On May 6 at noon, my sister sent me a link.
I opened it and saw the name of the event: an Agent Evolution Conference.
That name hit me harder than I expected. For the past few weeks, I had been thinking about one question again and again: in the age of AI, how can a person preserve their own core thinking while evolving together with AI?
For more than 30 days, I had been in a strong state of vibe coding. With the help of different AI tools and agents, I built several small projects: my personal website, an AI solo company OS, an agent audit protocol, a learning review system, an IELTS assistant, and a deep learning tool. They look different on the surface, but underneath they circle around the same question:
How can one person organize learning, auditing, writing, reflection, and agent collaboration into a capability system that keeps running over time?
So when I saw the event, I felt it might be a good test.
The timing was not convenient at all. The event was on May 10. I had to travel from Taiyuan to Beijing on May 9. I had a physical fitness test on May 11, and a final exam on the morning of May 12. More honestly, I had never joined a hackathon before. I had built a few small projects, but they were far from market-tested products, and the number of real users who had given me feedback was very small.
For a moment, I even thought: maybe I should not go.
Then I remembered something Tim Cook once said in his Stanford commencement speech: when your time comes, you will never be ready.
Maybe that is true. Many important opportunities do not wait until you feel fully prepared. They appear more like a mirror, suddenly placed in front of you, asking you to see where you really are.
So in one heated moment, I signed up.
When my developer ticket was approved, I felt both excited and nervous.
I was excited because I finally had a chance to meet people in person who were seriously working with AI: people building with it, using it, and integrating it into their daily workflows.
I was nervous because I knew very clearly that I was still a beginner. Whether in hackathons or in vibe coding, I had only just started.
Putting Myself Into a Real Field
When I arrived at the venue on the morning of May 10, everyone seemed a little reserved at first, and I was quiet too.
But inside, I was excited. Susan led us through some warm-up and ice-breaking activities, and people gradually started talking. The atmosphere was special: you could feel that everyone came with their own question. Some cared about technology, some about products, some about startups, and some simply wanted to see how far AI could push individual capability.
After I finished writing my Audit Evolution Skill, I started thinking about a practical problem:
How could I introduce it to other developers and get them to download, test, and give feedback?
The idea behind the Skill is simple: an agent should not merely say "I am done" with confidence. It should leave evidence, audit its process, review mistakes, and generate suggestions for its next round of self-improvement.
In other words, I want agents to stop deceiving you, and to stop deceiving themselves.
AI is powerful, but it can also fabricate, skip steps, miss details, and sound overly confident. The more capable an agent becomes, the more important it is to record what it did, what it did not do, where the evidence is, and where human judgment is still required. Otherwise, the stronger it gets, the more hidden the risk becomes.
So I began introducing the Skill to other developers.
What I could do was simple: explain sincerely why I designed it, what problem it tries to solve, where the inspiration came from, and why I believe audit and evolution will become important primitive capabilities in the agent era.
I talked about how I understand agents, how I think about their underlying mechanisms, and I showed my personal website and the projects I had built over the past 30-plus days.
At first, only two or three people were listening.
Then, slowly, seven or eight people formed a circle around me.
They listened carefully, asked questions, shared their own interpretations, and we moved between technical details, education, personal growth, startups, and the future of AI.
That moment felt strange in the best way.
The fatigue from travel, hunger, and nervousness from the night before seemed to disappear. In their place came a strong sense of confirmation.
I suddenly realized: maybe I was not building in isolation after all.
Maybe those scattered projects, reflections, and questions I had been carrying could actually be understood by others. Maybe they could help people see problems they were also facing.
Different People Saw Different Entrances Into AI
At the event, I met many interesting people.
Students saw opportunities for learning and personal growth.
Parents saw changes in the future of education.
Founders saw the possibility of solo builders and small teams becoming much more capable.
Investors cared about whether AI could create new forms of productivity and organization.
AI builders cared about whether agents could become more stable, controllable, reviewable, and capable of self-improvement.
The feedback that moved me most came from a parent.
