Madhu Guru：评估即产品规格，而非事后 QA
The most common mistake teams make is to treat evals as an additional QA step once the agent is built.
AI 导读
团队最常犯的错误，是把评估当成智能体构建完成后的额外 QA 步骤。 AI 产品本质上是不同的。 你的评估就是你的产品规格。
正文
当前显示原文，中文全文译文尚未就绪。
The most common mistake teams make is to treat evals as an additional QA step once the agent is built.
AI products are fundamentally different.
Your evals are your product spec.