This is the first episode of my podcast. Our guest is Weng Wei — known on Zhihu as "Gamer Weng Wei," a top food writer whose analytics ironically categorize him under politics and economics. His tagline: "A CTO who doesn't understand food isn't a good keyboard warrior."
He graduated from NUS Computer Science in 2006, rose from junior developer to architect to CTO, helped take a cross-border e-commerce company from Series A to a Nasdaq listing, then became regional CTO for Southeast Asia at an A-share listed company. After two decades in pure tech, he went back for a master's in economics.
Now in his forties, he says this is his absolute peak — in learning speed, output, and productivity. Because of AI.
From managing dozens to paying a subscription
As CTO, Weng Wei used to manage teams of several dozen — frontend, backend, ops, QA, product. Now he does the same work alone with AI.
"I didn't write code before, and I don't now. Before I told human engineers what to build, now I tell AI. The management work is identical — discussing technical architecture, product strategy, business positioning. Sometimes when AI can't get a critical piece right, I still jump in myself. Exactly like being CTO. The difference? Before I needed extremely expensive headcount. Now I pay a subscription fee."
The faster AI gets, the slower he goes
After a year with AI, Weng Wei made a counterintuitive choice — to slow down.
"AI's implementation speed is too fast, I can't even digest it all. Sometimes I build this and that without thinking — but what's the point? Should I even be doing this? I haven't thought it through."
He prefers typing over voice input because typing forces deliberation. "Efficiency has already exceeded what my brain can process. I need to make sure I'm still thinking."
He saw a post that morning: in the AI era, as long as you learn slowly enough, there's a lot you don't need to learn at all. He agreed. Over the past year, he avoided MCP and several other hot concepts. Six months later, the industry started turning against MCP too.
What is taste? Start with food
I asked him — everyone says taste matters most in the AI era. How do you define it?
"Food obviously requires taste. How do you build it? First, you eat a lot. If you haven't tasted flavors that exist, how can you judge?"
Same with technology. The book that influenced him most was "The Design and Evolution of C++" — a language he never used professionally. "That book traces every problem C++ faced at each stage and how they solved it. Once I saw the full arc, technology became three-dimensional to me."
Then he gave a cooking example: why do you add cold water three times when boiling dumplings? Because old wood-fire stoves couldn't quickly lower temperature, so you added water. With a gas stove, just turn down the flame.
"Most people just memorize recipes. You need to think about why a technique exists, abstract the underlying principle, then reconstruct the method for a better result."
He later learned this concept had a name — first principles. "I only found out after Elon Musk made it famous. Kind of funny."
This AI revolution will exceed all industrial revolutions combined
Weng Wei isn't an AI scientist, but he sees things differently through an economics lens.
"Keynes predicted roughly 100 years ago that in a century, society would reach a point of surplus productivity where people wouldn't need to work. I think that prediction is coming true."
"Many say old jobs disappear but new ones appear with every technological shift. I think that's wishful inertia. This time is genuinely different."
Why? "It's a demand problem. Past productivity gains created new desires, which created new jobs. But now supply is oversaturated and demand is declining. Japan's low-desire society, China's 'lying flat,' anti-consumerism — demand is falling. Supply up, demand down, jobs disappear."
The future's problem isn't abundance or scarcity — it's inequality.
Singapore should be a small but complete sparrow
On Singapore's role in the AI era, his view was clear:
"Singapore's success, honestly, is mostly thanks to its peers underperforming. It's not exceptionally strong anywhere — it's just lucky enough to avoid internal dysfunction."
Two critical judgments about the future: First, Singapore can't cling dogmatically to Lee Kuan Yew's playbook. Second, Singapore's existential question is whether it can continue not picking sides as geopolitics intensifies.
On AI industry strategy, he believes Singapore should go deep on the application layer with a full-stack approach — education, talent, industry development, financing, policy. "Singapore can't be big, but it can be strong. Can't build foundation models? Fine — do more applications."
He's also bullish on nuclear energy: "AI is ultimately an energy problem. Singapore should boldly pursue nuclear — if you can't build on the island, negotiate with Malaysia."
*This article draws from EP01 of my Chinese-language podcast on AI founders going global.