12-Factor 智能体:可靠 LLM 应用的模式 — Dex Horthy,HumanLayer

AI Engineer · · 时长 17:06

HumanLayer 的 Dex Horthy 提供的来源,内容涉及智能体可靠性阈值、开发者对生成代码的所有权,以及 LLM 应用开发中的人机-智能体协作模式。 阅读 3 条观点,查看支持证据与原始来源。

理解这篇

3 个要点

综合解读

  1. 70–80% 的质量不足以用于生产

    百分之七十到八十的功能足以让管理层兴奋并扩充团队,但不足以满足生产使用的可靠性阈值——超过这一点后,开发者将面对深层且不透明的调用栈,以及提示词或工具链的复杂性。

    支持这项说法 1

    这足以让 CEO 兴奋,并为你的团队再增加六个人。但随后你会有点意识到,百分之七十、八十还不够好;如果你想跨过那百分之七十、八十的质量门槛,你就得深入七层调用栈,去逆向工程这个提示词是如何构建出来的。

    Dex Horthy · 0:41

    原始摘录
    It's enough to get the CEO excited and get six more people added to your team. But then you kinda realize that seventy, eighty percent isn't quite good enough, and that if you wanna get past that seventy, eighty percent quality bar, you're seven layers deep in a call stack trying to reverse engineer how does this prompt get built
    回到原文语境 →
  2. 脚手架式自有代码优于封装层

    Dex Horthy 反对在内部抽象之上加封装层,而是主张采用脚手架式、由开发者拥有的代码——他将理想的工具链比作 shadcn,即生成的代码完全由开发者拥有并可修改。

    支持这项说法 1

    你不需要在内部组件外面套一层封装。你需要的是更像 shadcn 的东西,也就是把它用脚手架搭出来,然后由我来拥有它,我拥有这些代码,并且我对此没问题。

    Dex Horthy · 15:00

    原始摘录
    You don't need a wrapper around an internal thing. You need something more like shadcn, which is like scaffold it out, and then I'll own it, and I'll own the code and I'm okay with that.
    回到原文语境 →
  3. 智能体与人类配合更佳

    Dex Horthy 表示智能体与人类配合时表现更好,并鼓励构建者寻找让智能体与人类协作的方式。

    支持这项说法 1

    嗯,我的看法是,智能体与人类配合效果更佳,应设法让二者协同协作。

    Dex Horthy · 15:38

    原始摘录
    Um, and my take, agents are better with people. Find ways to let agents collaborate with humans.
    上下文

    嗯,构建智能体存在诸多难点,但现阶段你仍应着手解决;而且,大多数难点都应予以攻克。

    原始上下文

    Um, there are hard things in building agents, but you should probably do them anyways, at least for now, and you should do most of them.

    回到原文语境 →

关键时刻3

简短、标注来源的段落,并附有可验证上下文。完整对话保留在其发布者处。

智能体可靠性阈值

70–80% 的质量不足以用于生产

这足以让 CEO 兴奋,并为你的团队再增加六个人。但随后你会有点意识到,百分之七十、八十还不够好;如果你想跨过那百分之七十、八十的质量门槛,你就得深入七层调用栈,去逆向工程这个提示词是如何构建出来的。

原始摘录
It's enough to get the CEO excited and get six more people added to your team. But then you kinda realize that seventy, eighty percent isn't quite good enough, and that if you wanna get past that seventy, eighty percent quality bar, you're seven layers deep in a call stack trying to reverse engineer how does this prompt get built
开发者体验

脚手架式自有代码优于封装层

你不需要在内部组件外面套一层封装。你需要的是更像 shadcn 的东西,也就是把它用脚手架搭出来,然后由我来拥有它,我拥有这些代码,并且我对此没问题。

原始摘录
You don't need a wrapper around an internal thing. You need something more like shadcn, which is like scaffold it out, and then I'll own it, and I'll own the code and I'm okay with that.
人机-智能体交互

智能体与人类配合更佳

嗯,我的看法是,智能体与人类配合效果更佳,应设法让二者协同协作。

原始摘录
Um, and my take, agents are better with people. Find ways to let agents collaborate with humans.
上下文

嗯,构建智能体存在诸多难点,但现阶段你仍应着手解决;而且,大多数难点都应予以攻克。

原始上下文

Um, there are hard things in building agents, but you should probably do them anyways, at least for now, and you should do most of them.

来源与研究方法

这些观点均关联原始来源。转述已明确标注,不作为逐字原话展示。

打开转录或来源材料 (在新标签页中打开)报告问题

继续了解这些人物的观点