把工作交给智能体：人只负责引导、补上下文、修错误
Proton CTO 那份 AI 宣言被改成了人人能用的工作方法，三条职责讲得挺清楚。
这篇内容改编自 Proton CTO 的 AI 软件工程宣言，套用到日常电脑工作。核心是三条职责：引导智能体做事、改进它们拿到的上下文、修复反复出现的失败。文中还提醒，只加快单个任务可能把瓶颈挪到流程别处，要优化整个工作流，并把有效做法分享给团队。人的专业判断和担责仍然不可替代。
Our role in work now is:
1. Guide agents 2. Improve their context 3. Fix recurring failures
Other things:
- Guide agents to do better work faster. - Agents should keep working while you're busy. - Improve the whole workflow. Speeding up one task can just move the bottleneck elsewhere. - Share what works, so improvements help the whole team, not just stay in one chat. - Expertise still matters. AI handles more execution. You provide judgment, experience, and accountability.
This is an interpretation of the Proton CTO's AI manifest for software engineering, adapted into everyday computer work. https://t.co/m7M8zoGmmB
