All tags
Posts tagged with "AI Agent"
Intelligence Is Not Authority: Designing the Trust Boundary for Agentic AI
Starting from Satya Nadella's insider risk argument: why the agent execution plane and the authority control plane have to be separate, why identity is not authority, and which layer enforcement belongs in.
Intelligence Is Not Authority: Designing the Trust Boundary for Agentic AI
從 Satya Nadella 的 insider risk 論述出發，談 agent execution plane 與 authority control plane 為什麼要分開、identity 為什麼不等於 authority、以及 enforcement 該放在哪一層。
AI Doesn't Define Seniority. Engineering Judgment Does.
會不會使用 Claude Code、懂不懂 context window 或 tokenization，都不能直接決定一個工程師是否資深。AI 讓程式碼產出變便宜，但工程判斷、驗證能力，以及知道什麼值得做，仍然是資深工程師的核心價值。
AI Doesn't Define Seniority. Engineering Judgment Does.
Knowing Claude Code, or knowing how context windows and tokenization work, does not by itself make an engineer senior. AI made producing code cheap. Engineering judgment, knowing what to verify, and knowing what is worth doing are still what senior engineers are for.
Agent Identity Introduction
Why an agent needs an identity of its own, why identity is really four separate questions, what the ideal state looks like, and what a single tool call should actually be checked against.
Agent Identity Introduction
為什麼 agent 需要自己的身分、身分其實是四個不同的問題、理想狀態長什麼樣，以及一次 tool call 到底該被驗哪些東西。