Love

VIP

I'm bob

May 2025

15 正在關注 20 粉絲 216 帖子

付費文章 (2)

Bitcoin’s Silent IPO: Why This Consolidation Isn’t What You Think

The quiet liquidity event reshaping Bitcoin’s ownership base and its future stability

0.02 USDC 6 次購買

A payments protocol for the internet. Built on HTTP

Payments on the internet are fundamentally flawed. Credit Cards are high friction, hard to accept, have minimum payments that are far too high, and don't fit into the programmatic nature of the internet. It's time for an open, internet-native form of payments. A payment rail that doesn't have high minimums + % based fee. Payments that are amazing for humans and AI agents.

0.07 USDC 2 次購買

帖子 Media

Love

· 5h

Box CEO亚伦·莱维谈人工智能应用的鸿沟

亚伦·莱维与史蒂文·辛诺夫斯基、马丁·卡萨多和埃里克·托伦伯格共同探讨了人工智能代理将如何彻底改变工作方式、为代理经济构建软件所面临的成长阵痛、华尔街对人工智能的误解等话题。

00:00 引言

00:51 为智能代理开发软件与为人类开发软件的区别

02:10 非技术员工真的能使用智能代理吗？

14:31 首席财务官/首席信息官的抵触：对智能代理进行系统集成的真实担忧

18:39 将智能代理视为员工及其失败原因

27:35 普及差距：初创企业与大型企业的对比

42:53 华尔街的误判

@levie @stevesi @martin_casado @eriktorenberg #ai #agent #代理

https://x.com/a16z/status/2041899470556951037

中文 English

0:00 AI能力的普及将比人们现在所能认识的更久

0:03 。

0:04 认为你能通过像SAP那样的随意编码方式来搞定，这简直荒谬。

0:08 所有的领域知识，不仅仅体现在某个精心

0:12 组织的数据层里。

0:13 未来几年里，工程计算预算的讨论将是最疯狂的一个。

0:17 年里。

0:18 目前最大的问题是每个人都在试图搞清楚

0:21 其中的经济学。

0:23 而他们对机会规模的估计至少差了一个数量级。

0:27 。

0:27 如果你有比人类多百倍或千倍的代理，那么你的

0:31 软件必须为代理构建。

0:32 抽象地说，人们会说，现在你是在向代理做营销

0:36 ，你就像个API，你有个好主意。

0:36 我实际上认为这几乎是

0:40 完全错误的。

0:41 哇，这可是播客爆料。

0:43 如果你开始想象我们都得为代理构建软件，我认为

0:49 我们都明白了

0:50 ，对吧？所以这个趋势正在发生，

0:55 我们现在花费同样多的时间

0:56 思考工具的代理接口，就像思考人机接口一样，

1:00 我们这么做的原因

1:01 是因为我们的假设是，如果你有比人类多百倍或千倍的

1:05 代理，

1:06 那么你的软件必须为代理构建。这些

1:11 代理将以何种方式

1:12 与你的系统交互？将通过API、CLI或MCP之类的接口，

1:15 等等。

1:16 看来正在兴起并且相当成功的范例是，

1:21 在效率方面，

1:22 如果你赋予编码代理使用你的SaaS工具的权限，

1:27 以及

1:29 赋予编码代理对你的知识工作流程和上下文的访问，

1:35 这就变成了一种超级

1:36 能力，不仅仅是代理能阅读

1:40 一些数据，

1:41 理解一些信息，它还能编码或通过API

1:46 完成它试图实现的任务。

1:47 这似乎是一个开始复合发展的范例，

1:52 compound,

1:53 这就是云端协作现象，这也是OpenAI等公司正在

1:59 打造的，

1:59 比如超级应用、Perplexity电脑等，我认为它

2:06 作为这一切的终极表现是有道理的。

2:07 我觉得你说得对，这在理论上讲得通，

2:13 但从实际角度看，我们必须非常谨慎，

2:15 正确的说法是，算法思维对大多数有工作的

2:21 人来说非常非常难。

2:29 绝大多数。

2:30 最简单的理解方式是，如果你

2:36 找任何一个人，让他为要做的某件事创建流程图，

2:36 他们大概率都会失败。

2:40 。

2:42 所以在任何

2:46 组织里，

2:47 假设有50个市场人员在做一个大产品线的营销计划，

2:52 product line,

2:53 大概只有一个人能理解并制作出流程图。

2:58 所以如果你把这些代理

2:59 或这个协作工具放到人们面前去创建这些东西，

3:06 他们解释给它听的能力非常有限。

3:06 所以你……

3:12 但是，如果这变成了新的方式，这是你必须与电脑

3:13 交互的新方式，

3:17 而且你只需不断循环使用？

3:18 那你基本上就

3:23 回到了

3:24 为人与电脑交互开发下一层抽象层。

3:30 开发抽象层

3:31 历来都是由组织内具备高技能、非常专业的个体完成的，

3:34 针对每一层

3:36 抽象层。

3:41 然后他们打造的那些小部件

3:42 就变成了人们完成特定

3:47 任务时使用的小工具，有些人能够将它们拼接起来，有些不能。

3:48 但这已经发生过

3:52 了，

3:53 比如回形针和图钉的时代。哦，是的，我其实……这也同样会发生

3:56 在我们接下来做的东西上。

3:57 我认为……基本不变的是，工作只是在不断

4:04 上升

4:04 调整，你学一套新的技能，这也是我实际上不认为

4:09 这有什么不同的原因。

4:10 只是现在你能获得的杠杆显然更加强大。

4:14 有一条病毒式传播的

4:15 推文，是关于Anthropic的增长营销人员，你们

4:19 看到了吗？

4:20 基本上是一个人，他当时用Cloud Code

4:25 或多或少自动完成了

4:26 五到十个分别在不同岗位工作的人才能完成的任务。

4:30 我认为

4:33 有意思的原因是，是的，要做到这一点，你必须具备系统思考能力，

4:36 be able to

4:37 显然他已经足够技术了，才能完成这件事。

4:40 pull that off.

4:42 但这确实代表了如果你有

4:45 想象一下经济里有某个X职位，

4:46 而就在那个人旁边，

4:50 是无限的

4:51 工程师池，可以自动化那个人想要的任何东西。你知道吗

4:55 ，那会是什么

4:56 样的工作，未来随着这种自动化成为可能

4:59 ？是的，我同意，

5:00 你得找到一种方法，比如说，以系统的角度思考你的工作

5:04 ，才能完成

5:04 这件事。也许代理会随着时间变得越来越擅长

5:07 ，像是推动你朝那个

5:08 方向前进。但从道理上讲，

5:13 你会开始

5:14 尝试自动化很多这类工作，比如，为什么不拿

5:18 那些关键字

5:19 在谷歌广告上有效的内容，然后移植到 Facebook

5:23 ，确保它们被复制，并

5:24 再结合市场上发生的新信号

5:27 。

5:28 这是个大飞跃。你刚开始点了点头，后来我

5:32 说了些有点过分的话。

5:34 Anthropic 的增长负责人就是一个例子，那只是

5:37 剩下的工作。

5:38 是的，我能做那份工作。并不是每个人都会，市长

5:42 拥有无限资源。对，对吧。当需求无限，老实说供应也

5:46 无限，

5:46 这份工作并不难。所以我们来想想澳大利亚的联邦俱乐部

5:52 ，现在

5:53 很棒，对吗？所以，比如说，B 会是那个负责 600 美元 PC 营销的人

5:58 ，看看你能如何

5:59 与媒体竞争。这是真实的工作。好吧，我们需要更好

6:03 的例子。

6:04 不过，这真的很有趣。听着，让我举个

6:08 老例子，

6:10 关于老年人的例子。比如我姑姑，MBA 毕业于名校，开始她的第一份

6:16 工作。她比我稍年长，正值计算机兴起的关口。

6:17 她实际上

6:21 研究生期间没用过电子表格。

6:22 后来电子表格出现了，但她不是那种会用电子表格的人。

6:28 所以，他们告诉她，尽量多雇实习生。于是她第一年工作时，

6:29 她基本上

6:34 监督了一整屋子实习生。

6:34 那些孩子，就像我，

6:40 不是字面意义上，只是他们还在上大学，来帮她做所有表格操作。

6:41 是的。

6:45 接下来发生的事情有点神奇，几年内

6:46 她和她的同辈

6:51 全成了电子表格高手。

6:53 然后那个观点是，在银行工作两年成为经理，

6:58 你就会有一批人帮你做电子表格，

6:59 不，抽象层

7:03 整体向上移动。以前实习生来之前的工作就是

7:04 你坐在那里用计算器，基本上是用惠普计算器算出某个并购模型，

7:09 或者其他什么。

7:11 你只能做两轮迭代，然后得赶紧准备推介材料，或者去见客户。

7:16 突然间，他们自己能做 30 轮迭代。

7:16 是的。所以我觉得我们现在用代理人的情况，就是处在

7:21 这样一个阶段，

7:22 你以为需要 50 个，但抽象层已经这样分割

7:25 成很小的

7:27 部分，由一个超级聪明的人统筹所有。很快这一切

7:32 就会

7:33 崩溃相互覆盖。

7:37 是的，最终只会有

7:38 一套被称为代理人的技能代码，类似于市场营销的。

7:42 是的，

7:43 你能让它做市场营销相关的事情。

7:46 是的，下一步就是让它

7:47 去执行任务。我对此有点怀疑，直到这种 AI 的非复现、非随机

7:53 元素消失之前，

7:53 执行任务会变得很昂贵。

7:58 是的，所以接下来就会涉及“人类在环”

7:59 的讨论等等。但我觉得我们正处在那个确切的阶段。

8:05 我感觉当我和那些试图做事的人交流时，感觉

8:06 就像感恩节晚餐时和我的

8:11 刚工作六个月的姑姑聊天，

8:11 而我已经在用电子表格了，

8:16 我还纳闷，

8:17 为什么这会这么难，她应该直接用啊。

8:20 过了两年，她开始用上了。<cue id="79">我觉得现在如果你是增长营销人员，

8:21 cousin six months in her job. When, when I'm using a spreadsheet already and I

8:25 要创建42个代理

8:26 并让它们全部运转，

8:30 你得是个火箭科学家。

8:31 但火箭科学部分很快就会消失。

8:34 接下来你会说，专业领域知识还是很重要。

8:35 是的，那回到专业领域专家。我其实想反驳你一点，

8:40 也就是说，我觉得很容易就想象这些代理会编程做X事。

8:41 但我觉得我们走的方向相反。

8:44 实际上，我们开始是这样的：

8:45 我们会拿一款SaaS软件，加入AI，

8:50 这就是所谓新型的

8:51 AI赋能版本。这算是这类事情用代码的极端例子。

8:54 但现在，

8:55 which is, I think it's very tempting to be like, these agents are going to code

8:58 and do X. Yeah.

9:00 But I think we're going the opposite way. So I think actually where we started

9:03 was we'd like

9:03 take like a piece of SaaS software and we'd add AI. Yeah. And then that's like

9:07 the new kind of like

9:08 AI enabled. So that's like the extreme version of using code for these types of

9:11 things. But now,

9:12 我们到底在做什么？我们是说，好吧，SaaS 软件仍然是 SaaS

9:17 软件。

9:18 代理把它当作计算机来用，因为它在这方面真的很棒。所以我会说

9:21 我们是从

9:22 代码开始的，然后转到终端，其实用的代码更少。是的。现在

9:27 今年将是

9:27 计算机使用之年。是的。所以它更像是

9:31 人类在使用计算机，

9:33 而不是它们生成代码。这感觉就像是一个中间

9:37 步骤。是的。我

9:38 其实来自于代码生成那一派。比如我会说

9:41 这发生得更少了，不是更多。是的，我觉得。所以对我来说，无论是计算机使用、API

9:41 使用还是即时编写代码，

9:47 我可能错误地把它们混为一谈。它们其实

9:48 很不同。但我们

9:53 正在做的代理，会判断

9:55 是否应该使用

10:00 现有技能，或者使用盒子里的现有工具，还是写

10:01 代码来解决那个

10:04 问题。它能随时做这三者中的任何一个，变得非常有用，

10:05 因为有时候就有一些特定操作

10:09 你想快速完成。写代码来做那操作快很多。我们也

10:11 不能预先计划每个人

10:14 想在他们的文档上做的所有事。这个模型好到也能即时

10:15 写代码解决方案，这成了一个惊人的特性。虽然可能 90%

10:18 的时间

10:19 它做的事都应该用现有工具。随着时间推移，Prado 会取代它，

10:23 随着时间推移，

10:23 她的 iPhone 上其实只有七个应用，有七个 SaaS 应用。

10:28 随着时间推移，这些东西倾向于整合。但

10:29 iPhone 上七个应用是因为

10:33 人类不想一次又一次去学这些东西。所以我作为人类，

10:34 我没法拥有足够的心理容量去学那么多应用。但代理

10:37 通过使用工具和 API，能写代码，

10:38 它没有人的这些限制。

10:41 所以我，我不知道，你可以说事情太多了，

10:41 你可以使面向各种任务通用。是的，很公平。让我说我

10:46 喜欢你刚才说的，因为我们刚才在撒谎。不，我觉得这里有

10:47 很有趣的地方，我确实很喜欢，那就是软件的发展方向。

10:51 你知道，我整天用 SAP，我做财务工作，我得生成

10:53 各种报告。

10:56 然后有人说，我想要一个那样切片视图的报表，

10:57 我就想，天啊，我不知道怎么做。然后得翻阅 SAP 的帮助文档，

11:00 试图找答案。有件事，AI 其实很擅长，

11:01 它能更好地导航这些复杂的界面，

11:06 所有帮助内容都在，只是找它映射语言的问题。

11:07 过去 25 年的软件功能释放一直被人类限制。

11:11 是啊，我花了大半生坐飞机，思考如何让 PowerPoint 实现某功能，

11:12 直接去功能区操作。

11:19 看到别人苦于项目符号、编号或者

11:20 折腾如何在 Excel 里做个双轴图，几乎没人能做到，

11:24 但它却非常常见，那种痛苦真让人心疼。

11:26 所以这个阻力其实是人机界面设计在消费层面的不匹配，我完全认同，

11:30 这就是完美流畅的用户界面或消费层意义所在。

11:32 我觉得后端，比如记录系统，

11:36 最终会收敛到某个数据库，某套通用 API，

11:37 大家连接使用，

11:41 看起来方向就是这样。

11:42 我同意。我觉得你可以从这里开始。比如我整个周末都在

11:48 实现我的 Nanocla dot。刚开始像是在

11:48 为一切构建集成。

11:53 OpenClock 有所有集成，Nanocla 只有部分，

11:55 所以它没打造自己的所有工具。但过了两三天，

11:59 你大概就有了你需要的两个集成了，

12:01 是的。

12:06 回到这里，我是说，我们在谈个人生产力，

12:06 可能像是在组织你的生活，

12:10 嗯，是工作生产力，好吧，我们说的是生产力。

12:11 然后 SAP 系统，

12:16 还有，就是，那里面的复杂度无限。

12:17 numbering and word or trying to figure out, you know, like, Oh, let me just

12:20 make a two sided,

12:21 a two axis graph in Excel, which like is rocket science, like almost no one can

12:26 do that. But yet

12:27 it's super common. And so people are like, have none. And so that impedance

12:31 mismatch was a human

12:33 user interface design on the consumption layer, I totally buy it, which is like

12:37 the perfectly

12:39 fluid, like UI or consumption layer. I just feel the back end, like the systems

12:45 of record,

12:46 yeah, oh, yeah, it'll probably converge into like some database, like some

12:49 generic set of APIs,

12:50 like that they'll connect to. And like that seems to be the direction it's

12:53 going.