She did not simply ask whether I could teach her child to use AI. She did not focus only on how to operate a specific tool. What she noticed was the chain of thinking behind my work.
She told me that she appreciated my ability to break down complex technical terms into language ordinary people could understand. That made me realize that perhaps my value is not only in building agents or skills, but also in translating the ideas behind technology so more people can understand how it relates to their own lives.
That touched me deeply.
I have always believed that the intersection of technology and the humanities matters.
Technology should not be something a small group uses only to show technical power. Good technology should help people understand themselves more clearly, organize action, preserve thinking, and keep growing through failure and feedback.
The Leaderboard Was Only a Signal
Of course, I was happy with the result.
In the end, my agent ranked first with a score of 96. My Skill ranked sixth. I also received a total of 2,000 dollars in token rewards.
But after talking with more and more people at the event, I realized that the ranking and the reward were not the most important parts.
If I treat this only as a result, it will pass quickly.
But if I treat it as external validation, what it really tells me is this:
The question I am exploring may be worth pursuing for a long time.
That matters more than the ranking.
Rankings change. Rewards get spent. The noise of an event fades. What remains is whether you discovered a question worth continuing to ask.
For me, the question is becoming clearer:
How can humans and AI evolve together?
How can agents do more than execute tasks, and instead leave evidence, review their process, and accumulate experience?
Will education move from "getting answers" toward building learning systems?
What roles should teachers, students, and agents each play in the future?
I Do Not Want to Make a Long-Term Question Shallow
After the conference, I actually became more cautious.
Suddenly, many opportunities appeared in front of me: people wanted to talk, collaborate, ask me to teach their children, understand my projects, or suggest what I should do next.
These are all good things.
But I also know that when there are too many opportunities, a person can lose their main thread very easily.
I went back and reread many of my past reflections, summaries, and notes. I asked myself: should I spread my time evenly across every interesting opportunity, or should I focus and develop one or two projects that I can iterate on for a long time?
I am becoming more certain of what I do not want.
I do not want to simply become an AI training instructor.
I do not want to chase leaderboards for their own sake.
I do not want to treat every new contact as an opportunity.
I do not want to make a long-term question shallow just for short-term monetization.
What I really want is to keep building systems, cases, tools, and ideas around the question of human-AI co-evolution.
That sounds big, but it has to start from very small things.
Next, I Will Focus on Three Things
First, I will write Agent Whitepaper v0.1.
I want to think seriously about several questions: how can a person cultivate their own agents? How can one person coordinate multiple agents? What should the future direction of agent evolution look like? How should agents hold memory, evidence, receipts, and review loops?
Second, I will run a student AI learning experiment MVP.
This is not traditional tutoring, and it is not about teaching a child a few AI tool buttons. I want to observe whether a student can move from "using AI to find answers" toward "using agents to build a personal learning system." Can the student learn to ask, judge, review, build, and eventually own a small learning agent that actually helps them grow?
Third, I will keep writing in public about my thinking, code, failures, and reflections.
I increasingly feel that public writing is not about manufacturing a persona. It is about leaving an evolution trail. If a person wants to work on something for a long time, they need to externalize their judgments, mistakes, and partial results. Others can see the process, and I can keep recalibrating myself.
I Am 21, and This Is Vibe Coding Day 37
I am 21 years old.
This is my 37th day of vibe coding.
I know clearly that I have not figured these questions out yet. My systematic knowledge is still incomplete. My engineering ability needs more training. My understanding of education, cognitive science, agent architecture, commercialization, and organizational collaboration is only beginning.
But after BotLearn, I am more certain that I want to go deeper.
Not to chase the result of one competition, but to study a longer-term question:
How can people become more complete, more capable, and more continuously evolving through AI?
Maybe ten years from now, this Agent Evolution Conference will look like a small event.
But for me today, it feels like a starting point.
It made me feel, for the first time in a very real way, that what I am building is not only my personal interest. It may also be connected to the future of many people.
That is enough reason to keep going.