Entrepreneurship, intent, data, security, human nature, org design and the development process. Each essay comes in two forms: the original text, and an interactive version you can play with as you read.
The essays are written in Chinese.
Why Silicon Valley-style fundraising inverts the logic of building — and how investors, founders, and talent can each act rationally yet add up to a collective zero. Field notes from what I have seen, tried, and learned.
Brain-computer interfaces, eye tracking, voice, EMG — every flashy new input modality runs into the same wall: the weight of habit. The real game isn't swapping hardware; it's reading what you actually want to do right now. That's intnt.
Every click, every dwell, every late-night search is a leak of intent. Stitch the fragments together and you can know what someone wants before they ask. That's predictive commerce — and the very thin line that runs through it.
Most of the security you've enjoyed was underwritten by one fact: attacking you wasn't worth it. AI has floored the cost of attack — and also the cost of defense. The danger isn't the technology; it's which side the scales tip toward.
What users say they want is never what they actually want. Peel the requirement back layer by layer and you find: the surface varies endlessly, but there are only a handful of strings underneath. AI changes how those strings get plucked — not the strings themselves.
Once code gets cheap, the decisive edge is the combination of 'deep demand insight + complete delivery' — and those two things almost never live in the same person. Why the structure of a law firm turns out to be exactly the right answer for this moment.
What does 'owning it' actually mean? Between AI-generated output and something you dare to ship, there is a gap. Whether you cross it by 'understanding it yourself' or by 'independent verification' — that is the judgment call we make on every project.