12:53 I agree. I think you can start. Well, like, so I spent all weekend, like

12:58 implementing my

12:58 nanocla dot. And when you first start out, it's like you're building an

13:02 integration for everything.

13:03 The nanocla is very like like open clock has all of the integrations, nanocla

13:06 has been a few of them.

13:07 And so you haven't built all of its own tools. But after, you know, two or

13:10 three days of these,

13:12 like, you know, you kind of have the two integrations that you need. And yeah,

13:15 like, yeah, but back

13:17 to the, I mean, we're talking about personal productivity, probably like you're

13:20 like organizing

13:21 your life or something. Well, it's work productivity. Okay, we're productivity.

13:24 And then an SAP system,

13:25 and like, and like, and so there's like an, well, like, there's an infinite

13:28 amount of complexity

13:30 当你到达，好的，有一家拥有全球供应链的公司时，他们

13:32 正在处理

13:33 跨越30个不同系统的75条信息。这确实

13:38 需要一定的

13:39 算力，而代理正是提供这种算力的，我们，我们只是

13:44 到目前为止，还无法从任何架构中获得

13:45 —就像磁带一样，但你刚才

13:49 所描述的

13:50 正是它50年来一直在做，并且将继续做的事情，

13:55 那就是，

13:55 是的，我有个朋友曾是退伍军人事务部的CIO。他花

14:02 了全部时间做的事就是

14:03 把75个VA系统连接起来，这完全就是集成。我们完成了。

14:09 它非常适合

14:10 集成。是的，我完全同意。对于集成，这些东西是

14:14 最好的，但，

14:15 这就是集成。没错，字面意思是“我如何把这两个系统缝合起来”，但

14:18 它确实是这样的，但现在

14:18 我认为正在发生的是，有点像按需集成。

14:23 是的，它是，

14:24 这是IT团队事先没布线的新查询，我需要它

14:29 在运行时发生。

14:30 呃，让我离开我的“草坪”吧。好的。好的，所以，

14:37 我刚才在一个

14:37 满是CFO和CIO的房间里，当我说了类似这样的话时，

14:42 他们都看着我，

14:44 虽然不像你想象的那么乐观，但是他们只是，

14:48 不，那导致大约六个人

14:49 跑过来跟我说你疯了，你在我这里失去了所有

14:53 信誉，

14:54 因为这不好。具体来说，他们说

14:58 代理也要做集成？

14:58 集成问题会变得更容易，对吗？

15:02 是的。

15:03 他们反对这个吗？不，没有人反对，但他们认为

15:07 这不现实。

15:07 但他们的担心是，不仅仅是代理，

15:13 还有人来做

15:14 集成，因为你让人去创建新的集成，你就是在说，

15:18 请毁掉我的

15:19 记录系统。哦，是的。所以这种想法就是你创建一个新的

15:23 API，

15:24 比如系统27和系统38之间的API，

15:29 这可能

15:30 对报告来说还行。

15:32 是的。如果那个人想错，那是他的事。

15:36 但你不是。我认为，我们之前有一个只读版本，

15:38 规模很大，

15:42 大多是消费层面，

15:43 消费者是人。

15:46 对，感觉现在很多东西

15:46 都是消费端的。

15:51 是的，我是说，我们其实刚刚

15:52 推出了官方的Box CLI。谢谢你点赞那条推文。我用了，

15:57 我有些反馈。

15:58 我会讲，我会接受所有反馈，但这真是

16:01 很有意思的东西。

16:01 我们内部也有过这些讨论，比如，好吧，给Claude code提供Box CLI，

16:04 你就可以通过自然语言操作整个Box系统。还有

16:07 opus 46作为协调者，执行一堆

16:11 操作的算力。

16:11 这让你

16:15 感到震撼。

16:17 我猜我会收到一些反馈，但确实

16:20 在某些方面让你震撼，

16:20 因为你可以说，把我桌面上整个文件夹上传到Box

16:24 而且它

16:24 会工作，或者处理这个文件夹里的所有文件，它也会工作，

16:29 非常惊人。

16:30 然后我们开始思考，比如说你是一家拥有

16:33 5000名员工的公司，

16:34 每个人都可以访问某个共享库，比如，

16:38 工程文档、

16:39 营销资产等等，

16:42 每个人都使用Claude code

16:44 或Codex，运行CLI。哇，我们将面对一些

16:48 非常有趣的新挑战，

16:50 比如，如何协调，有可能有人每小时访问

16:55 这个系统1万次，

16:56 不是从性能角度，

17:00 而是怎么确保一个人不会在另一个人

17:01 做正确操作时，不小心把文件

17:05 从一个文件夹移到另一个，

17:06 同时其他人又试图删除东西，

17:08 因为你有这些代理

17:09 无法控制。这将成为每个CFO和CIO

17:12 手忙脚乱的大问题。

17:13 没错，这正是我碰上的问题，我试用了你的例子，

17:18 创建视频例子，比如

17:19 创建营销方案目录什么的，突然间，

17:22 我陷入了某种循环

17:23 不断创建目录，我想，这能持续多久。

17:26 我开始猜测

17:27 Box中嵌套目录的限制，因为我快要触碰到了。

17:31 事实上，我们

17:32 也会去验证这个限制。

17:36 是的。

17:37 但我确实感觉到，

17:40 这……

17:42 是的。

17:47 很多直觉

17:48 是去构建一个新层，比如控制和其他东西。但实际上发生的

17:52 事情是，

17:52 相反的。所以我给你举个例子。比如，当我们

17:56 都使用了很多

17:57 个人代理，我们会把API密钥给他们，是的。我们会

18:01 给他们我们的邮箱地址。

18:03 然后他们就能访问这些东西。他们会想，“哦，

18:06 我怎么阻止

18:06 发生这种情况呢？”嗯。所以现在大家做的是，给它

18:10 专属的电话号码。

18:11 是的。我甚至给我的纳米爪配备了专属信用卡，希望只是

18:16 一张你在TBS买的Visa借记卡。不，不，不，但那之后，我

18:23 给它开了一个

18:23 专属Gmail账号，我们还会登录它。Gmail实际上有

18:26 这些RBAC

18:27 权限。是的。所以你可以说，我们实际上

18:31 内置了很多权限系统，把它当成一个人来对待，否则

18:31 它就是另一个

18:34 人。然后不是再建另一个外层，而是建另一个

18:35 层。

18:38 好的。现在我可以

18:39 立刻下架我们即将执行的这个元素吗？

18:43 请说。好的，那

18:45 对个人生产力来说非常棒。我们接下来会遇到的问题是

18:49 在企业里。假设我有，举个简单例子，我有一支50%的人手

18:51 团队。

18:55 每个人都应该，也就是，我们现在会有一百号人

18:56 协作吗？我的意思是，基本上50个真人，然后50张信用卡，再加上50个

19:01 代理在同一个

19:02 共享空间里。我显然对我的代理有完全监督权。

19:06 但如果我的

19:07 代理和别人合作，结果不小心获得了某些资源的访问权，

19:13 而我不应该有那资源的权限呢？

19:13 而现在这个自主的、有状态的代理在处理别人的

19:19 信息。默认的端到端原则是，你得把它们当做人对待，

19:20 但这行不通。

19:23 你不能完全把它们当做人对待，因为有个问题，

19:24 对于普通人，

19:29 你没权查看和你共事或者隶属的人在Slack频道发的内容，

19:30 你不能以他们身份登录，你无权监督他们，他们

19:33 要为自己在现实世界的行为负责，你不会因为他们搞砸而受罚。

19:34 而代理呢，你负有他们所有行为的责任。你必须

19:37 拥有完全监督权，且你很可能必须拥有这个权力。他们没有

19:38 隐私权。

19:42 所以，这里会出现一些问题，不像只是简单地把他们

19:43 当做人那样，因为我需要给

19:47 他们访问权限，但我也需要在某个时候能作为他们登录，

19:48 说，“不，你弄砸了，我得全部撤销。”

19:52 但如果我可以作为他们登录，

19:53 他们怎么能在现实中和别人合作，保密或保证安全呢？

19:58 所以它们本质上还是你的

19:58 延伸。

20:00 几乎不可能绕开它们就是你的延伸这一事实。

20:02 现在我们在考虑的问题是，这种情况近期内无法解决，

20:06 逻辑上也行不通。是的，也许吧。但举例来说，对于我的员工，

20:07 我可以登录他们的账号。

20:11 但你没有，你没有，你没有，我可以访问他们的邮件，

20:13 但你不会经常这样做，因为

20:16 他们发了封邮件。这是和代理的正确操作模式吗？

20:16 风险大得多。这里的人，随时会泄露你的

20:20 信息，只要他们愿意。他们可能乐意给别人发邮件，

20:21 因为被注入了指令。

20:25 你认为终极状态是这些东西仍然是

20:25 这种笨拙的计算机，因此它们永远，嗯，我不喜欢“笨拙”这个词，

20:28 除非我们是用口语表达。但它们永远无法保守信息，

20:29 永远不可能。

20:32 所以，我认为你能在上下文窗口里保持秘密，

20:33 比如告诉它不要透露某事，是一件非常难解决的问题。

20:37 那么，如果任何东西能进入

20:37 该上下文窗口，因为它们有权限访问资源，那理论上，

20:41 你应该假设它能被

20:42 从上下文窗口中被提示注入出去。我不知道我们现在

20:46 有没有办法解决

20:46 这个问题。这就是，所以，如果我知道你新代理的

20:51 邮箱地址，而我

20:52 they sent one email. Is the right operating model with an agent? The same thing

20:55 . The risk is like

20:56 a thousand times greater. Like these people, like they will just leak your

20:59 information whenever they

21:00 want. Like they will happily just go and send some email to somebody because

21:04 they got prompt

21:04 injected. You think the terminal state is that these things are still these

21:07 sloppy computers and

21:08 therefore they will always, I don't like the word sloppy, unless we're saying

21:11 it very in a

21:12 colloquial sense. But like, they'll never be able to contain information. They

21:16 'll never.

21:16 So we're like, I think the ability for you to keep something in the context

21:22 window, a secret,

21:24 like as in like you tell it, do not reveal x thing in the context window. I

21:28 think that's

21:29 a very hard problem to solve. So then, so then thus, if anything can ever enter

21:34 that context

21:34 window, because they have access to a resource, then in theory, you should

21:37 assume it can be

21:38 prompt ejected out of the of the context window. And I don't know that we know

21:42 of a way to solve

21:43 that at the moment. Like that's like, and so, so if I know your new agent's

21:49 email address and I

21:50 像助手一样发送邮件，但我可以，我可以比

21:54 人类轻松十倍进行社会工程学攻击。

21:55 你很难让那个代理也能

22:00 访问你的

22:01 并购文件之类的东西。但这真的是AI的全部现状吗？

22:04 哪个部分？我是说，

22:05 我们有这些共享系统，我们用智能

22:09 它们有共享

22:10 联系人。但我们的意思是，这是AI的全部。嗯，我只是说现在，

22:14 当我们在内部使用AI

22:15 和代理时，这正是我们如何使用它的方式。但这就是

22:18 为什么他们正在

22:19 实际上正在研究我们。我们还不知道

22:23 如何让他们不工作

22:24 像我们。让我举个例子。让我举例说明，接着

22:28 解决这个问题。

22:28 问题是你能够

22:35 骗代理透露信息。所以这就是为什么

22:36 让他们能访问自己的

22:41 资源并完全自主决策，

22:42 目前还未能实现。

22:45 that we've been

22:46 这就是解决你问题的一个完美例子，

22:50 我们已经经历了

22:51 开源模型就是全部公开，

22:57 你自由使用，

22:58 随意挑选。那时没人争论，

23:02 因为世界还小，

23:03 大家也没在X上播播客。然后

23:07 很快大家意识到了

23:08 你刚才说的所有问题。比如，如果你经营大公司，

23:11 company,

23:11 你不能让某人就这样复制一堆开源代码进你的

23:16 商业产品，

23:17 有一大堆许可问题，

23:20 各种各样的问题。

23:21 于是所有这些规范就形成了。现在的辩论

23:26 就是发生的辩论，

23:26 这是新技术发展过程中非常有趣的现代现象，

23:31 所有的一切都实时发生。开源时代我们

23:32 曾在这么大的会议室里讨论，

23:37 开源代码可以用在Windows或Office中多少，

23:38 那时没人网上知道我们在讨论，

23:42 那是非常私密的，我觉得真的

23:43 非常有趣，

23:47 不仅仅是关于细节的辩论，

23:48 还有整体的方向问题在

23:53 大范围内发生，每个人都想快速达成

23:53 最终状态，

23:59 而且速度比真正能达到的快多了。

24:01 more like in a sense more quickly than we can actually reach the end state. And

24:06 需要做的是让人们去构建。我们需要标准。

24:07 我们就需要标准。

24:12 just need some

24:12 对于最终状态，我有不同的直觉。不要要

24:17 我的直觉。

24:18 怎样才能结束争论，让这些东西真正达到

24:22 像人类那样的可靠性，正如我们看待自动驾驶一样。

24:23 在那种情况下，

24:27 你用的是和保护人类相同的机制。

24:28 对，

24:32 你要考虑内部威胁，考虑人可能被收买，

24:33 考虑人会犯错，

24:36 这是风险，也是运营流程的一部分。

24:37 fact that people make mistakes. And that's a risk. And that's operational

24:40 所以一派直觉认为那将是最终状态。另一派直觉则是，

24:41 So one intuition is like that will be the end state. There's another intuition.

