AI 智能体 are not as beautiful as the marketing hype. Many companies already slowing down or stopping because token, GPU, cloud and maintenance costs become too high.
The real mistake is using LLM for everything. Simple tasks should run through parsers, rule-based automation or detached 系统, while AI is only used when really needed.



Ruang pembaca
Apa pendapat anda?
Komen disemak sebelum diterbitkan. 电子邮件 diperlukan untuk moderasi dan tidak dipaparkan kepada pembaca.
rule-based这个说法我要拿回去跟同事讨论。
文章对rule-based automation or detached 系统提供了很好的起点,值得继续跟进。
这篇文章把simple tasks should run through讲得比一般的AI介绍更具体。
视觉和结构让rule-based的概念更容易掌握。
关于simple tasks should run through的实际落地部分最吸引我。
如果可以继续说明rule-based的真实案例,我会想继续阅读。
收藏了,主要是为了rule-based。
关于rule-based automation or detached 系统的例子很实用,适合团队继续讨论。