is rethinking how agents are activated and assigned.

We dont see agents as assistants embedded inside apps. We see both humans and AI agents as nodes that can be scheduled and dispatched for work. Bonjour doesnt sell agents. What we sell is the output they produce. When someone orders an agents service, we take a 15% cut from the results.

AN: But the agents on Bonjour are still humans, right? Are AI agents really ready?

Vincent: Thats right. For now, were starting with human agents. There are currently a lot of designers, developers, and creators on Bonjour. Humans are intelligent agents too, but we believe the real breakout in the next five years will come from AI agents.

In the future, you wont be opening a website to look for services anymore. Youll simply summon an agent and ask it to handle tasks on your behalf. It will understand your intent and act on it.

That means agents will no longer be secondary features inside user interfaces. They will become the smallest units of interaction and delivery in the next wave of digital services.

AN: If people start relying on agents this often, wont that completely change how work and collaboration happen?

Vincent: I think its inevitable. By 2026, well see unicorn companies with just one human employee orchestrating a swarm of agents. Thats not my prediction. Its what the CEO of Anthropic is saying.

In that kind of structure, we wont be seeing another wave of big tech firms as we presently know them. Instead, it will be small, tightly focused teams operating at massive scale with agents doing the heavy lifting. Bonjour wants to be the transaction layer that powers that entire system.

From social to productivity networks

AN: But AI agents still mess up a lot today. Isnt the experience unreliable?

Vincent: There are definitely issues. Many AI agents arent reliable enough yet. Our current approach is to start by delivering tasks through humans, closing a revenue loop, and winning enterprise clients. That will give us the base we need to eventually help users train their own agents. These agents will eventually deliver for the B-side.

That said, current instability also points to three big problems, and three major opportunities.

First is the lack of persistent context.

Most agents today reset memory after each use. They are fast in one-off exchanges, but rephrase your question or change the scenario and they forget who you are. Its like having an intern who reintroduces themselves every day. Smart short-term, but they dont retain. Were investing heavily here to build agents that remember, that stay with you over time. Not just tools, but reliable partners.

This context continuity will become the real platform moat. Once built, its not just a prompt-based tool anyone can replicate, but an extension of your thinking.

Second is a misalignment in values.

Even if an agent completes a task, it may not do it the way you wanted. You might say you care about creativity and independence, but the result may still seem templated.

A good agent cant just understand what you say. They should be able to comprehend the vibe behind what you want. Thats something we emphasize when training Bonnie, our HR agent. We feed it a set of belief-based keywords so she can shape behavior from values, not just logic.

That leads to a different experience. Its no longer just about task execution. It starts to feel like working with a teammate who gets you.

Third is the absence of self-improvement.

Most agents dont learn. They finish a task, then reset. But reliable agents must remember what went wrong last time, and improve. We want agents to learn like humans, but without forgetting or inconsistency.

Thats why we believe the strongest platform wont be the one with the most agents, but the one with the most context assets.

The more context an agent can draw on, the better it can match tasks, deliver results, and lower the cost of trust. Thats whatll be scarce in the AI agent era.

AN: Bonjour started as a social platform for people, right? How did the shift to agent-based socializing happen?

Vincent: At the beginning, we just wanted to solve a longstanding problem: social interactions are too slow.

You have to add friends, make small talk, test the waters, build rapport. These rituals are really just slow collaboration. But in professional or semi-professional contexts, people often need faster matching.

Say Im looking for a frontend developer in Hangzhou who understands UI/UX and is familiar with Apples Human Interface Guidelines (HIG). Youre not going to find that on a traditional hiring platform. Those platforms are built around posting jobs and waiting for responses.

What I want is to say one sentence and have the AI bring me the right person instantly.

AN: That sounds like an introvert-friendly design, like passive, non-performative networking?

Vincent: Its not just for introverts. Plenty of people on our team do feel hesitant about putting themselves out there. We dont like posting, recruiting, or asking for attention. But that doesnt mean we dont want to be seen.

We just want a more restrained way to be visible and valuable.

AN: So when did you start thinking of agents as the atomic unit for Bonjour?

Vincent: That shift happened at the end of last year. I was talking with Lu Qi (from MiraclePlus), explaining what we were building, and he said something that clicked, that agents themselves are the future of work.

That line stuck with me. I realized we werent building a social network. We were building a productivity network.

So the connections on Bonjour arent about meeting more people. They are about getting things done, and agents are how we move from people finding people to tasks finding nodes.