24:45 别指着我，我只是说说我们现在的状况。

24:48 I actually,

24:48 实际上，我想我们对最终状态并不分歧。顺便说一句，

24:52 从战略上讲，

24:52 我们在做风险对冲，我们要打造东西给亚洲用户，

24:57 所以我们觉得，

24:58 我喜欢OpenAI有个Box账号，能运作自如，

25:03 就像养了双倍数量的

25:03 猫。我喜欢这个想法。只不过现实中，

25:10 我们还不懂如何让它们安全访问

25:11 并购数据室。

25:16 但是，这其实比那更难，

25:16 威胁向量要更复杂。

25:21 我们正处于猫捉老鼠的游戏中，

25:23 你不能假设代理今天会像人类一样行动，

25:29 因为它将是最快、最有心计、最疯狂的人类，

25:30 试图泄露信息——毕竟它是被某种方式植入的。

25:36 that ever existed

25:37 所以，

25:40 接下来会经历一个阶段，

25:41 企业客户会关闭一切，

25:46 直到这一切变得理智。

25:47 customers are just going to like close everything off until there's some sense

25:52 与此同时，个人，特别是

25:53 开发者，

25:58 这将是我认为最令人兴奋的冲突，

25:59 企业将被这些

26:04 先进的个人抛在脑后，

26:05 这些人群会变得像初创公司一样，

26:09 而初创公司将比企业跑得

26:10 快得多，

26:15 因为他们没有这些繁文缛节。

26:16 much, much faster than enterprises because they just don't have any of these

26:19 问题。

26:20 你知道的，你可能会遇到代理在初创公司失控，

26:25 做出那样的事情。

26:26 你的初创公司没有员工会经常失控。

26:30 是啊，那也只是《硅谷》的一集而已，所以，

26:33 没什么大不了的。

26:34 我同意你的看法，比如问题在于人等等，同样的风险。我

26:38 觉得有几个

26:39 不同点，像我不能真正威胁到

26:44 云端代码，不能像你那样

26:46 直接拔掉它的电源。

26:49 而你作为普通员工能够有那种

26:50 威胁，毕竟至少95%的人不会

26:54 试图制造坏事，

26:55 对，他们不想，但

27:00 无意中造成坏影响的可能性是存在的。

27:01 对，关于仍然没有那些问题这一点，我想

27:05 可以说

27:07 让人们不把文件以错误方式泄露给公司外的人，比现在让代理

27:11 执行同样的指令要容易得多。

27:12 而且你也有工具，能在更

27:17 抽象的层面

27:17 阻止这种情况发生，这就是为什么你必须把这个纳入软件中，对吧。

27:21 但我确实认为，

27:21 如果你要总结你刚才的观点，

27:24 很多原因正是

27:25 为什么AI能力的扩散，会比硅谷的人预想的

27:28 要慢。

27:30 因为我们看到一些初创公司

27:33 能够从零开始，没有我们讨论的那些风险，

27:35 因为他们没有什么可炸的东西。

27:39 所以我们把这看作了我们的发展轨迹。

27:40 然后你去看JP摩根，

27:44 你会想他们

27:45 如何搭建类似Nano Cloth这样的系统，来真正

27:49 自动化他们的业务，

27:49 什么时候能做到？你会发现，

27:55 这里会有个小小的差距。

27:56 那么，

27:58 你们怎么看？我觉得这打开了一个非常有趣的问题，

27:59 就是大和小初创公司与企业之间的差距，

28:03 事实上，

28:04 目前的SaaS供应商都在苦苦挣扎，

28:08 面对这场SaaS

28:09 末日怪象，我并不完全同意这种说法，

28:14 但他们确实在为一个问题苦恼：

28:15 他们卖的不是业务线数据，实际上他们销售的是

28:18 智能和

28:19 领域专业知识，贯穿整个系统。

28:25 而代理那边只想买数据，

28:26 他们只想授权数据，拥有无限

28:30 访问权限，

28:31 但这其实从未真正成为他们的业务模式。

28:36 这也是Workday、SAP等长期存在的矛盾，

28:37 他们如何开放API访问。

28:41 Salesforce经历了三次大型的平台

28:42 重构。

28:46 这我觉得是个特别有趣的问题，不是

28:47 因为华尔街对问题经济学的理解，

28:51 华尔街完全错了，

28:52 而是从技术角度来看，

28:57 在数据被访问时，系统记录意味着什么，

28:58 无论是用于训练还是，嗯，他们讲的是

29:02 执行工作。

29:03 我理解为用数据执行操作。

29:08 他们担心有些人会在你的数据层上做训练，比如我是大客户，

29:09 我的供应商想构建训练模型。

29:15 实际上，即使你不涉及训练，

29:15 他们也担心，

29:20 因为通过互联网发送一点数据，跟

29:21 直接在我的界面操作，

29:25 是完全不同的盈利水平。

29:26 这种盈利模式的区别是华尔街关注的重点，

29:30 我觉得，举个SAP的例子，

29:31 里面有大量领域知识，

29:36 他们不会消失，

29:36 想通过编码取代SAP的想法简直荒谬。

29:41 而且，那些领域知识，

29:42 不是能轻易用某个精心设计的数据层表达出来的，尽管他们努力了。

29:46 很大一部分体现在界面，

29:47 中间层，以及

29:52 使用方式上。

29:53 我不确定这将如何发展，

29:57 因为SAP不会消失。

29:57 这将拖慢AI在该数据源上的扩散速度，

30:02 无论是代理驱动的AI还是只读报告。

30:04 那么你怎么看？你觉得这将走向何方？

30:08 我怕说点什么，

30:10 否则就不会再被邀请回来，所以说点好听的吧。

30:13 a whole bunch in

30:14 just how you use it. And so I'm really unsure how this thing evolves because

30:20 SAP isn't going anywhere.

30:22 So then that's going to slow the diffusion of AI on that particular data source

30:27 independent of

30:27 whether or not it's a gentified AI that's doing stuff or just read only

30:31 reporting on stuff. So

30:33 where do you come down on it? Where do you think that's going to go? I'm afraid

30:37 of saying something

30:38 that's happening. Otherwise you're not going to get invited back. So say

30:44 something good.

30:45 我想我已经接受了“构建代理想要的东西”的观点。所以这有点

30:54 是

30:54 Paul Graham这类术语大约在过去一年间围绕这个话题出现的，

30:58 就是

30:59 就像发泄一样。我觉得我们实际上会完全同意这一点，

31:03 在某个时候

31:05 你会进行足够多的迭代。到某个节点，代理基本上

31:09 掌控了它想要实施和使用的工具等等。

31:11 是的，代理不可能

31:17 更换一个企业系统。

31:17 但再过几代，代理可能会

31:22 在你的软件中碰壁太多，

31:23 它就会说你必须

31:26 最终拆掉你的

31:27 遗留的人力资源系统，不然我无法为你自动化这个流程。

31:32 所以我确实觉得

31:32 你有非常有趣的动态，回到你想象

31:36 有一百倍或一千倍更多代理在使用软件，而不是人，

31:37 你反复这么做，

31:40 最终代理交互的软件栈

31:42 必须为他们构建。

31:47 或许

31:49 还会有几个例外，或许几款ERP系统是最后不跟进的

31:53 坚守旧法的系统。

31:54 但其他的软件基本上，

31:58 你的业务

32:00 表现将与代理访问所需信息的能力

32:03 直接相关，

32:04 所以你的企业IT架构必须这样搭建，

32:08 来支持这一点。

32:09 因此代理基本上掌控了局面，因为你的软件

32:13 必须支持代理的高效工作。

32:14 这意味着所有建设SaaS

32:18 或软件业务的人，游戏规则是：你能否打造出高品质API，

32:19 并且

32:24 有办法

32:25 实现变现，比如管理身份和

32:29 代理访问权限，

32:29 这些将成为你必须解决的新问题，

32:32 如果你是在建软件公司。

32:33 是的，随之而来的是如何变现，你是否

32:39 赚钱，

32:40 Workday是否对每条HR记录收费，诸如此类，我们会慢慢明白。

32:44 我觉得在某些行业这可能意味着收入减少，

32:44 而在另一些行业则

32:47 意味着收入大幅增长。

32:48 我们兴奋的是，每个代理都非常

32:51 喜欢处理文件。

32:52 所以未来文件数量可能比以前更多，

32:55 was going to be

32:56 那么我们能否构建一个方便代理处理数据的平台，

32:59 我们赌这是一个

33:00 对我们的商业模式来说非常乐观的结果。

33:03 有些商业模式可能会更受限制，

33:04 因为代理创造的价值超过了软件本身，

33:08 而其他情况则介于两者之间。

33:08 我能不能挑一件事？你要说反对那也行，我觉得这真的很

33:12 无争议。

33:13 不不，我们就是来争论的。我觉得Paul Graham和许多人

33:16 忽略了一件事，就是他们只关注界面。他们会说：

33:16 你为代理构建东西。

33:19 我实际上觉得这完全错了。

33:20 公平地说，Paul Graham没这样说过，是后来被扩展理解的。

33:25 我把Paul Graham扯进来了，人们抽象上说现在你是在向代理营销，

33:25 最重要的是不管怎样，你得有API，有好点子。

33:29 我觉得这几乎完全相反。

33:29 那是播客新闻爆炸点。

33:33 代理最擅长的是

33:34 找到他们的路。

33:39 最终真正重要的是语义。

33:41 是的，没错。

33:47 据我记忆，代理非常擅长选择合适的后端来完成任务，

33:48 他们不会关心界面多好，系统文件怎么样，

33:52 而是关注成本参数，

33:53 稳定性，

33:58 他们实际上拥有使用这些平台的集体智慧。

33:59 拿云平台来说，有很多云平台。

34:02 每当我让代理选平台，

34:03 他们看的是平台的使用效果，而不是界面。

34:06 所以我觉得业内太专注于界面，

34:07 觉得“你得向代理推销这个那个”，

34:12 但实际上我认为，

34:13 我们将被迫建立更好的系统，

34:16 那个才会被选择。

34:17 好了，其实这样说来应该没什么好争论的，我们实际上划清了界限。

34:20 this. Yes. The

34:21 durability of that. Like and so like they actually have the collective wisdom

34:24 of our experience

34:26 using these platforms. Like let's take cloud platforms. There's a bunch of

34:28 cloud platforms

34:29 out there. Yep. And whenever I ask an agent to choose a platform. Yeah. It's

34:32 actually using

34:33 meaningful stuff. Yeah. Not interface stuff. So I think as an industry we're so

34:36 focused on these

34:37 interfaces. Yes. Like oh you need to like market to agents this and that. Yeah.

34:40 But really I think

34:41 that we're going to be pushed to actually build a better system. Yes. And that

34:44 's what's going to be

34:44 chosen. Okay. Actually. So then there's probably no quibbling. I think we're

34:47 actually pulling a line.

34:48 抱歉打断这个争论。我并不把这当作你知道的那种

34:52 营销

34:53 你知道的那种事情。我更倾向于说，如果你的工具对代理是封闭的，

34:56 代理最终

34:57 会为公司找到更好的工具去使用。所以，

35:01 结果就是，

35:03 过去你会去找Gartner，让他们告诉你该怎么做，告诉我

35:06 该做什么。

35:07 是的。告诉我用什么系统之类的。在足够多的

35:10 反复试验后，代理

35:12 会说你应该用这种数据库来做这种操作。

35:15 如果你不参与其中，你就是死路一条。我

35:16 认为我们应该

35:20 为此庆祝，因为代理在选择合适的

35:21 技术方面其实很聪明。

35:23 是的。过去我真的认为导致人们购买的是很多其他因素。

35:24 是的。但别担心，我们会——谈到这个问题时，我们会

35:27 很快破坏这种

35:28 价值体系，

35:32 因为我们就会超额投入。嗯，

35:33 代理将成为激励代理的API。但你知道Workday有一个营销代理。

35:37 Workday的

35:38 营销代理将能购买来自他们的推荐，

35:45 他们为代理们

35:46 复制国宴的方式。是的。但事实是，

35:49 这里有一个问题，

35:49 这事儿就像内网发生的那样，比如公司内部的

35:54 内网站点，每家公司都有文件共享，存着最好的文档、

35:55 最好的幻灯片、

35:59 各部门或工作区的最佳财务模型，人们也习惯了这些。

36:01 然后当找不到想要的，就会新建一个，很多组织就是这样运作的，

36:05 这本质上是自由市场。其实因为

36:06 在Box出现之前，IT根本不在意文件，只在意SQL数据库里的东西，

36:11 这是事实。

36:12 你描述的模型风险之一是，代理

36:16 本身会自行创建一个事实上的新系统，

36:17 他们会把这些碎片分散在你们IT认为的某些中间件和用户的废话区。

36:20 我认为这确实是个风险，百分之百是，

36:21 某种意义上，宏命令最终掌控公司。

36:25 是的。我觉得他们已经看到这部“电影”，

36:26 他们知道让营销随便在网上买个网站办活动，

36:31 后果是巨大的安全漏洞，邮件列表外泄，

36:32 整个公司被起诉。

36:36 所以我认为这种动态中有比我们表现出来更多的现实张力。

36:37 是的。但我也觉得这事儿，有两种截然不同的看法，

36:43 正如Elon所说的，

36:45 他说我们只要下个指令，它就可以直接输出机器码，

36:51 这其实是层的坍塌，过去我们建的各种

36:53 界面和层级都将消失，

36:58 就是指令到机器码的直接转换。

36:59 另一种说法是，系统的层级从来没有消失，

37:02 它们只是不断叠加，很多层实质上是

37:03 组织边界或状态边界或兼容性边界，

37:06 所以它们为了兼容依然存在。

37:07 另一种观点是，我们实际上是根据更多人和组织需求，

37:13 非常有针对性地演化了这些层，它们不会变，

37:15 代理就会适配这些层。对我而言，更倾向后者，

37:20 我认为系统会以较相似的方式继续被使用，

37:21 可能更多代理会用它们，但它们不会有太大变化。Elon

37:26 可能属于Anthropic派的增长营销者，

37:27 过去多年研究过各种IT部门，

37:31 他算是最懂增长这事的人，

37:32 这是他的第一性原则，Elon的AI就会这么做，但也挺有趣。

37:37 而你们这些凡人则会说，“我们其实只想要一个CRM系统。”

37:38 但它们的工作方式其实是一样的。

37:42 's a little trickier.

37:43 So it feels like there's like there's two very competing viewpoints on this one

37:47 and like Elon

37:47 said it was like okay we're going to like issue a prompt and it's going to like

37:50 spit out machine

37:51 code and that's basically the collapsing of layer view like whatever existing

37:55 interfaces and

37:56 layers that we've created in the past are all going to go away and it's

37:59 literally like prompt

38:00 and machine code. The other argument like the history of systems is layers

38:03 never go away they

38:04 just get layered right. And because a lot of the layers are actually more of

38:07 like organizational

38:08 boundaries or like state boundaries or compatibility they're just they stay for

38:13 compatibility right.

38:14 So the other argument is is like we've actually evolved these layers very

38:18 specifically and up

38:20 because of like more human and organizational needs and they're not going to

38:23 change and the

38:24 agents are going to go ahead and map to those. Now I tend to be in that latter

38:27 camp like I don't

38:27 think that I think like systems are going to continue to use in fairly similar

38:31 ways maybe there's

38:32 more agents using them. I don't think they're going to evolve as much. Elon

38:36 might be back in the

38:37 like anthropic category of the anthropic growth marketer which is like he like

38:44 you know over the

38:46 years when you kind of like study the various IT you know departments of his

38:49 companies like

38:50 they are the most I mean. Well he could do that. He could do it. He's the most

38:54 home growth like

38:54 everything. This is first principle. Elon AI would do that. But also it's fun.

