An engineering notebook on AI, systems, and everyday experiments.

观察到当代码成本趋近于零时，为一次性决策构建专用工具反而是最优策略。提出AI Native的本质是在信息获取成本坍塌后采用全新策略——从直觉驱动转向高分辨率的数据驱动决策。

Observes that when code cost approaches zero, building disposable tools for one-off decisions becomes optimal. Argues AI Native means adopting new strategies enabled by collapsed information costs—shifting from intuition-driven to high-resolution, data-driven decisions.

提出用户生成软件需要新的范式。软件公司不再交付成品，而是交付"生成内核"——包括核心套件、引导知识和杠杆工具集，专为AI使用而设计，最大化表达范围、意图保真和生成效率。

Proposes that User-Generated Software requires a new paradigm. Instead of finished products, companies deliver "Generative Kernels"—core kits, guiding knowledge, and leverage tools designed for AI to use, maximizing expressive range, intent fidelity, and generation efficiency.

分享3分钟语音prompt如何让AI完成资深科学家一整天的工作量。详解招聘、任务委托、入职培训、过程指导和产品验收五大管理环节如何释放AI的全部潜力。

Shares how a 3-minute voice prompt led to AI completing a senior scientist's full-day work. Details five management principles—hiring, task delegation, onboarding, process guidance, and product acceptance—that unlock AI's full potential.

AI"偷懒"的本质是LLM输出长度限制导致的注意力分散。Wide Research通过多轻量模型并行处理子任务、主LLM汇总的方式解决，分享为Codex构建该能力的设计思路。

Why AI slacks off on large tasks: LLM output length limitations cause attention drift. Wide Research solves this by parallelizing with lightweight models, then aggregating results with a primary LLM.

深入解析电车与油车底盘操控之争：油车通过战略自由塑造性格，电车用战术科技对抗物理惯性，两种截然不同的工程哲学。

A deep dive into EV vs ICE car chassis engineering, exploring different philosophies: ICE cars shape character through strategic freedom, while EVs fight physics with tactical technology.