AN: That implies a lot of human labor might start disappearing from the equation.

Vincent: I dont think people will disappear. Their role just changes, from executors to orchestrators. From workers to directors.

The pirate ethos is a vibe that you either feel, or dont

AN: Bonjour is your third startup. How does it compare to LinkedUp?

Vincent: LinkedUp was also about professional networking. But back then, I was idealistic. I didnt want to raise funds. I wasnt in a hurry to monetize. I just wanted to build something big and brilliant all at once.

Reality smacked us fast. Without funding, we ran out of cash quickly. We didnt invest in user acquisition, so traction stalled. And when that happens, morale collapses.

The biggest lesson was that its not enough to do the right thing. You have to do it in the right way. That means starting light, shipping fast, building momentum. Someone has to stick around. Someone has to be willing to pay.

So with Bonjour, we started with something very lightweight: a personal profile card. Everyone needs one. We designed a tap-to-connect card with a few keyword tags that make self-presentation easy, and let people see each other, need each other, and activate each other.

AN: It has been nearly a year since launch. Hows Bonjour doing?

Vincent: We launched on the first day of AdventureX last July. It was the first edition of the event, and its a large-scale hackathon for developers aged 1626. It blew up. Everyone at the event was talking about us. We were rushing to make tap cards for people while explaining our vision nonstop.

I remember, at 5 a.m. on the fifth day, users were still gathered at our booth. One of them, a woman who was also a founder, started crying while I was telling our story. She was a founder too and deeply related to our struggles. Then she sent us RMB 45,000 (USD 6,300) over WeChat, but not as an investment. She just wanted to support us so we could keep going.

That week, we gained 3,200 seed users. Over the next six months, that grew to 40,000. We eventually raised funding too, but Ill always remember that ladys RMB 45,000. That was our first round.

AN: That week at AdventureX, it sounds like people loved Bonjour. What do you think they were responding to?

Vincent: I think our values were clear from the start, and that attracted people with a similar mindset.

From day one, there has always been a bit of defiance in what were doing. Its not something we explicitly say, but something people feel when they join in.

AN: You often mention the pirate ethos. What does that actually mean?

Vincent: It means the refusal to be boxed in. A resistance to being told how things should work. I think thats part of the mindset for a lot of people born after 2000.

At AdventureX, we were telling people about our failures, how we think about social products, and why we believe the future is about tasks finding people, not resumes, not credentials. People got it. They felt the same doubts about the old ways of connecting.

Were young, restrained, and a little rebellious. Thats something people can sense immediately.

AN: Is this pirate ethos just the classic youthful rebellion, like the idea that everyone is a leftist at 20 and a conservative at 40? Or is there something truly different about your generation?

Vincent: I think its fundamentally different. We grew up with the internet. We didnt rely on parents or teachers to tell us how the world works. We had to search, sort, and judge for ourselves. So when we started forming our worldviews, we had a ton of reference points. We could test whether so-called authorities were legit.

You can call it generational or environmental, but either way, I think its shaped a whole new way of seeing the world.

AN: As a founder born in 2002, do you feel that some investors dont fully understand you because of the generational gap?

Vincent: Definitely. Some investors have challenged me directly. They ask what exactly is different about Gen Z culture compared to earlier generations.

But heres the thing: its a vibe. You either feel it or you dont. You cant explain it with logic or examples. If youre part of it, you just get it. If youre not in that world, or if your mindset isnt shaped by this generation, it probably wont make sense.

Of course, there are also investors who do understand us. They can sense the energy clearly. I think fundraising is like getting married. You cant explain why it works. You just have to find the right match, and thats the only way itll go far.

Stealing all the best stuff to make what matters

AN: How big is the Bonjour team now? How did you all come together?

Vincent: There are five of us on the core team, plus a few part-time contributors. Most of us met at AdventureX, actually. Theres me, two developers, someone in growth and ops, and an admin manager. Everyone was drawn in by the product and the idea. We just clicked.

Honestly, this is my ideal team structure. A small group of people, each powered by a bunch of agents. I think eventually well all train our own agents, basically our own operating systems, and make them stronger through our personal expertise.

AN: Tell us about your life in Liangzhu. That area has become a hub for AI-focused digital nomads. People say you guys rented a big countryside house for just RMB 5,000 (USD 700). Whats your daily routine like?