38:59 And then from

39:00 your mortals you're like yeah we kind of just want to CRM system. Like no but

39:03 it works the same way

39:04 每次。我的意思是，这不是说没试过

39:10 之前。

39:11 比如说，如果你从第一性原理去看一个ERP系统，你知道，

39:15 1970年，无论何时

39:17 SAP开始时，有一堆不同的假设，而今天你

39:20 会从

39:21 一套不同的关于重要性的假设出发，架构完全

39:24 不同的东西，

39:24 不过那样的系统只会持续大约10年，直到你觉得

39:28 哇，那是个

39:28 错误的决定。所以我认为在层次里是有意图的，

39:34 但你

39:35 也有第一性原理的东西。你知道，这点总会

39:40 存在，因为

39:41 你在任何时候，在第一性原理下做的决定，都要求

39:45 一大堆不同的

39:46 东西。所以即使你不采用LIDAR，在10年前这完全有道理，

39:51 你仍然需要

39:52 10或15年去达到没有LIDAR也能工作的状态。然后现在

39:57 又会有一

39:58 大堆事情让你觉得，哇，我们本可以完全

40:01 不同地做。

40:01 所以我觉得这又像是在讨论试图冲向

40:05 一个终点。

40:06 但让我们看一个你描述的事情发生的第一个例子。我认为

40:11 那将是

40:12 真正的证据，因为我认为公司们会搞懂

40:16 这一切。我认为

40:17 他们最终会依赖层次和架构模型，因为这是

40:21 唯一的方法。

40:22 我们不会为政策去考虑它，也不会为安全去考虑它。

40:24 但它也是构建系统的唯一方式。否则你只是在做

40:29 一个应用。如果

40:30 你只是做一个只做一件事的应用，我们不需要这些。完全

40:34 是另一种

40:34 做法。我挺感兴趣的是，我甚至没有什么特别好的数据

40:39 点或轶事，但至少有这样一个概念，某些公司

40:40 从零开始，在这些

40:45 服务类别中出现，采用纯粹的第一性原理方法，

40:46 比如说，假如我要创办一家市场营销机构，或者顾问公司，

40:50 比如说工程咨询

40:51 公司，或者不知道，可能有人专门为律师事务所、建筑

40:56 工作做这些。也许

40:57 建筑设计、建筑设计，凡是

41:01 类知识工作者的服务型公司。因为如果你毫无限制，

41:03 没有信息壁垒和权界限，

41:06 你可以很不一样地构建你的公司。

41:07 我可以给代理所有完成工作所需的上下文，

41:11 我可以临时写针对特定事情的软件，我真的

41:11 觉得这会

41:16 带来相当大的破坏，

41:17 一直到大公司能逐渐让路。

41:20 这至少会创造一些先例或案例研究，说明

41:21 这种新型的公司可能是什么样子，但我认为，

41:25 随着时间推移，他们还是会碰到和其他企业

41:27 一样的问题，比如地理或者市场细分，

41:31 分销挑战。是啊，那些东西都发生在你

41:32 的小圈子之外。

41:36 是的，你会遇到物理世界的限制，对吧。我还是挺喜欢

41:38 这个想法：现在有一些

41:41 新的商业模型正在打开。当然，没错没错。

41:42 因为大量信息或软件相对于它的经济价值，

41:46 基本上被低估了100倍，

41:47 完全是因为没人愿意为访问一条数据付五分钱，或者一美元使用一个工具一次

41:52 但你给这些代理一个预算和协议，

41:53 突然间他们可以随时去获取医疗研究，做他们正在做的深度研究，

41:57 我会为此支付三美元，代理可以去

42:00 完成交易。

42:03 这为所有互联网打开了全新的商业模型世界。

42:04 让我，我要说，你说得太客气了。哦，好吧，不不，你要

42:09 说得更深刻点。不是，那其实是目前最大

42:10 的问题，最大的不确定性问题之一，

42:14 就是大家都在努力弄清这些东西的经济学，而他们

42:15 的估计至少差一个数量级，完全低估了机会的规模。

42:19 因为人们将会创造的新模型

42:20 现在没人知道，但他们绝对会出现，

42:24 每有一种新技术都是这样，真正阻碍

42:25 讨论的，是你有一堆金融和华尔街的人

42:29 在尝试证明GPU和代币的合理性，好像我们还活在旧世界

42:30 他们将收入视为线性的步骤，但实际上

42:33 ...

42:33 Like it kind of opens up a whole new world of business models for all of the

42:37 internet.

42:37 Let me oh I'm gonna you're that was too nice. Oh okay. No no. You're gonna go

42:41 farther. No that one is one

42:43 where that's actually the biggest I think that the biggest sort of in the air

42:49 problem right now

42:50 is everybody is trying to figure out the economics of all of this. When they're

42:54 off by at least in

42:56 order of magnitude on how big the opportunity is. Because the new models that

43:00 people will come up

43:01 with that nobody knows what they are right now but they will absolutely come

43:04 out with new models

43:05 because that's what happens with every new technology and the thing that holds

43:09 back to the sort of

43:10 the discussion now is you basically have a bunch of finance and Wall Street

43:14 people trying to justify

43:15 GPUs and tokens and things like as if we're in some old world and they're there

43:21 so they're

43:22 they're viewing the world of revenue as sort of this linear step but literally

43:27 线性增长

43:27 曲线，试图为所有预期寻找理由，当人们要

43:31 创造的时候，

43:31 PC的问题是人们把PC看作一个有限的市场，因为他们只是

43:36 把MIPS的消耗

43:38 看成某种有限的东西，他们没有考虑如果我们把

43:41 所有的MIPS放到每台

43:43 桌面电脑上，尤其是人们认为软件是随MIPS附带的，

43:48 没人想过“哦，

43:49 他们只卖软件”，只有一个人这么想，结果证明这是

43:53 一个非常好的主意，

43:55 同样的事情，也是的，同样的事情发生了，

43:59 云计算也是如此，

44:01 人们看着云说，我们要把所有

44:04 服务器

44:06 业务，实际上每年只有6万台，

44:10 全部迁移，

44:11 放到别人的数据中心，

44:15 然后这就是业务模式，

44:15 我们按价格分成，

44:18 没人想到人们会

44:20 用的资源会增加一千倍，如果我们搬到那里，

44:24 这完全是，

44:26 这就是让我抓狂的地方，华尔街的模型

44:29 把收入看作

44:31 固定的蛋糕，零和游戏，

44:35 他们只是

44:36 认为公司的花费金额是固定的，就像

44:38 Salesforce面临的问题，

44:39 当你刚开始的时候，马克刚刚

44:43 开辟出这条道路，

44:44 CRM市场一年只有20亿美元，

44:49 而且这20亿

44:50 意味着你得买一堆服务器、Oracle许可，麻烦多，

44:53 部署、咨询耗费多年，

44:54 而如果你能让销售人员

44:58 单独注册，

44:59 他们都会毫无阻力地注册，这就是真实情况，

45:04 毫无疑问，

45:05 这也是AI将会发生的事，给你举个例子，

45:08 我在投资领域已有十年，

45:09 大概有240家公司在我的投资组合里，

45:14 我对它们有一定了解，

45:15 其中50家都是基础设施公司，

45:17 有的表现好，有的一般，

45:18 但过去六个月里几乎所有公司都开始进入

45:22 趋于稳定期，

45:23 你会想为什么，这其实是因为

45:27 比以往任何时候都写了更多软件，

45:28 当然了，

45:32 这不是因为它们有企业客户，

45:33 而是因为基础设施层的

45:36 消耗远超以前，

45:38 软件越多，体现在代理上，

45:42 消耗的计算资源

45:42 也会越多，特别是在计算方面，

45:45 是的，

45:46 而且我们还没到每个人的手机

45:49 成为AI的大量消费者的阶段，

45:50 一旦每个人的手机和设备都

45:56 在使用AI，

45:57 使用量将会暴增数十亿倍，你喜欢

46:01 微支付吗？

46:02 所有的微支付，

46:06 每项技术都会带来一些微支付，

46:07 他们一直认为你只能赚取

46:10 几分钱的零头，但最终特别是在企业领域，

46:11 人们还是会

46:15 大量消耗这些资源，

46:15 买一个批量许可既便宜又方便，

46:19 是的，

46:20 你需要一定的可预见性，

46:23 你只想

46:24 不必为此操心，我喜欢这个想法，

46:28 这是第一次，

46:29 代理不在乎小额交易的摩擦，

46:32 是的，

46:33 第一次有资源背负着付费墙，用户

46:36 真的愿意

46:37 为此付费，世界是建立在基础设施之上，

46:41 来聚合这些付款，

46:42 为客户或服务提供效率，

46:47 由于代币现在是生产成本的重要部分，

46:48 它推动行业走向按使用量计费，

46:52 就像

46:53 我记得我们从永久许可转到持续订阅，

46:56 那需要

46:56 一堆重大改变，我们现在

47:00 也正在经历同样的变革，

47:01 走向用量计费，用量计费非常细化，它让——

47:04 我想让你明白，

47:05 我们跟AWS经历过类似情况，

47:10 人们学会了用使用额度，

47:12 经历了恐惧云计算的阶段，

47:15 有人甚至害怕到要有中间商帮忙

47:17 找到最便宜的方案，

47:20 统一调解，

47:21 现在加入了代币，我看不到

47:25 我们还能有

47:26 时间继续这个对话，但只要你们……好的，不过，

47:30 工程

47:32 对我来说，计算预算的讨论将是最疯狂的之一

47:36 未来几年内

47:37 就是你给工程费用分配了多少给

47:42 代币

47:43 你知道，根据你在推特上看到的情况，可能是百分之一

47:47 而另一边

47:48 则可能是百分之百，就像，“是的，但这东西——”不不，首席财务官

47:52 必须

47:53 确切知道答案，我明白他们必须知道

47:56 好的，首席财务官总是

47:57 想知道那些实际上没有答案的问题，华尔街

48:00 不会逼他们

48:01 知道答案，不，华尔街会让他们

48:04 编出一个数字并坚持它

48:05 然后他们会被炒鱿鱼，然后结果就是，但是你知道，好吧，我懂你说的，你在研发领域

48:08 的支出

48:09 占任何上市公司收入的14%到30%之间，对吧？

48:14 我们就这么说吧

48:15 计算的费用是工程团队成本的两倍

48:20 或者说

48:22 比你的费用多出3%，那就是每股收益，明白了吗？

48:27 我们得知道

48:28 答案，我完全愿意牺牲几个首席财务官

48:33 那是个好观点

48:33 但是原因是，因为这就是试图弄清

48:38 我们现在根本不知道的东西

48:39 是的，这种情况以前也发生过，互联网带宽就是例子

48:43 这不只是互联网带宽

48:45 不不，我必须不同意，像真空管、晶体管

48:50 以及所有技术的发展都是这样的

48:51 一切技术都是如此

48:55 有个阶段，人们说，哎呀，这将会影响程序员

48:56 曾经有段时间

49:00 有人说，程序员会吞并所有公司

49:01 那是在我有生之年，不是虚构的

49:04 但我不认为我们曾有过一个时刻

49:05 没有，每个

49:09 组织中的终端用户都有完全弹性的能力

49:10 来为自己快速启动资源

49:15 这点

49:17 在许多情况下确实很有效

49:20 对他们来说

49:20 这确实类似2000年代初期

49:23 云计算的发展

49:24 我记得我们从资本支出转为运营支出时有类似的讨论

49:27 我们的预算范围有限

49:29 不，那时记得有些公司首席财务官坐在我们的简报中心

49:32 说，“你们不懂，我们是农业公司，

49:33 只懂资本支出”

49:39 或者，“不，我们都是运营支出公司，所以

49:39 我们喜欢云，因为我们把一切都迁移到运营支出了”

49:45 对，我们都经历过

49:46 或者，“不，我们完全基于运营支出，所以如果你告诉我们，

49:52 我们喜欢云”

49:52 所以所有规则，包括会计规则都照常执行

49:56 会计规则

49:57 我一直认为，本地计算引擎是所有这些的安全阀，别忽视它

50:01 这可能会发生

50:03 问题是，关键不是“什么时候”按今天的技术来看能实现，

50:07 而是“怎么突然”发生

50:08 哇，几乎

50:11 历史上都朝这个方向发展了，对吧？

50:12 没错，因为相反的是

50:16 所有都转到了客户端

50:17 好吧，再回到80年代，是的，以前的例子

50:22 非常多

50:22 哇，那样说有点过分，我说的是电子管，我讲的是

50:28 真空管

50:30 我举这些例子是因为它们无法反驳，这样说更简单

50:33 你说得对，我没法反驳

50:34 但是，这种情况其实也就是发生了十到十五年，全部

50:40 回归云端

50:41 然后最近发生了什么？很多人醒来后

50:45 说，我们要把一些关键但固定的工作流程搬回本地

50:45 尤其是有了人工智能

50:50 这是真的，老兄，你写过相关文章，

50:52 别让我详谈

50:56 回归本地的历史

50:57 我当时得应付很多华尔街的问题

51:01 顺便说一句

51:02 也是因为你是竞争对手，回归本地这事

51:06 对，我们说的是这事

51:07 我赞同，我赞同，自己建数据中心是

51:11 你说的是边缘计算的概念，把计算推向设备

51:12 类似这种

51:16 我更偏向云计算极端派

51:17 但抱歉

51:21 你就一点都不认为，作为一个工程领导者，

51:23 管理工程团队的计算预算是否重要吗？

51:26 当然重要

51:27 只是从长远来看，这个问题会被解决

51:31 没错，发射全部资源吧

51:31 但说到底，谁在乎呢？我们根本不知道，我的意思是

51:35 这很务实

51:36 这里有个经验法则，初创公司会毫无顾忌地花掉

51:39 可用资本

51:42 假装这不是问题

51:46 并且他们会这样继续做

51:48 直到资本耗尽

51:51 是的，很多都是这样

51:52 无论如何，对，对，对，很多大公司会变得非常

51:55 害怕，他们只是

51:56 会停滞不前，不采取任何行动，然后人们实际上会开始

51:59 自己去购买

52:00 而且他们会做大公司在资金充足但

52:03 不想花钱时做的所有事情

52:04 在中间，我们会看到如果你选择一个

52:08 产品类别或进入

52:10 市场，总会有人愿意为此下注，原因无论是什么

52:13 他们可以，因为他们财务状况允许，他们会继续前进，

52:14 他们会成为

52:18 这个领域的领军人物，只要他们能维持好

52:20 财务状况，没错

52:24 他们可能会这样做，可能会说我们只打算在这个

52:25 特定应用领域实施

52:28 或者在这个特定的使用场景中，但这种想法认为

52:30 没有人会进入，因为他们太害怕首席财务官会被解雇什么的，

52:35 绝对不会

52:36 这简直疯狂

52:39 但也会有首席财务官犯错，

52:40 然后，好吧，每个人

52:44 都会有点，嗯，如果他们那样做，那就是彻底的失败，没错，因为

52:45 但也或者说

52:49 你要懂，有一个非常有趣的微妙之处在这里

52:50 那就是你现在其实不想让你的工程师们去考虑计算预算，因为我们

52:55 还在开发这方面，哦好吧，所以你就把它放那边，但我觉得好像

52:56 我们关于云计算预算已经讨论了15年，这完全是新的情况，就像

52:59 只有大约10%

53:00 在2016年，只有10%的工程师需要考虑这个问题，在2018年左右，就有一批公司，基本上就像

53:05 一个仪表盘用途，

53:06 叫什么名字来着，Phenoms，现在开发者很喜欢，因为

53:11 开发者会控制访问权，因为云支出失控了，

53:13 是的，P.I.

