I build AI products and write down what I learn.

我做 AI 产品,把学到的写下来。

I’m MJ. Mostly I write about context engineering and how LLMs behave once they are inside a real product.我是 MJ。写的主要是上下文工程,以及大模型放进真实产品之后的表现。

Latest essay最近一篇 · 2026-02-17The General Tso's Chicken Theory of AI Adoption
NOW THINKING · bureaucracy as organizational memory LAST RUN · 2026-10-10 · 4.8 km PORTFOLIO · 10 positions since 2026-01-02

Running 跑步

Every run since 2021, synced from Strava.2021 年以来的每一次跑步,从 Strava 同步。

618Runs
3,968Kilometers
Open the running log →打开跑步记录 →

Portfolio 组合

A paper-trading portfolio, priced every weekday.一个模拟盘组合,每个交易日更新价格。

10Positions
+35.39%Since 2026-01-02
  • MU
  • NVDA
  • GLW
  • RMBS
  • INOD
  • SANM
  • PLAB
  • TTMI
  • AEP
  • HROW
See all positions →查看全部持仓 →

About 关于

I care about how AI products actually work: why LLMs hallucinate, how context shapes what a model does, and why some AI rollouts stick while others stall.我关心 AI 产品到底是怎么运转的:大模型为什么会幻觉,上下文怎样决定模型的行为,以及为什么有的 AI 项目能落地,有的停在半路。

What I write comes from shipping AI products at work and from helping teams adopt AI in large companies.这些文章来自我在工作中交付 AI 产品的经历,也来自帮大公司里的团队把 AI 用起来。

I write about
Context engineering AI product management Enterprise AI adoption LLM behavior
I build Run Card Generator ↗An iOS app that turns your running stats into shareable artwork.

Ask MJ

Answers come only from what I’ve written. If it isn’t there, it says so.只根据我写过的内容回答;没写过的,它会说不知道。