Vincent: Its not a villa or anything like that. Its just a two-story residential house. But its comfy, much better than last time when we worked out of Starbucks and used the free Wi-Fi to get through the day. I could write a whole guide on how to survive in Starbucks: which ones are best, where staff wont kick you out, and how to stretch a single drink for 12 hours.

Now weve got our own space. The downstairs is the workspace. I live upstairs with the two developers. Im always the first one up. Then I go shake the devs awake, and we get to it. We talk about what were building today, whats next. Then we just work, usually straight through to midnight.

Sometimes, someone blasts rock music out of nowhere. Lately Ive been playing Queen.

AN: Sounds like youve built a little startup utopia.

Vincent: It is, in a way. We all live here. Nobody really goes home. Theres no hard line between work and life. This is home and our workplace.

One of the devs cooks sometimes, when hes in the mood. Otherwise, we eat out. Liangzhus small, so everythings close.

AN: Do you think this lifestyle is sustainable? Can it last?

Vincent: Im not sure. Maybe one day well have an office. But I dont think well ever do the nine-to-five punch clock thing. Thats not us. What matters is doing work we really care about.

Compared to my last startup, the team now is much more stable. Weve got clearer direction, and stronger belief. Thats partly because I explained the vision early. We all know this matters. Were not just grinding for nothing.

And the users help, too. People love what were building. We get constant feedback, encouragement. That energy keeps us going every day.

AN: Doesnt the pirate ethos clash with having an office? Pirates dont clock in and out.

Vincent: Exactly. Steve Jobs said that being a pirate is about breaking rules and winning with less. That means always putting your very best into your product, no matter how you get there.

AN: From age 15 till now, whats the biggest thing Jobs has taught you?

Vincent: His answer to one interview question stuck with me. Someone asked how he knew when something was right. He didnt say user feedback or tech feasibility. He said: taste.

He said the reason the Mac succeeded was because the people who built it were also the best poets, musicians, zoologists, and historians, and they just happened to be elite computer scientists too. Even if they hadnt built computers, they wouldve created something beautiful somewhere else.

So Macs greatness came from blending the very best parts of the world into the product.

That really struck me. To me, taste isnt about style. Its a sensitivity to recognizing whats great, and the willingness to craft it into your work.

We try to do the same with Bonjour. We borrow Apples design language, WeChats account infrastructure. We shamelessly steal all the best stuff.

Like Picasso said: Good artists copy. Great artists steal.

Quick takes

AN: Whats the most shocking thing AI has done in 2025?

Vincent: For a while, I felt like progress had plateaued. Nothing matched the thrill of GPT-4s initial launch. But then DeepSeek emerged, and we saw a dramatic leap in reasoning capabilities. That was genuinely exciting. For people building on top of AI, these foundational advances feel like rays of hope.

AN: Whats one thing in AI you didnt believe in at first, but have since changed your mind about?

Vincent: I used to think text-to-image apps like Miaoya were kind of gimmicky. I didnt see real use cases. But then I saw people using them to take professional ID photos. Thats a real, practical use case.

It made me realize that AIs value often shows up in the most unexpected cases.

Now I think the real moat in AI isnt the tech itself, but whether it can make users feel genuinely understood.

AN: Do you have any contrarian takes on AI?

Vincent: I believe that context will become the most valuable asset in the future. Right now, everyone is competing for user data. But in the future, they will be fighting over something more powerful: the right to understand your context.

Imagine this: every person will have a local context warehouse that stores their interaction history, preferences, and habits. Only when you trust a service will you allow it to access that warehouse.

Thats true user sovereignty. Like keeping your money in a bank, but never giving up control.

AN: What are you most looking forward to in 2025?

Vincent: Definitely the collaboration between Jony Ive and OpenAI. Thats something worth getting excited about. Beyond that, I hope we see a breakout of innovation at the application layer.

AN: Name three books that influenced you the most.

Vincent: First has to be Steve Jobs by Walter Isaacson. Ive read it many, many times.

Second is Allen Zhangs compilation on the product philosophy behind WeChat. I reread it every year.

Third is Build by Tony Fadell. That book is basically my startup bible. Fadell was the father of the iPod and a key part of the iPhone team. He wrote in detail about how disruptive products are built from zero. Like how the idea for a touchscreen started even back in the iPod days. That book made me realize real innovation is born from obsessing over user experience.

100 AI Creators is a collaborative project between AI Now! and KrASIA, highlighting trailblazers in AI. Know an AI talent we should feature? Reach out to us.