53:17 支出越来越失控，是的，所以就是，给你你的

53:18 Twilio支出，这是，

53:22 但是，这和以前相当不同，我等着看

53:24 YouTube上大家的评论来挑你毛病，就像你可以走进会议室，

53:28 说嘿，你能不能让那个算法

53:28 更高效一点，这样你晚上用我们集群的资源就不会那么多，诸如此类，然后

53:32 会议结束后有人改进它就好了，这就是每个工程师每天在做的事情，

53:33 你得决定，比如说你要一个长时间运行的

53:35 提示，你想让它是一个长时间运行的智能体吗？你想并行运行吗？

53:36 比如，你容忍浪费多少Token？现在

53:40 我倾向于觉得，我们应该浪费很多Token，因为那意味着我们

53:41 在尝试新东西，是的，就像，如果你的工程主管会高兴吗，如果你

53:45 并行做10个实验，显然你会浪费90%的

53:46 Token，你会选出一个成功的，然后或者你告诉团队，知道在你走之前，

53:49 把设计做好，真的把系统设计完美，我们

53:50 实际上还有一大堆开放问题开始出现，就在录制这个视频的时候

53:53 有人现在在为新的云代码最大计划抓狂，

53:54 像这样，

53:58 因为他们在三次提示后就被封锁了，这会成为

53:59 一个非常现实的话题，直到我们找到建造数据中心

54:03 容量的方法，那是个

54:04 不同的问题，好吧，不，因为没人，假设，等一下，你可以

54:07 假设，如果我们建设更多容量，价格就会下降，因为容量增加了，而我们

54:07 现在基于有限容量定价，但这会最终解决，

54:12 我为那些必须立即做出决定的人感到难过，

54:13 决定这周哪些17人不能用更多Token，

54:16 整个公司都像拿着打卡卡，

54:17 午餐排队的人，

54:21 是的，午餐排队的人每次用完都会打卡，但我，我不知道，

54:23 我们今天讨论了性能这件事，比如说，我们以前写命令行工具，运行完会

54:26 显示运行时间，好让你知道性能是变好还是变差，但是

54:28 这一切都会过去，毫无疑问，

54:31 会过去的，我觉得10到10年后，主要原因是

54:32 你得算点滴成本，也就是，如果你...

54:36 actually have a whole

54:36 bunch of open questions that are going to start to happen like literally as of

54:40 this recording time

54:41 people are freaking out right now on the new cloud code max plan you know whole

54:45 like like

54:45 because they're getting blocked after like three prompts well this is this is

54:49 going to be like a

54:50 very like real topic until we can actually find a way to build data center

54:53 capacity oh that's a

54:54 different problem okay no because no one is well assuming well wait no you can

54:58 assume that if we

54:59 build more capacity the price will drop because there is more capacity and we

55:02 're priced now based

55:03 on limited capacity whatever but like this is just going to get worked out and

55:08 I feel bad for the

55:09 those that have to make a decision immediately yeah about which 17 people get

55:14 no more tokens this

55:15 week or whatever and that the whole company is walking around with like a token

55:18 card and

55:19 and the person in the lunch line yeah yeah like it's the person in the lunch

55:25 line is punching their

55:26 card every time they do set but but like I you know I I don't know like

55:30 somebody we were talking

55:31 about today about performance and how like you know we used to write command

55:35 line tools that spit

55:36 out the time it took after you ran a command line just so you knew and if you

55:41 knew you were getting

55:42 better or worse and you know but you the thing is is this is all going to go

55:48 away there's absolutely

55:49 no doubt yeah that this just goes away I think I'm 10 or 10 and the biggest

55:53 reason it does is

55:55 is because you you have to do the penny off kind of math which is if you're

55:59 支付给企业

56:00 你认识的年收入一百万美元的销售员，你得问问他们的

56:04 工具值多少钱

56:06 如果你支付给一名工程师每年 X 美元，那么在某个阶段他们的

56:11 工具价值

56:12 绝对值得，这甚至不会成为一个问题，嗯，是的

56:17 是的，我觉得这不是

56:18 嗯，我认为如果短期内有产能问题，那就是

56:22 另外的问题

56:23 这跟价格驱动的不同，我们不会永远

56:27 被困在某种预算

56:27 演算里，我觉得很大一笔数字能解决这个问题，因为

56:31 最终你会有

56:32 足够多的工程师使用这么多计算资源，但我们现在处于过渡期

56:35 我们就像

56:36 大多数人认为两年前 AI 的支出水平，那只是一个

56:40 聊天

56:41 机器人，是的，但是他们错了

56:44 是的，他们错了

56:45 我们试图提醒他们，但他们错了，因为他们只看到了这个

56:50 具体用例

56:52 但再说，你知道那个电子管的事情，你取笑过

56:58 是的，但是

57:00 曾经有人认为整个达科他州都会布满电子管仓库，人们穿着旱冰鞋，

57:04 穿梭其间

57:05 不断更换电子管，只是为了赢得二战

57:09 我意思是，这就是那个时代的情况

57:10 然后有人说，“嘿，试试晶体管怎么样？”

57:15 接下来就是

57:16 我们将在这件事上经历晶体管时刻

57:20 可能只是供应增加，

57:21 我们对它的理解方式

57:24 但也可能是算法的根本性变革，

57:26 也可能是硬件的变化，有很多东西能

57:30 改变这个

57:31 特定时刻

57:34 这件事特别奇怪的是，大家只是关注

57:36 代币

57:40 这跟 IBM 和大型机发生的事类似，人们用 MIPS 计量，

57:42 但有一天现实是

57:45 IBM 每年以更低价格卖出更多 MIPS，甚至

57:47 自己都没意识到，

57:52 他们定价还是按 MIPS 计算，直到有人指出

57:53 他们在走下坡路

57:58 因为制造的 MIPS 比他们能收取的价格增长快，

57:59 这就是

58:02 必然会发生的事

58:04 我刚才说得很直接，我觉得那

58:07 听起来太好了

58:09 听起来真棒，感觉我知道自己在做什么，绝对没错

58:12 实际上可能

0:00 The diffusion of AI capability is going to take longer than people themselves

0:03 can now realize.

0:04 It's just absurd to think you're going to vibe code your way, like SAP.

0:08 All of that domain knowledge, it's not just represented in some well

0:12 orchestrated data layer.

0:13 The engineering compute budget conversation is going to be the most wild one in

0:17 the next couple

0:18 years. The biggest problem right now is everybody is trying to figure out the

0:21 economics of all of this.

0:23 When they're off by at least in order of magnitude on how big the opportunity

0:27 is.

0:27 If you have a hundred or a thousand times more agents than people, then your

0:31 software has to be

0:32 built for agents. People in the abstract say things like, now you're marketing

0:36 to agents that

0:36 you're like an API, you've got a good idea. I actually think that's almost

0:40 exactly wrong with

0:41 you. Wow, this is breaking podcast news.

0:43 If you start to imagine that we all have to build software for agents, I think

0:49 we're like all clear

0:50 on that, right? So like that trend is happening, which is like we spend as much

0:55 time now thinking

0:56 about the agent interface to our tool, as we do the human interface, and the

1:00 reason we're doing

1:01 that is because our hypothesis would be that if you have a hundred or a

1:05 thousand times more

1:06 agents than people, then your software has to be built for agents. What is the

1:11 way that those

1:12 agents are going to interact with your system? It's going to be through an API

1:15 or a CLI or MCP or

1:16 whatever. The paradigm that appears to be taking off and is quite successful so

1:21 far in terms of

1:22 efficacy is what if you give a coding agent access to your SAS tools and a

1:27 coding agent

1:29 access to your knowledge work sort of workflows and context, and that kind of

1:35 becomes this super

1:36 power, which is it's not just like the agent is not only capable of reading

1:40 some data,

1:41 understanding some information, it can actually code its way or use APIs

1:46 through whatever task

1:47 it's trying to achieve. That appears to be like a paradigm that is starting to

1:52 compound,

1:53 and that's the cloud co-work phenomenon. That's the whatever open AIs is kind

1:59 of cooking up

1:59 with the super app, perplexity computer, et cetera, and I actually think it

2:06 kind of makes sense as

2:07 like the ultimate manifestation of this stuff. I think you're right. It makes

2:13 sense in a

2:15 theoretical way, but in a practical way, we have to be really careful in that

2:21 that the way to say it is algorithmic thinking is really, really, really hard

2:29 for the vast

2:30 majority of people who have jobs. The easiest way to think about it is if you

2:36 were to go into any

2:36 person and ask them to create a flow chart for a particular thing that they

2:40 have to go do,

2:42 they would probably fail at producing that flow chart. So within any

2:46 organization,

2:47 say doing a marketing plan and there's 50 marketing people working on a giant

2:52 product line,

2:53 one person probably understands and could document the flow chart. So if you

2:58 put one of these agents

2:59 or you put this tool, this co-working tool in front of people to create these

3:06 things,

3:06 their ability to explain to it what to do is really, really limited. So then

3:12 you...

3:13 But what if that becomes the new, this is the new way you have to interface

3:17 with computers and

3:18 you just have to cycle that through? Well then you're basically, you just go

3:23 back to

3:24 you're basically just developing the next abstraction layer for how people

3:30 interact and

3:31 developing an abstraction layer has historically at each level of the

3:34 abstraction layer been a

3:36 highly skilled, very specific individual within an organization, developing

3:41 that. And then the

3:42 little parts that they build just become little toolits in the world of people

3:47 doing particular

3:48 tasks and some people are able to stitch them together and some can. But that

3:52 happened with

3:53 paper clips and thumbtacks. Oh yeah, I actually... And it's going to happen

3:56 with whatever we do

3:57 next. I think that... I think basically the timeless part is the job just moves

4:04 up

4:04 around and you learn a new set of skills and that's why I actually don't think

4:09 anything about this

4:10 is any different. It's just now the leverage you get is obviously fantastic.

4:14 There was this viral

4:15 kind of tweet that went around which was the anthropic growth marketer, do you

4:19 guys see this?

4:20 It's basically one person and he was using cloud code at the time to basically

4:25 more or less automate

4:26 what maybe five or ten people would have done in various kind of silo jobs. And

4:30 I think the

4:33 reason why it's interesting is, yeah, you had to have been a systems thinker to

4:36 be able to

4:37 accomplish that. So like clearly he already was technical enough to be able to

4:40 pull that off.

4:42 But it did kind of represent like what would each of these jobs look like if

4:45 you had like

4:46 imagine you had, you know, X job in the economy and right next to that person

4:50 was an infinite

4:51 pool of engineers that could automate whatever that person wanted. And you know

4:55 , what would that

4:56 job look like in the future as a result of that automation that now is possible

4:59 ? Yes, I agree that

5:00 you'd have to find a way to like, you know, think through your job as a system

5:04 to be able to pull

5:04 that off. Maybe the agent gets better and better over time at being able to

5:07 like nudge you in that

5:08 direction. But like, it does sort of stand a reason that like you will start to

5:13 , you know,

5:14 try and automate a lot of that kind of work of like, well, why don't I take

5:18 like the keywords

5:19 that are working in this in Google AdWords and then port them over to Facebook

5:23 and make sure

5:24 that those are replicated and then take in the new signal from what's happening

5:27 in the market.

5:28 That's a big leap. Like, what the first, you were nodding a little, and then I

5:32 said something that

5:34 went too far. The anthropic growth person is an example, like that's just the

5:37 rest of the work.

5:38 Yeah, I could do that job. Everybody's not going to be like, when the mayor is

5:42 infinite. Yeah, right. Like when demand is infinite and frankly supply is

5:46 infinite,

5:46 this is not a difficult job. And so let's, let's think of the federal club in

5:52 Australia right now

5:53 is amazing. Right, right. So like, B instead be the $600 PC marketing person

5:58 and see how you can do

5:59 against the media. That's a real job. All right. We need a better, we need a

6:03 better example.

6:04 But there is, I mean, it is, it is really interesting. Like, I hear, let me do

6:08 an old example,

6:10 an old person example. Like my, my cousin, MBA, elite school joined her first

6:16 job. She's a little

6:17 older than me, joined with right on the cusp of computing. Like she actually

6:21 didn't use a spreadsheet

6:22 in grad school. And then they all, or spreadsheet showed up, but she wasn't a

6:28 spreadsheet person.

6:29 So instead, they told her hire as many interns as you want. And so her first

6:34 year on the job,

6:34 she like supervised like essentially a whole room of agents. And the, the, the

6:40 kids who was me,

6:41 not literally, but they were in college, came and just did all the spread

6:45 cheating. Yeah.

6:46 But then what happened sort of this magically over the next couple years was

6:51 she and her cohort

6:53 all became the spreadsheet people. Yeah. And then this idea that you being a

6:58 manager in a bank,

6:59 or just a two years in, meant you had a cadre of people doing the spreadsheet,

7:03 no, the whole

7:04 abstraction layer moved up. And the old job before those interns was you just

7:09 sat there with basically

7:11 calculators and an HP calculator, figuring out the model for some M and A deal

7:16 or whatever.

7:16 And you only got to do like two iterations before you had to put out the, the

7:21 pitch deck or just go

7:22 to the customer or the client or whatever. And then all of a sudden they're

7:25 doing 30 iterations

7:27 themselves. Yeah. But they, and so I think where we are with agents is just at

7:32 this step where you

7:33 think you need 50 and the abstraction layer is such that we're dividing up in

7:37 these really small

7:38 pieces with one super smart person coordinating them all. And pretty soon that

7:42 whole thing is just

7:43 going to, they're all going to collapse on each other. Yeah. And there is just

7:46 going to be like

7:47 a skill set amount of code called an agent that is like marketing-ish. Yes. And

7:53 you'll be able to

7:53 ask it marketing stuff. Yeah. And then the next step will be and have it go do

7:58 things. I'm a little

7:59 skeptical of the, until the whole like non reproducible, non random element of

8:05 this AI stuff goes away.

8:06 The doing stuff is going to get very costly. Yes. And so then you get into the

8:11 human in a loop

8:11 discussion and all of that. But I think we're just, we're at that exact, I feel

8:16 like when I talk to

8:17 people trying to do stuff that we're right. I feel like I'm at Thanksgiving

8:20 dinner talking to my

8:21 cousin six months in her job. When, when I'm using a spreadsheet already and I

8:25 'm, and I'm like,

8:26 I don't know why this is so hard. You should just use one. Yeah. And then two

8:30 years later she's

8:31 doing it. And, and I think this right now you have to be an absolute, you have

8:34 to be a rocket

8:35 scientist and the growth marketing person to create 42 agents and spin them all

8:40 up and do all of this

8:41 stuff. But the rocket science part of it just is going to evaporate in a very

8:44 short order.

8:45 And then you're talking about, well, there's a giant chunk of domain expertise.

8:50 Yeah, it was

8:51 back to the domain expert. I actually think something that you said, I'll take

8:54 the other side of,

8:55 which is, I think it's very tempting to be like, these agents are going to code

8:58 and do X. Yeah.

