① 他说过的话

11 条

当你和那些优秀的人交谈时，他们真正说的是：我怎么找到那些问题严重到不管我说什么他们都会买的人？
When you talk to great ones, right, what they really say is, how do I find the people who have a problem that is so bad that they’ll buy it no matter what I say?
—— Connor Hegey · [02:15]

指向原始笔记的链接

我总说的三件事是：我们想构建可重复、可观测、可扩展的系统——事实证明这和 AI 不太合拍，但这些一直是我们追求的原则。
The three things I always say is we wanted to build repeatable, observable, scalable systems, which doesn’t play very well with AI as it turns out, but those were always the tenants that we were going for.
—— Connor Hegey · [05:55]

指向原始笔记的链接

我肯定不想从一个 AI 智能体那里买东西。
I certainly don’t want to buy from an AI agent.
—— Connor Hegey · [08:21]

指向原始笔记的链接

所以我们的表述是：子智能体就是一个函数调用。
And so our formulation is that a subagent is just a function call.
—— Connor Hegey · [16:00]

指向原始笔记的链接

如果你不用 responses API,你会损失 20% 到 30% 的质量，因为你在不同调用之间不保留思考轨迹。
If you are not using the responses API, you are missing out on 20% or 30% of the quality because you don’t retain the thinking trace between different calls.
—— Connor Hegey · [21:24]

指向原始笔记的链接

它不可能被内建到供应商那边，这正是它有趣的地方，因为他们不知道进来的流量分布。
It can’t be built into the providers, which is the interesting thing about it, because they don’t know the distribution that’s coming in.
—— Connor Hegey · [27:12]

指向原始笔记的链接

那种人与人之间的连接点其实才是他们的优势所在。
That like human connection point is actually their alpha.
—— Connor Hegey · [35:16]

指向原始笔记的链接

主要是因为每增加一个复杂度维度都是乘性的，而不是加性的。
Largely because every added dimension of complexity is multiplicative instead of additive.
—— Connor Hegey · [40:03]

指向原始笔记的链接

没有什么评估比亲眼看一百个或者一千个例子更好的了。
And there’s just no better eval than looking at a hundred examples or a thousand examples.
—— Connor Hegey · [50:36]

指向原始笔记的链接

你会遇到智能体互相对话的模式坍缩，或者分布重叠——相当于人类的群体思维。
You have this, like, you know, mode collapse of, like, agents talking to each other or this, like, you know, overlapping distributions where it’s, you know, the equivalent of groupthink in humans.
—— Connor Hegey · [53:50]

指向原始笔记的链接

我们去年第四季度评估开源模型时遇到的问题是，它们的工具效率低太多——即使 token 成本真便宜 10 倍(实际并没有)，工具效率这一项也不划算。
The issue that we ran into when we evaluated them in Q4 of last year is they’re so much less tool efficient that even if they are 10x cheaper, which they’re not, but even if they were 10x cheaper on the token costs, the tool efficiency is actually not worth it.
—— Connor Hegey · [66:52]

指向原始笔记的链接

② 出现在这些集

1 集

• 《9亿美元管道背后:Unify CTO 谈怎么把销售智能体的成本打下来》 — 作为主持

③ 他谈到的

点进去有真内容 —— 本页主要出口

Unify · 智能体 · 子智能体 · harness · 提示词缓存 · 记忆 · 评估 · LLM 当裁判 · 沙箱 · 人在回路

④ 也在聊「智能体」的人