← Back to Home · All Annual Letters
June 6, 2026
Hey there,
This is Liwei Wu again. It has been a year since my last letter, and what a year it has been. Today, June 6, marks three years since TradingTransformer Technologies LLC was incorporated in Delaware on 2023/6/6.
The most important thing that happened this year is not a number. After years of having no product to put in your hands, we finally shipped one: NoArk — Infinity Memory, our general-purpose AI agent, is now officially live on the App Store — on every iPhone, and on iPad, Mac, and Apple Vision Pro too. For me, this milestone matters more than any return we posted. A return is the past; a product people actually use is the future.
That said, the numbers were good as well. In 2026 we generated roughly 90% return in the Shanghai stock market — our third market, after the U.S. and Hong Kong in 2025. That the method generalizes to a third, very different market is strong evidence we’re onto something real, since generalization is the true test of whether something is genuinely intelligent or merely lucky. But if I had to pick one thing I’m proudest of this year, it’s NoArk, not the percentage.
A few things have been on my mind this year. Let me share them.
Last year was a great year for vibe coding. I was probably an earlier adopter of agent mode than most — well before the term “vibe coding” was even coined. What surprised me is who adopted it first. You might expect young students to embrace it before senior engineers, but in my experience the opposite was true: upper management often adopted it earliest. If you asked me why, I’d point to a slogan I learned as an intern at Facebook a decade ago — “move fast and break things.” Breaking things simply isn’t that scary in most cases, especially for people senior enough to have broken things before and survived. Vibe coding is a genuinely promising direction when done right; just look at how many stars a vibe-coding repo can rack up on GitHub, returns or no returns.
So I’ve been asking myself: is vibe trading the next wave after vibe coding? D.E. Shaw — the firm where Jeff Bezos worked before founding Amazon, for anyone who doesn’t recognize the name — is already hiring for roles that look a lot like this.
Running TradingTransformer, I’ve felt it personally: AI is starting to outsmart us. Take SanDisk, one of our example trades. On paper I’m supposed to be the expert here — and yet the AI read that position better than I did. It grasped the non-linear dynamics at play while I, the so-called human expert, was still reasoning in straight lines. That single trade convinced me how poorly we humans reason when the world changes at non-linear speed. We like to say the AI is “grokking” something profound, when the simpler explanation may be that we are just slow to understand a world moving faster than our intuition. After watching enough of these, I’ve stopped doubting it: vibe trading is here to stay.
It has become clear to me that AI makes more informed decisions than I do. That leaves an uncomfortable question: do I actually have the authority to accept or reject the AI’s trades, or is that just an illusion of control? Increasingly I suspect I should do what I already do in vibe coding most of the time — accept the diffs, and only revert when something is truly, obviously wrong.
Maybe I should start calling it agentic trading rather than just an algorithm. Whatever we call it, the Shanghai result I mentioned above wasn’t a fluke: over the past year it has worked exceptionally well in China — even more so than in the U.S. I can’t say for certain why. My best guess is that there’s simply less competition: the Chinese market still has a large population of amateur traders, whereas the U.S. market is dominated by professionals.
Let me also clarify something about “trading” versus “investing,” since people often treat them as opposites. To me they are two sides of the same coin. How can you invest without buying and selling? How can you trade without holding? There’s no real difference between the two — words like “value” or “algorithmic” just add flavor to the same underlying act.
In last year’s letter I admitted we were still nowhere close to a consumer product. A lot can happen in twelve months. Not only is that product finally here — I’ve also changed my mind about what our highest priority should be. Launching NoArk, and what I think of as our fiat lux mission, now rank above improving the agentic trading algorithm or simply making more money.
Living in Shanghai is like attending business school every single day; everyone here is a businessman, and you learn something just by watching how people operate. Those observations went straight into the product. NoArk — Infinity Memory is a general-purpose AI agent that lives in your pocket. You chat with “Charlie” — yes, named after Charlie Munger — across more than 40 preset workflows spanning markets, productivity, travel, and everyday life admin, using text, voice, photos, or files. You can also sit down with AI mentors modeled on personal heroes of mine: Charlie Munger, Claude Shannon, Steve Jobs, and dozens more. Everything is stored on your own device — we collect no data.
The name carries a double meaning. It echoes Noah’s Ark — a vessel that carries what matters safely through a flood of change. But read it as “no ark” and it becomes a challenge: in a fast-moving agentic era, no one is coming to save you with a ready-made ark. You have to become your own ark, and NoArk’s agents are there to help you do exactly that.
We launched on the App Store before Android, tailored to the high-end market. Honestly, I still don’t understand how Gemini Spark charges $100 a month for fairly basic agentic functions. NoArk is free to download, with paid plans starting at $9.99/month — 50% off as an early-bird offer — for features I believe already work better than what’s out there. This is where I spend most of my time now, because we are in a once-in-a-generation opportunity. The agentic era is arriving, we are still very early, and that genuinely excites me.
This is the heart of that fiat lux — “let there be light.” I plan to allocate half of our future profits toward education, to help bridge the wealth gap that AI companies have created and widened, in America and in the other countries I’ve traveled through. Making more money has never been the point; using it to light the way for others is.
I’m always surprised how many people know Claude — the AI — but not the brilliant man it’s named after: Claude Shannon. He is one of my role models, and a big part of why I founded TradingTransformer. Beyond founding information theory, Shannon was a remarkable investor, reportedly compounding over 28% annually for decades. If you’d like to read more about him and what we’re building, visit tradingtransformer.com.
Back in the 1980s, he predicted that one day AI would generate returns comparable to Moore’s Law: if AI is truly intelligent, then roughly every 18 months it should be able to double the money it manages. That idea has stayed with me ever since — it’s the bar I hold our own work against.
Still, a target to chase — even one as audacious as Shannon’s — has never been what gets me out of bed. What does is purpose. I work far harder once I remember the real goal: using AI to narrow the wealth gap by creating more equal access to education, like the agentic education we’ve built into NoArk. It’s the same mission I felt at Coursera years ago, and it moves me in a way a purely money-driven mission never could. Setting a higher purpose, I’ve found, matters more than any number.
On a personal note: I’m working day and night these days — often down to one meal a day, coding past midnight, sometimes not leaving home for an entire day. And yet I’ve never felt freer. With AI alongside me it’s as if I’m traveling the world without leaving my desk — a genuinely surreal feeling. I’m even writing a paper right now with AI, side by side with a young and ambitious intern. Life, honestly, feels wonderful; some days I think I could live to a hundred.
The serendipity is the best part. With AI’s help I recently connected with an American New York Times columnist here in Shanghai, and through him I’ll be joining a reading club in the city — cooler, I have to admit, than the Prada fashion show I was invited to earlier this year. I can’t imagine a better life than the one I have now — except, perhaps, my eventual dream of returning to university as a professor, mentoring young people. And there’s more to look forward to: at the end of this month I’m flying to Los Angeles to watch the football World Cup — held only once every four years — for the first time in my life, and then on to San Diego for ACL 2026.
Before I go, I'll leave you with something special: a song called “Better Days Ahead,” written and performed entirely by NoArk. Give it a listen.
Machines that learn and grow with us, a future we can share…
It gives us time to love and dream, more than we can give;
no need to fear the change, my friend — it is here to set us free,
a better world for everyone, for you and for me…
A symphony of code and heart…
And the future's bright, it is in our hands; AI brings us to new lands…
together we will understand.
As always, if you ever find yourself in Shanghai, let me know — maybe we can hang out for fun.
Dr. Liwei Wu
CEO, TradingTransformer Technologies LLC
Shanghai, China
2026/6/6