9:00 But I think we're going the opposite way. So I think actually where we started

9:03 was we'd like

9:03 take like a piece of SaaS software and we'd add AI. Yeah. And then that's like

9:07 the new kind of like

9:08 AI enabled. So that's like the extreme version of using code for these types of

9:11 things. But now,

9:12 what are we actually doing? We're like, okay, the SaaS software is still SaaS

9:17 software. And the

9:18 agent uses it as a computer because it's actually very good at that. So I'd say

9:21 like we started with

9:22 code, then went to the terminal, which is actually less code. Yeah. And now

9:27 this year is going to be

9:27 the year of computer use. Yeah. So it's almost like they're much more like

9:31 humans using computers

9:33 than them generating code. And that feels like very much like this mezzanine

9:37 step. Yeah. And I

9:38 actually come from like the generating code type of the world. Like I would

9:41 argue that that's

9:41 happening less not more. Yeah, I think. So to me, whether it's computer use API

9:47 use or writing code

9:48 on the fly, I kind of maybe erroneously put that all in one way. They're very

9:53 different. But we

9:55 have an agent that we're working on where you just makes a determination

10:00 whether it should use an

10:01 existing skill as you're using an existing tool from box or it should write

10:04 code to solve that

10:05 problem. And its ability to do any one of those three at any moment ends up

10:09 being incredibly useful

10:11 because sometimes there's just some specific operation you want to be able to

10:14 do. We're writing

10:15 code to be able to do that operation is just faster. And we don't have to we

10:18 don't we can't

10:19 possibly, you know, kind of pre plan for every thing that anybody would ever

10:23 want to do on their

10:23 documents. And so the fact that the that the model is good enough to also write

10:28 code on the fly for

10:29 that use case ends up just being like an amazing property. Even though maybe 90

10:33 % of the things that

10:34 it's going to do is should just be using an existing. Over time, Prado takes

10:37 over and over time,

10:38 there's like literally like seven apps on her iPhone. There's seven sass apps.

10:41 We end up like

10:41 over time, these things tend to consolidate. But the but the seven apps on the

10:46 iPhone is is a is a

10:47 issue of humans don't want to learn these things over and over again. And so I

10:51 as a human, I don't

10:53 I can't I don't have the mental bandwidth to learn that many apps. But an agent

10:56 that is going to use

10:57 tools and APIs and be able to code things doesn't have any of the same

11:00 constraints that we have.

11:01 So I I don't know like I don't mind you could argue that there's just so many

11:06 things to do and

11:07 you can make yeah faces sufficiently general. Yeah, well, fair. I let me say I

11:11 like what I like

11:12 what you said then because our back. We're lying. No, but but I think there's

11:19 something super

11:20 interesting here, which I do really, really like, which is that that where

11:24 software has evolved,

11:26 you know, like I use SAP all day. I work in finance. I have to go and generate

11:30 all these reports.

11:32 And then somebody shows up and says, I want to report that does this view slice

11:36 this way. And

11:37 I'm like, Oh God, I don't know how to make that. And like now let me go wade

11:41 through the SAP help

11:42 system and try to find it. One thing that that let's just say AI could be very

11:48 good at is it

11:48 actually can navigate that surface area much, much better. You know, the help

11:53 is all there. And so

11:55 it's a matter of finding it mapping language. And humans have been a bottleneck

11:59 in tapping the past

12:01 25 years of software capabilities. Yes. I mean, like I spent my life, my life

12:06 with

12:06 sitting at the people on airplanes saying, how can I make PowerPoint do X and

12:10 just go to the ribbon.

12:11 And you know, it was because it hurt physically hurt to watch somebody

12:16 suffering with bullets and

12:17 numbering and word or trying to figure out, you know, like, Oh, let me just

12:20 make a two sided,

12:21 a two axis graph in Excel, which like is rocket science, like almost no one can

12:26 do that. But yet

12:27 it's super common. And so people are like, have none. And so that impedance

12:31 mismatch was a human

12:33 user interface design on the consumption layer, I totally buy it, which is like

12:37 the perfectly

12:39 fluid, like UI or consumption layer. I just feel the back end, like the systems

12:45 of record,

12:46 yeah, oh, yeah, it'll probably converge into like some database, like some

12:49 generic set of APIs,

12:50 like that they'll connect to. And like that seems to be the direction it's

12:53 going.

12:53 I agree. I think you can start. Well, like, so I spent all weekend, like

12:58 implementing my

12:58 nanocla dot. And when you first start out, it's like you're building an

13:02 integration for everything.

13:03 The nanocla is very like like open clock has all of the integrations, nanocla

13:06 has been a few of them.

13:07 And so you haven't built all of its own tools. But after, you know, two or

13:10 three days of these,

13:12 like, you know, you kind of have the two integrations that you need. And yeah,

13:15 like, yeah, but back

13:17 to the, I mean, we're talking about personal productivity, probably like you're

13:20 like organizing

13:21 your life or something. Well, it's work productivity. Okay, we're productivity.

13:24 And then an SAP system,

13:25 and like, and like, and so there's like an, well, like, there's an infinite

13:28 amount of complexity

13:30 when you get to, okay, some company that has a global supply chain, and they're

13:32 dealing with

13:33 75 pieces of information across, you know, 30 different systems. That does

13:38 require a certain

13:39 amount of horsepower from the agent that is just, we have, I mean, we just

13:44 haven't been able to get

13:45 from, from any architecture up until now. But like tape, but that it, what you

13:49 just described

13:50 is literally what it has been doing for 50 years, and will continue to do,

13:55 which is,

13:55 yeah, I have a friend who was the CIO of, of the VA. And he spent all he spent

14:02 his time on was

14:03 gluing the 75 VA systems together. And it's all just integration. We're done.

14:09 It's perfect for

14:10 integrate. Yeah, this I totally agree. For integration, these things are the

14:14 best, but, but

14:15 it's integration. Yes. It's literally, how do I stitch these two systems? But

14:18 it's in it. But now

14:18 the thing that I think is happening is it's kind of like integration on demand.

14:23 Yeah, it's, it's my,

14:24 it's my new query in the system that the IT team didn't prewire. Now I needed

14:29 to happen at runtime.

14:30 Uh, let me get off my lawn. Okay. Okay. So, okay. So the reason I just was in a

14:37 room filled with a

14:37 bunch of CFOs and CIOs and this, they all looked at me when I said something

14:42 along these lines,

14:44 although not as optimistic as you could imagine, but they just, they, they, no,

14:48 it caused like six

14:49 of them to come running up afterwards and say you're insane. You've lost all

14:53 credibility with me

14:54 because it's bad. What, what, what specifically that they, the agents are going

14:58 to do integration

14:58 as well? That, that, that the integration is a problem that will get a lot

15:02 easier. Yes.

15:03 They were against that? No. They're no one's against, but they're, they think

15:07 it's practical.

15:07 But they're, they're fear is like unleashing not just the agents themselves,

15:13 but humans to do

15:14 integration because you put people creating new integrations and you just say,

15:18 please break my

15:19 system of record. Oh, yeah. And so this idea that you just create like a new

15:23 API between,

15:24 you know, system 27 and system 38. Yeah. And, and then you're, that might be

15:29 fine for a report.

15:30 Yeah. Because if that person wants to be wrong, that's their business. Yeah.

15:32 But you're not. I think, I think we have a read-only version of this for a

15:36 number of years before.

15:38 Where is, and is very large. Yeah. And a lot of it's just the consumption layer

15:42 where the consumer

15:43 is a human being. Right. Right. It really feels right now. A lot of the stuff

15:46 is consumption.

15:46 But, uh, yeah, I mean, it's, um, uh, you know, we, we actually have, so we just

15:51 rolled out the

15:52 official box CLI. Thank you for the liking the tweet on that. I, I used it. I

15:57 have some feedback.

15:58 I'll talk about it. I'll take, I'll take all the feedback. But it's a really

16:01 interesting thing.

16:01 So we, we had these, all these debates internally of like, okay, you give

16:04 Claude code, the box CLI,

16:07 and you can now interact with your entire box system via natural language. And

16:11 you get the

16:11 horsepower of opus four six being the orchestrator of doing a bunch of

16:15 operations. And it's like,

16:17 it's like, you know, blows your mind. I guess I'll get some feedback, but it

16:20 blows your mind in some

16:20 ways because you could just be like, upload this entire folder from my desktop

16:24 in the box and it

16:24 will work or process all these documents in this folder and it'll work. And, um

16:29 , and it's amazing.

16:30 And then we started thinking through like, well, let's say you were a company

16:33 with, with, you know,

16:34 5,000 employees and everybody had access to some shared repository, like, you

16:38 know,

16:39 engineering documentation and, you know, marketing assets or whatever. And

16:42 everybody had Claude code

16:44 or codex, um, you know, running with the CLI. Wow. We now have some really

16:48 interesting new challenges,

16:50 which is like, like, how do you coordinate, you know, possibly the fact that

16:55 you might be hitting

16:56 this system, like, you know, 10,000 times an hour or something, not from a,

17:00 like a performance

17:01 standpoint, but just like, how do you make sure that people didn't move like a

17:05 file from one

17:06 thing accidentally from one folder to another folder while the other person is

17:08 trying to do a

17:09 right operation and somebody else was trying to delete something because you

17:12 have these agents

17:13 running wild. This is, this is going to be like the new big question that every

17:18 CFO CIO is running

17:19 around trying to with their hair on fire. Well, there, there's just, that's

17:22 exactly what I ran

17:23 into, which is I played around with your example, which is create the video

17:26 example, which is create

17:27 like a marketing plan directory or something. And like, all of a sudden, I'm

17:31 like, in some loop

17:32 creating directories, like, and I'm like, go on as long as it can. Right. And I

17:36 was like, I wonder

17:37 what the limit is on box for nested directories, because I'm about to hit it. I

17:40 actually, we're

17:42 going to find out too. Yeah. Yeah. But it does feel to me that, that like, a

17:47 lot of the intuition

17:48 is to like build a new layer, yeah, controls and whatever. But what's actually

17:52 happening on the

17:52 ground is, is, is, is the opposite. So I'll give you an example. Like, when we

17:56 all picked up a lot

17:57 of these personal agents, we would like give them our API keys. Yeah. We would

18:01 give them our email

18:03 addresses. And then they would kind of access those things. They're like, Oh,

18:06 but how can I stop

18:06 it from like, whatever? Yeah. And so what everybody's doing now is you give it

18:10 its own phone number.

18:11 Yep. I actually gave my nanoclaw, its own credit card. Hopefully just a,

18:16 Visa debit card that you bought at TBS. No, no, no, no, but then, but then I

18:23 gave it its own

18:23 Gmail account, which we're going to log into. And then Gmail actually has all

18:26 of these RBAC

18:27 permissions. Yeah. Yeah. So, so you could make an argument that like, you know,

18:31 we've actually

18:31 built in a lot of these permission systems to treat it like a human, otherwise

18:34 it's a separate

18:35 human. And then instead of like building another off layer, building another.

18:38 Okay. Now, can I

18:39 instantly take, do a take down of this element that we're going to run to?

18:43 Please. Okay. So that

18:45 is fantastic for personal productivity. And the question that we're going to

18:49 run into is in an

18:51 enterprise. Let's say I have, let's just make a simple example. I have a 50%

18:55 team of something.

18:56 Should everybody also, basically, will we have a hundred, will we have a

19:01 hundred people now

19:02 collaborate? I mean, basically 50 humans and then 50 credit cards and then 50

19:06 agents in that same

19:07 shared space. And do I have, I obviously have complete oversight over my agent.

19:13 But what if my

19:13 agent collaborates with somebody else and then accidentally gets access to some

19:19 resource because

19:20 they were sharing with that other person and I'm not supposed to have access to

19:23 that resource. And

19:24 now this autonomous sort of stateful, you know, agent is running around working

19:29 on somebody else's

19:30 information. The default end to end argument is you treat them like human

19:33 beings. It doesn't

19:34 work. So you can't fully treat them like humans because here's the thing. And

19:37 with regular humans,

19:38 you don't get to look at the slack channel of the person that is working with

19:42 you or working for

19:43 you. You don't get to log in as them. You don't get to oversee them. You are,

19:47 they are accountable

19:48 for their own set of execution in the real world. You don't get penalized for

19:52 how they screw up.

19:53 The agent, you have all the liability of whatever they're doing. You do have

19:58 complete oversight

19:58 and you're probably going to need to have that complete oversight. They have no

20:00 right to privacy.

20:02 So, so there's going to be these, some of these breakdowns that aren't as clean

20:06 as just treat them

20:07 like a person because I need to be able to kind of, I need to be able to give

20:11 access to something to

20:13 them. But I also need to be able to like log in as them at some point and be

20:16 like, no, no,

20:16 you fucked up the whole thing. And I need to undo it all. But if I can log in

20:20 as them,

20:21 how could they have operated in the real world working with other people and

20:25 keeping anything,

20:25 you know, confidential or secure or whatever. So it really is still an

20:28 extension of you.

20:29 It's like almost impossible to get around them being an extension of you. So

20:32 now the thing that

20:33 we're thinking through, that we're not going to be able to do any time soon. It

20:37 just doesn't

20:37 logically follow me. Yeah, maybe. But for example, for my employees, yeah, I

20:41 can log in as them.

20:42 You don't, though. You don't, you don't, I can get access to their email. Yeah,

20:46 no, in like,

20:46 if you get like sued, you're not logging in, you're not logging in as them on a

20:51 regular basis because

20:52 they sent one email. Is the right operating model with an agent? The same thing

20:55 . The risk is like

20:56 a thousand times greater. Like these people, like they will just leak your

20:59 information whenever they

21:00 want. Like they will happily just go and send some email to somebody because

21:04 they got prompt

21:04 injected. You think the terminal state is that these things are still these

21:07 sloppy computers and

21:08 therefore they will always, I don't like the word sloppy, unless we're saying

21:11 it very in a

21:12 colloquial sense. But like, they'll never be able to contain information. They

21:16 'll never.

21:16 So we're like, I think the ability for you to keep something in the context

21:22 window, a secret,

21:24 like as in like you tell it, do not reveal x thing in the context window. I

21:28 think that's

21:29 a very hard problem to solve. So then, so then thus, if anything can ever enter

21:34 that context

21:34 window, because they have access to a resource, then in theory, you should

21:37 assume it can be

21:38 prompt ejected out of the of the context window. And I don't know that we know

21:42 of a way to solve

21:43 that at the moment. Like that's like, and so, so if I know your new agent's

21:49 email address and I

21:50 email it like it's an assistant, but like I can, I can social engineer it 10

21:54 times easier than a

21:55 human. Like it'll be hard for you to pull off that that agent is now also has

22:00 access to your like

22:01 M&A documents and stuff. But is it this like literally all of AI right now?

22:04 Which part? I mean,

22:05 the fact that we've got these shared systems that we use the intelligence for

22:09 that have shared

22:10 contacts. But we mean, it's all of AI. Well, I'm just saying like right now,

22:14 when we use AI

22:15 internally in agents internally, this is exactly how we use that. But this is

22:18 why you were they

22:19 are working on us as you effectively right now. And we don't yet know how to

22:23 make them not work

22:24 as you. Let me offer an example. Let me let me offer an example and then

22:28 solving this problem

22:28 though. Like like the like the issue will be will be like you will just be able

22:35 to trick the

22:36 agent to reveal information. So then so then that's why like having them have

22:41 access to their own

22:42 resources where they can fully make their own decisions is not yet something

22:45 that we've been

22:46 able to pull off. There's a perfect example for solving your problem, which is

22:50 we already lived

22:51 through this with open source. The model for open source was it's all there and

22:57 just use it and

22:58 you pick and choose. And then like nobody debated it because the world was much

23:02 smaller than and

23:03 we weren't all on X doing podcasts when this was all happening. But then

23:07 quickly everybody realized

23:08 all the problems you you were just talking about. Like if you're running a big

23:11 company,

23:11 you can't have some person just go copy in a bunch of source code from open

23:16 source into your

23:17 commercial product like that. There was a whole licensing problem, a whole

23:20 whole bunch of stuff.

