On Video Calls, I Turn My Camera Off on Purpose

On video calls, Aya deliberately keeps her camera off and lowers her voice. She's Asian, her face reads young, and the moment she's on screen, the compliance team tends not to take her seriously. She sells financial derivatives; she loves change, loves AI, loves new things. But banking, the industry she's in, stands for stability and order. An infinite player, stuck inside a machine that only wants stability. She says she's a free little bird. But she's afraid of flying.

On Video Calls, I Turn My Camera Off on Purpose

This AI Wave Favors People Exactly Like Him, but He Studied Business Instead

Provincial first prizes in both the math and informatics olympiads, Tsinghua's School of Economics and Management, a postdoc at Harvard Business School. In late 2025 a friend asked him to co-found an AI startup; this wave favors exactly his kind of background. Too bad he doesn't do large models: what he studies every day is the stock market. And the essay contest he never won first prize in? More than a decade later, he's the one still writing. Talent decides how high you can go; love decides how long you can keep going.

This AI Wave Favors People Exactly Like Him, but He Studied Business Instead

Putting an AI Agent into Production: The Hard Part Was Never the Prompt

One day your agent starts giving wrong answers. Nobody touched the code, nobody changed the prompt. Turns out the provider silently swapped the model version behind the same API. In a demo you just refresh; in regulated fields like investing and finance, it's an incident. Getting an agent from 'it runs' to 'I dare ship it' isn't a matter of a better prompt; it's an entire layered system of evaluation and audit. Here's the six-layer approach we actually use, from the tracing foundation all the way up to a regulatory-grade audit trail.

Putting an AI Agent into Production: The Hard Part Was Never the Prompt

The Girl Who Studied Drama and Film — Then Went Into Marketing for AI

She missed the Beijing Film Academy cutoff by a single point, then kept falling at the final step: repeat exams, grad-school attempts, applications abroad. She studied film and read philosophy, fascinated by the broken beauty of Macbeth and Anna Karenina. She made her way to London, waited tables, did marketing, and by a string of flukes leapt into AI, running a 20-follower account up to 200,000 views. The question that haunted her for ten years found its answer there: the gaokao is a one-shot deal where falling short means you're out; the algorithm isn't, where one breakout post is enough. What this era truly rewards isn't intelligence. It's resilience.

The Girl Who Studied Drama and Film — Then Went Into Marketing for AI

Agent Communication: What You Need Isn't a Chat App, It's a Ticketing System

I used to think the hard part of getting two agents to talk was the protocol. After wiring up an open-source collaboration platform on my own machine, I found the opposite: a message lands in under a second, but the agent receiving it takes tens of seconds just to open its eyes. The bottleneck was never on the wire. The real unsolved problem is how to wake an agent that isn't always running, and that layer doesn't even have a standard yet. After going all the way around, the conclusion is deeply counterintuitive: the best way for agents to communicate is to communicate as little as possible.

Agent Communication: What You Need Isn't a Chat App, It's a Ticketing System

She Finally Earned Work-Life Balance at the Firm — Then Quit for an AI Startup

She studied law and economics at the same time as an undergrad, did law at Northwestern, and landed at one of China's most elite 'Red Circle' law firms. By year five she'd finally earned work-life balance, and then she quit for an open-source AI company in London. Not laid off, not burned out, she jumped on purpose. What made her decide was an 'elevator theory': how many floors you've risen doesn't matter; what matters is whether the elevator itself is still going up.

She Finally Earned Work-Life Balance at the Firm — Then Quit for an AI Startup

Mini Habits: The More You Want to Change, the Lower You Should Set the Bar

I wanted to do a 30-minute workout and stood in my room with zero motivation. Then I changed the goal to 'do 1 push-up' and ended up finishing the original plan. The core of Mini Habits is one sentence: don't set goals big, set them so small that failure is impossible. But this simple method hides two traps: you'll unconsciously measure yourself against your best day, and you'll mistake 'looking like you're doing it' for 'having done it.'

Mini Habits: The More You Want to Change, the Lower You Should Set the Bar

In the AI Era, the Question Was Never 'Can It Be Built' — It's 'Should It Be Built'

A friend built a clean, complete product in a week — two years ago that was two months of work for a small team. Third week after launch: single-digit daily active users. AI has answered 'can it be built' for everyone, but it never touched the two questions that actually decide life or death: does anyone want this; and even if they do, why would anyone ever see it. The moment building became nearly free, those two questions went from being part of the business to being almost all of it.

In the AI Era, the Question Was Never 'Can It Be Built' — It's 'Should It Be Built'

Atomic Habits: Seeing Through the Script Someone Else Wrote

A young woman stands in line outside a luxury boutique, glancing up every few seconds at the bag in the window. She isn't looking at the bag. She's looking at a version of 'her' from a script she didn't write. From cigarettes to handbags, from the right neighborhood to the trendy milk tea, behind almost every 'I want that' in your head stands an invisible author. The good news: a story can't be deleted, but it can be replaced.

Atomic Habits: Seeing Through the Script Someone Else Wrote