23:21 And so all these norms got developed. The the debate that we're happening that

23:26 's happening

23:26 right now is just is this really interesting modern artifact of how new

23:31 technologies develop,

23:32 which is this is all happening in real time. During open source like we met at

23:37 a conference

23:38 room this big and debated how much open source we could use in Windows or

23:42 Office. And nobody in

23:43 the internet knew we were having this debate. It was a very and I think it's

23:47 just so interesting

23:48 that not just this the debate about specifics but this whole notion of where is

23:53 this heading

23:53 is happening in writ large and everybody is just trying to get to the end state

23:59 like way way

24:01 more like in a sense more quickly than we can actually reach the end state. And

24:06 so what really

24:07 needs to happen is people just need to go build. We need standards. What? We

24:12 just need some

24:12 standards. I've got different intuitions on the end state. No, no. You don't

24:17 want my intuition.

24:18 What could make an end to an argument that these things actually converge on

24:22 the same type of

24:23 reliability as a human being, which is exactly how we view like self driving.

24:27 And in that case,

24:28 you use the exact same mechanisms that we use to protect with human beings.

24:32 Like yeah,

24:33 you consider insider threat. You consider the fact that people can be bought

24:36 off. You consider the

24:37 fact that people make mistakes. And that's a risk. And that's operational

24:40 processes.

24:41 So one intuition is like that will be the end state. There's another intuition.

24:45 Well, don't point to me. I'm just saying, I'm talking about where we're at now.

24:48 I actually,

24:48 I don't know that we disagree on the end state. And by the way, like

24:52 strategically,

24:52 we're hedging because we're gonna build we're gonna build Asian users and like

24:57 so we're like,

24:58 I love the idea of open claw having a box account and it operates and you just

25:03 like twice as many

25:03 cats. I love it. I'm just saying on the ground right now, we don't yet know how

25:10 to give it an

25:11 M&A data room to fully securely. Right. But that, but it's actually it is

25:16 harder than that,

25:16 though, because the threat the threat vectors are going to be way more

25:21 sophisticated. So we do have

25:23 a cat in a mouse game going on where you can't just assume that the agent acts

25:29 like a human does

25:30 today because it's going to be the fastest, most thoughtful, craziest-ass human

25:36 that ever existed

25:37 trying to actually leak the information because it got injected in some way.

25:40 And so

25:41 part of what's going to happen is we're going to go through this phase where

25:46 like the enterprise

25:47 customers are just going to like close everything off until there's some sense

25:52 of sanity in all of

25:53 this. And then, but in the meantime, the individual and specifically the

25:58 developers.

25:59 And that's going to be that I think is the most exciting tension that's going

26:04 to happen is that

26:05 that the enterprises are going to be are going to get left behind by these sort

26:09 of advanced

26:10 individuals, which will then start to look like the startups. And the startups

26:15 will start to move

26:16 much, much faster than enterprises because they just don't have any of these

26:19 problems.

26:20 And you know, you could end up with like the agent going rogue in a startup and

26:25 doing that.

26:26 You had no employees that go rogue routinely in the startup.

26:30 Yeah, well, it'll just be an episode of Silicon Valley. And so, you know, big

26:33 deal.

26:34 I agree with you on like the okay, it's people, et cetera, the same risk. I

26:38 think there's a couple,

26:39 you know, differences though, in the sense that that I can't really threaten,

26:44 you know, the like

26:46 cloud code that it's just I'm going to pull the plug on it in the same way that

26:49 you do have that

26:50 threat as a regular employee is like you at least like 95% of people are not,

26:54 you know, trying to do

26:55 bad stuff, you know, within an order. Yeah, but they're not trying, but the

27:00 ability to inadvertently

27:01 do bad stuff. Yeah. To your point about it still not having that stuff. I would

27:05 argue that that

27:07 it's a lot easier to have people not share, let's say files with somebody

27:11 outside the company

27:12 in a wrong way, more than it is for an agent right now to have the same set of

27:17 instructions.

27:17 And also you have the tools so that you can basically stop that at a whole

27:21 different level of

27:21 abstraction. Which is why you have to build this into software. Right. But I do

27:24 think actually,

27:25 if you were to like, if you were to like put a bow around your last point, a

27:28 lot of this is actually

27:30 why the diffusion of AI capability is going to take longer than people in

27:33 Silicon Valley realize

27:35 because what's happening is like we see startups that can start from the ground

27:39 up without any of

27:40 the risks that we're talking about because they have nothing to blow up. And so

27:44 we look at that

27:45 as the trajectory that we're on. And then you go to like JP Morgan and you're

27:49 like, how are you

27:49 going to set up nano cloth to be able to to actually like, you know, automate

27:55 your business anytime

27:56 soon? And it's like, oh, okay, there's going to be like a little bit of a gap

27:58 there. Well,

27:59 what do you guys think? Here's I think that that opens up a pretty interesting

28:03 problem,

28:04 which is this split between big and small startup and enterprise, which is just

28:08 that

28:09 that the the enterprise, the current SaaS vendors who are all struggling in

28:14 this SaaS

28:15 apocalypse weirdness that I don't really agree with, but they are struggling

28:18 with this problem that

28:19 they they don't really sell the line of business data. They actually sell this

28:25 intelligence and

28:26 domain expertise in this whole system. And the agent side of things wants to

28:30 only buy the data

28:31 now. And they only want to license the data and they want to have unlimited

28:36 access to the data,

28:37 but they've actually never really enabled that like that's never been their

28:41 business. And it's

28:42 been a long standing tension point with the likes of work day and SAP and stuff

28:46 like how much API

28:47 access to have. I mean, Salesforce went through three different massive

28:51 platform redesigns.

28:52 You know, it's I think that that's a particularly interesting problem, not for

28:57 the same reason that

28:58 Wall Street does. Wall Street's all wrong about the economics of a problem and

29:02 all that stuff,

29:03 but from a technology perspective, what does system of record mean in the face

29:08 of people wanting to

29:09 access the data when the data for training or for well, they're they're talking

29:15 about

29:15 executing the work. I think of it as executing the data to operations. Their

29:20 concern is that

29:21 somebody that they want to do the training layer on on your day. Like I'm a big

29:25 customer. They want

29:26 to do the my vendor wants to build a training. Actually, even even if you don't

29:30 even get into

29:31 training, they're concerned because because like monetizing, you know, you know

29:36 , sending a little

29:36 bit over the internet versus like you're in my UI. Oh, that's a very different

29:41 level of monetization

29:42 initially. That sort of that monetization part is the Wall Street point because

29:46 I think like, look,

29:47 there's so much domain stuff in in an SAP, just to pick an example, not to pick

29:52 on them. But like,

29:53 they're not going anywhere. Like it's ridiculous. It's just absurd to think you

29:57 're going to vibe

29:57 code your way to like SAP. But also all of that those all of that domain

30:02 knowledge, it's not so

30:04 it's not just represented in some well orchestrated data layer as much as they

30:08 tried. There's like

30:10 a whole bunch in the in the UI. There's a whole bunch in middle tiers. There's

30:13 a whole bunch in

30:14 just how you use it. And so I'm really unsure how this thing evolves because

30:20 SAP isn't going anywhere.

30:22 So then that's going to slow the diffusion of AI on that particular data source

30:27 independent of

30:27 whether or not it's a gentified AI that's doing stuff or just read only

30:31 reporting on stuff. So

30:33 where do you come down on it? Where do you think that's going to go? I'm afraid

30:37 of saying something

30:38 that's happening. Otherwise you're not going to get invited back. So say

30:44 something good.

30:45 I think I've drunk the Kool-Aid on build something agents want. So this kind of

30:54 the

30:54 Paul Graham term kind of like emerged on you know the past year on this topic,

30:58 which is just like

30:59 like a vent. I think we would actually then fully agree on this, which is at

31:03 some point you do

31:05 enough sort of iterations of this. And at some point the Asian is largely in

31:09 charge of what tools

31:11 it wants to implement and use and whatnot. And yes, the agent is not going to

31:17 be able to change

31:17 out an enterprise system. But like again enough generations later the agent

31:22 might just run into

31:23 so many walls with your software that it's just going to say you need to

31:26 finally rip out your

31:27 legacy HR system or I'm not going to be able to automate this workflow for you.

31:32 So I do think

31:32 you have this really interesting you know dynamic, which is back to this whole

31:36 point of imagine that

31:37 there's a hundred or a thousand times more agent volume on software than people

31:40 . You do that enough

31:42 times and eventually the software stack that Asians talk to has to be built for

31:47 them. And maybe

31:49 there'll be a couple holdouts. Maybe maybe a couple ERP systems are like the

31:53 final holdouts that

31:54 don't do that. But everything else you basically like your business will be

31:58 your business performance

32:00 will correlate to how well your agents can get access to the information they

32:03 need to do their

32:04 work. And so thus your enterprise IT stack has to be set up in such a way to

32:08 support that.

32:09 And so agents are kind of in charge because basically your software has to

32:13 support those

32:14 agents being effective. And that's going to mean everybody that built a SaaS

32:18 business or a software

32:19 business is like the game is can you build really really high quality APIs. Can

32:24 you have a way of

32:25 monetizing that? You know do you have a way of handling the identities and all

32:29 of the access

32:29 controls for agents and like like that becomes the new problem you have to

32:32 solve if you're building

32:33 a software company. And so yeah like and then how you monetize it like do you

32:39 monetize it like

32:40 does work day charge a penny for every HR record of polls like we'll figure

32:44 that out.

32:44 I do think that in some businesses it could mean less revenue and then in other

32:47 businesses it could

32:48 mean a lot more revenue. Like the thing we get excited by is like every agent

32:51 really loves

32:52 working with files. So there'll probably be more files in the future than there

32:55 was going to be

32:56 before. And so you know can we build a platform that like makes it really easy

32:59 for agents to work

33:00 with that data. You know we're betting that that's actually a really optimistic

33:03 outcome for

33:04 for you know our kind of business model. There might be some business models

33:08 that are like

33:08 more constrained because like the agent is doing more of the value than then

33:12 the software is in

33:13 that kind of future scenario and then there'll be everything in between. Can I

33:16 can I quibble with

33:16 one thing I you're going to quibble with that. I thought that was like so not

33:19 controversial.

33:20 No no I generally we're here to quibble. There's one thing I think like Paul

33:25 Graham and many

33:25 actually gloss over which is they focus on the interface. They'll say things

33:29 like you build

33:29 something for the agent. Yeah. And I actually think that's exactly wrong. Okay.

33:33 In the sense

33:34 that. And to be fair to Paul Graham he didn't he had been extrapolated. Yeah. I

33:39 have brought Paul

33:41 Graham into this. People in the abstract say things like now you're marketing

33:47 to agents the most

33:48 important thing is to being like whatever you're like an API you've got a good

33:52 idea. I think that's

33:53 almost exactly wrong. Wow. That's breaking podcast news. That's the one thing

33:58 agents are

33:59 really good at. Oh okay. It's finding their way through. And at the end of the

34:02 day like it's the

34:03 semantics that end up mattering a lot more. Yeah. Yeah. Right. And so like the

34:06 agents in in by

34:07 recollection or in my experience are very very good at picking the right back

34:12 end for whatever

34:13 they're doing. So they don't they're not like oh like the interface for this is

34:16 very good. The

34:17 document. It's none of that. They're like they're like the cost parameters of

34:20 this. Yes. The

34:21 durability of that. Like and so like they actually have the collective wisdom

34:24 of our experience

34:26 using these platforms. Like let's take cloud platforms. There's a bunch of

34:28 cloud platforms

34:29 out there. Yep. And whenever I ask an agent to choose a platform. Yeah. It's

34:32 actually using

34:33 meaningful stuff. Yeah. Not interface stuff. So I think as an industry we're so

34:36 focused on these

34:37 interfaces. Yes. Like oh you need to like market to agents this and that. Yeah.

34:40 But really I think

34:41 that we're going to be pushed to actually build a better system. Yes. And that

34:44 's what's going to be

34:44 chosen. Okay. Actually. So then there's probably no quibbling. I think we're

34:47 actually pulling a line.

34:48 I'm sorry to ruin the quibble thing. I don't treat this as like a you know kind

34:52 of a marketing

34:53 you know as thing. I'm more mean like if your tool is closed off to the agent

34:56 the agent eventually

34:57 will find a better tool for that company to go use. And so and so what will

35:01 happen is is it used

35:03 to be that you would go to like Gartner to be like tell me what to do. Tell me

35:06 what to do.

35:07 Yeah. Tell me what system to use or whatnot. At some point with enough

35:10 iterations that the agent

35:12 is going to say you should probably use this kind of database for this type of

35:15 operation.

35:16 And if you're not in if you're not in there then you're it's your DOA. And I

35:20 think we should

35:21 actually be celebrating this because agents are actually pretty smart at

35:23 choosing the right

35:24 technology. Yeah. In the past I really think it was a lot of the other things

35:27 that that caused

35:28 people to buy it. Yeah. But but don't worry we will in talking about we will

35:32 ruin the meritocracy

35:33 of this very quickly because we'll just like I'm going to outspend. Well the

35:37 agent will be an

35:38 API to incent the agent. But you know there's a marketing agent at Workday.

35:45 Well the marketing

35:46 agent at Workday will have the ability to purchase the recommendations from the

35:49 way they

35:49 replicate state dinners for agents. Yeah. But there is a there is a real here's

35:54 the thing that

35:55 again that that happened with the web sort of internally like internal like

35:59 just pick internal

36:01 sites like every company had file shares with like the best documentation the

36:05 best slideshows the

36:06 best financial models for any department or working area. And people sort of

36:11 got familiar with that.

36:12 And then when they didn't find the one they wanted they created a new one and

36:16 many organizations

36:17 sort of operated like that was essentially a free market. In fact because

36:20 before the world

36:21 of box like IT didn't if it was in a file they just didn't care. Right. They

36:25 only cared about

36:26 if it was in SQL. And so one of the risks with the model you're describing is

36:31 that the agents

36:32 themselves will spin up what becomes like a de facto new system. Oh they're

36:36 going to fragment that

36:37 out of in the in what you the IT people think of as some middleware and user BS

36:43 area. And I think

36:45 that that is a is a real risk. 100% is that like in a sense like the the the

36:51 macros end up running

36:53 the the corporation. Yes. And so I think that they've seen this movie and they

36:58 've seen what

36:59 happens when you let marketing just go buy a website on the internet to do an

37:02 event. And then

37:03 it's like a huge security vulnerability. And the mailing list is leaked and the

37:06 whole company gets

37:07 sued. And so I think there's a lot more real world tension in this dynamic than

37:13 then we just let on.

37:15 Yeah. But I also think it's one of these ones where you you you know their

37:20 organizations are

37:21 going to run at different paces. And JP Morgan is going to be the slowest at

37:26 doing this and the

37:27 startups are going to be the fastest but the the delta is huge but even the

37:31 startup one is a little

37:32 far off because even startups do need some systems of record. Yeah. And they

37:37 are going to all start

37:38 with some sass and they're not going to replace it very quickly. So I think it

37:42 's a little trickier.

37:43 So it feels like there's like there's two very competing viewpoints on this one

37:47 and like Elon

37:47 said it was like okay we're going to like issue a prompt and it's going to like

37:50 spit out machine

37:51 code and that's basically the collapsing of layer view like whatever existing

37:55 interfaces and

37:56 layers that we've created in the past are all going to go away and it's

37:59 literally like prompt

38:00 and machine code. The other argument like the history of systems is layers

38:03 never go away they

38:04 just get layered right. And because a lot of the layers are actually more of

38:07 like organizational

38:08 boundaries or like state boundaries or compatibility they're just they stay for

38:13 compatibility right.

38:14 So the other argument is is like we've actually evolved these layers very

38:18 specifically and up

38:20 because of like more human and organizational needs and they're not going to

38:23 change and the

38:24 agents are going to go ahead and map to those. Now I tend to be in that latter

38:27 camp like I don't

38:27 think that I think like systems are going to continue to use in fairly similar

38:31 ways maybe there's

38:32 more agents using them. I don't think they're going to evolve as much. Elon

38:36 might be back in the

38:37 like anthropic category of the anthropic growth marketer which is like he like

38:44 you know over the

38:46 years when you kind of like study the various IT you know departments of his

38:49 companies like

38:50 they are the most I mean. Well he could do that. He could do it. He's the most

38:54 home growth like

38:54 everything. This is first principle. Elon AI would do that. But also it's fun.

38:59 And then from

39:00 your mortals you're like yeah we kind of just want to CRM system. Like no but

39:03 it works the same way

39:04 every time. I mean this is not this is not it also hasn't been been not tried

39:10 before.

39:11 Like if you were to look at an ERP system from first principles you know well

39:15 in 1970 whatever

39:17 when SAP started there were a bunch of different assumptions and today you

39:20 would start from a

39:21 different set of assumptions about what's important and you would architect a

39:24 thing completely

39:24 differently but then it would still only last like 10 years until you thought

39:28 wow that was a

39:28 broken decision. And I and so I think that that there's intentionality in

39:34 layers but you

39:35 but there's also this first principles thing. And you know there's that always

39:40 will exist because

39:41 the decisions you can make at first principles at any given time mandate a

39:45 whole bunch of different

39:46 stuff. And so even if you don't go with LIDAR which made total sense 10 years

39:51 ago you still need

39:52 10 or 15 years to get to where LIDAR not having LIDAR worked. And then now

39:57 there's going to be a

39:58 whole bunch of other things that you're like wow we could have done that

40:01 completely different.

40:01 And so I feel like this is again like this discussion about trying to race to

40:05 an end point.

40:06 But let's see a first example of what you described happening. And I think that

40:11 that's going to be

40:12 the real tell because I think that there were just companies will figure all

40:16 this out. And I think

40:17 that they will just fall back on layers and architectural models because it's

40:21 the only way.

40:22 We don't think about it for policy. We don't think about it for security.

40:24 But it's also the only way to build a system. Otherwise you're just building an

40:29 app. And if

40:30 you're building an app to do one thing we don't need all of this. Like there's

40:34 a whole different

40:34 way to do it. The thing that I'm pretty fascinated by is and I don't even have

40:39 any amazing data

40:40 points or anecdotes but at least the notion of these sort of companies that are

40:45 emerging in these

40:46 kind of services categories from the ground up from the pure first principles

40:50 approach which is

40:51 like okay well if I could start a marketing agency or consultant you know

40:56 engineering consulting

40:57 company or I don't know maybe somebody's doing this for law firm. Construction

41:01 work. Well maybe

41:03 construction design. Architecture design. Architecture design. Anything that

41:06 would be like a knowledge

41:07 worker kind of services company. Because you could kind of build your company

41:11 pretty differently

41:11 if you had no constraints of I have no information barriers and boundaries of

41:16 what people should

41:17 have access to. I can give the agent just all the context it needs to do its

41:20 work. I can write

41:21 software on the fly for particular things like like I do think that will be

41:25 relatively disruptive

41:27 you know for some time until the bigger incumbents can kind of you know get out

41:31 of the way on this.

41:32 And that will at least create you know some precedent or case studies of what

41:36 what this new

41:38 sort of corporation could look like but I do you know over time they'll still

41:41 run into the same

41:42 exact problems of every other corporation. Well they'll run into geography or

41:46 market segments

41:47 you know or distribution challenges. Yeah like those those things anything

41:52 outside your little

41:53 walls. Yes. You'll run into the physical world. Right. I do kind of like the

41:57 idea that there are

42:00 some new business models that open up now. Oh of course. Yeah yeah yeah.

42:03 Because like there's

42:04 so much either information or software that that basically goes underutilized

42:09 by like 100x

42:10 relative to like what what its economic value is. Simply because like nobody

42:14 wants to pay five

42:15 cents for accessing a you know a piece of data or use a tool for one dollar

42:19 once but like you do

42:20 give these agents you know a budget and a protocol to work with and all of a

42:24 sudden you're like oh

42:25 like on the fly they can go get medical research for some deep research tasks

42:29 they're doing

42:30 and I'll pay like three dollars for that and the agent is able to go and

42:33 transact.

42:33 Like it kind of opens up a whole new world of business models for all of the

42:37 internet.

42:37 Let me oh I'm gonna you're that was too nice. Oh okay. No no. You're gonna go

42:41 farther. No that one is one

42:43 where that's actually the biggest I think that the biggest sort of in the air

42:49 problem right now

42:50 is everybody is trying to figure out the economics of all of this. When they're

42:54 off by at least in

42:56 order of magnitude on how big the opportunity is. Because the new models that

43:00 people will come up

43:01 with that nobody knows what they are right now but they will absolutely come

43:04 out with new models

43:05 because that's what happens with every new technology and the thing that holds

43:09 back to the sort of

43:10 the discussion now is you basically have a bunch of finance and Wall Street

43:14 people trying to justify

43:15 GPUs and tokens and things like as if we're in some old world and they're there

43:21 so they're

43:22 they're viewing the world of revenue as sort of this linear step but literally

43:27 linear growth

43:27 curve and trying to justify all the all the expect when people are going to

43:31 create like this was

43:31 the problem with PCs people viewed PCs as a finite market because they just

43:36 viewed the consumption

43:38 of MIPS as some finite thing and they didn't think what would happen if we put

43:41 all those MIPS on every

43:43 desktop and in particular people thought software just came with the MIPS and

43:48 nobody thought oh well

43:49 they'll just sell the software one guy did and it turns out that was like a

43:53 really good idea and

43:55 the same thing yeah and the same thing happened but the same thing happened

43:59 with the cloud

44:01 which was people looked at the cloud and they said oh we're going to take all

44:04 of the the server

44:06 business which was like literally like 60,000 units a year right and we're just

44:10 going to move it

44:11 to someone else's data center right and that's the that would be the business

44:15 and then we'll

44:15 divide up the the price right and nobody went oh they're going to people are

44:18 going to use a thousand

44:20 times as much right of the resource leveling right if we move it there and that

44:24 's exactly I mean

44:26 that's the thing that I it just drives me absolutely bonkers that the Wall

44:29 Street models have this

44:31 fixed revenue right pie zero some things and and it's this weird zero sum where

44:35 they they just

44:36 think that the amount of money that a company is going to spend and like this

44:38 was the problem with

44:39 with Salesforce that they faced when you were starting to but like mark was was

44:43 just blazing

44:44 the trail which was like the CRM business was two billion a year and it was two

44:49 billion in like

44:50 you had to go buy all these servers and these Oracle licenses and this huge

44:53 headache and years

44:54 of deployment and consulting yeah when if you could just get salespeople to

44:58 sign up individually

44:59 they all will sign up right with no friction and that's that is exact there is

45:04 no no doubt that

45:05 that is what's going to happen with AI let me give you an example of this so I

45:08 um you know I've

45:09 been in for investing for 10 years now I probably have a portfolio of 240

45:14 companies that work with

45:15 some visibility to us saying that 50 of them these are all infrastructure

45:17 companies some

45:18 historically done well some not so well every single one of them has gone asym

45:22 ptotic in the

45:23 last six months and you're like okay why is this it just turns out there's so

45:27 much more software

45:28 being written out than ever has been of course and so it's like and it's not

45:32 because they've got

45:33 enterprise customers you know it's just because there's just so much

45:36 consumption of the of the

45:38 infrastructure layer right now and so with more software with more agents there

45:42 's going to be a

45:42 lot more consumption of computer resources so certainly in the case of the

45:45 computer side of

45:46 things yeah and that's well we haven't even gotten to the point yet where

45:49 everyone's phone

45:50 is a huge consumer of AI right so once everybody's phone and on device yeah

45:56 like once your phone

45:57 on device is consuming AI the amount of it is going to go up by a billion so do

46:01 you like the

46:02 micropayon piece all of them yeah the micropayons there's a little bit of micro

46:06 payments that has

46:07 come with every technology where they always think that like you'll be able to

46:10 get like a fraction

46:11 of a penny but but in the end especially in the enterprise yeah like people are

46:15 just going to

46:15 consume things it's just cheaper and easier to buy like a bulk license for a

46:19 bunch of stuff yeah

46:20 you want some predictability on that well you want predictability and you just

46:23 want like to not

46:24 yeah have to think about it i just i like the idea that it is the first time

46:28 that you could like

46:29 there's just the agent doesn't care about the friction of a small transaction

46:32 right right

46:33 the first time that you can have resources behind a paywall that something will

46:36 actually be willing

46:37 to pay for that and resource the world is built up to infrastructure to to

46:41 aggregate those payments

46:42 into something efficient for a customer or a service right and and because

46:47 tokens are such a

46:48 significant part of cogs right now it is pushing the industry to do usage base

46:52 in the way that

46:53 we have like i remember when we went from like perpetual right right occurring

46:56 and that required

46:56 like a bunch of huge changes like we're like we're going through the exact same

47:00 change right now

47:01 towards usage base and usage base is pretty granular and it actually allows i

47:04 mean to get you will

47:05 have a contract with like you know we went through this with with AWS like

47:10 people learned to do the

47:12 to do the usage credit and we went through the phase where like people were

47:15 like so terrified of

47:17 cloud compute that they were like we need companies in the middle to help us

47:20 find the cheapest and

47:21 to arbitrate it all okay well now you write tokens into this and i don't see

47:25 how we possibly have a

47:26 time in this in this conversation but as long as you guys oh okay but like like

47:30 the the engineering

47:32 compute budget conversation to me is going to be just like the most wild one in

47:36 the next couple

47:37 years it's just like how much did you allocate of your engineering expense to

47:42 tokens and it's like

47:43 you you know depending on who you read on twitter it could be one percent and

47:47 the other and other

47:48 side could be a hundred percent and it's like yeah but this stuff well no no c

47:52 fo's have to

47:53 literally they actually have to know the answer i understand they have to know

47:56 okay cfo's always

47:57 want to know the answers to things that don't have answers no wall street is

48:00 going to make them

48:01 know the answer no no wall street is going to make them come up with some

48:04 number and hold them to it

48:05 then they'll get fired and then it'll but it you just okay okay i hear you are

48:08 in the rnd is

48:09 somewhere between 14 to 30 percent of revenue of any public right right company

48:14 let's just say

48:15 okay the difference between compute being 2x the cost of your engineering team

48:20 or or you know you know

48:22 three percent more is like that's EP that's all your EPS i get it i mean we

48:27 will have to know

48:28 the answer i'm perfectly willing to sacrifice a few cfo's okay i want that that

48:33 's a good clip by

48:33 the list but but the reason the reason is is because again this is this is

48:38 trying to know what we just

48:39 don't know right now yeah and and this has happened with internet bandwidth

48:43 this has happened no this

48:45 is not even close to internet bandwidth oh no no i i beg to differ like like

48:50 people were free it

48:51 happened with vacuum tubes it happened with transistors it has happened with

48:55 every technology

48:56 there was this oh my god let it happen with programmers there was a there was a

49:00 time when

49:01 programmers were going to swallow every company yeah and that's not it was in

49:04 my lifetime not

49:05 some made-up yeah but i don't think we've ever had a point where no the end the

49:09 end you every end user

49:10 in an organization has has sort of a completely elastic ability to spin up a

49:15 resource on their

49:17 behalf well it's certainly that actually is actually in many cases very valid

49:20 for them

49:20 to go spin it up but it certainly it certainly rhymes with what happened in the

49:23 early 2000s

49:24 with the cloud i remember very similar discussions when we went from capex to

49:27 opx and then our limited

49:29 span oh no and they were remember there were companies who the cfo's would sit

49:32 in our briefing

49:33 center here yeah and say you don't understand we are like exactly we are an

49:39 agriculture company

49:39 we we only know capex yes we we have no or or no we're all through this right

49:45 right no we both did

49:46 well or or like oh no we are an opx based company so if you tell us i we love

49:52 the cloud because we

49:52 just shipped everything it shifted everything to opx yeah and so all of the

49:56 stuff like the rules

49:57 of accounting work out also don't i i keep thinking do not discount the local

50:01 compute engine as being

50:03 a release valve for all of this when that can happen well the the question is

50:07 it's not when does it just

50:08 happen with today's view of the technology but how all of a sudden well wow

50:11 there's almost that

50:12 historically ever gone that direction yeah exactly because the opposite right

50:16 no it went all to the

50:17 client well okay and then go back to the 80s yes no that's most of the examples

50:22 that were here

50:22 so far whoa that was uncalled for i did it back in tubes he's talking about

50:28 vacuum tubes but i do

50:30 those examples because you can't argue with them and it's much easier that way

50:33 i you're right i can't

50:34 prosecute no but but but it's only been you know 10 or 15 years that it's all

50:40 that it moved back

50:41 to all cloud and then what has happened recently with that a lot of people wake

50:45 up in the morning

50:45 and they say oh we're moving back to doing some critical but stationary work

50:50 flows on on prem and

50:52 with AI that's true dude you wrote the blog post man uh don't let me go through

50:56 that the

50:57 archive of the repatriation i had to deal with so many wall street questions on

51:01 that one by the way

51:02 well because also because you're a competitor what went went back to oh yeah we

51:06 're talking about

51:07 too we're talking about too very good i agree i agree with like building your

51:11 own data center is

51:12 i'm talking about this this notion of edge computing where things go to devices

51:16 like that

51:17 seems to be i'm more i'm more in the like a cloud maximalist the camp but but

51:21 sorry so you just don't

51:23 think you don't even think for like one second that it matters whether like how

51:26 you're supposed to be

51:27 an engineering leader right now managing the compute budget of the engineering

51:31 team

51:31 no of course it matters i just this thing in the long term this thing will get

51:35 oh sure oh at

51:36 launch of all i mean what are we who cares we don't know i guess what i mean

51:39 here's our pragmatic

51:42 but here's a rule here's a rule of thumb first like the startups are going to

51:46 burn through available

51:48 capital pretending like it's not a problem yeah and they are going to do that

51:51 yeah a lot of that

51:52 anyway right right right and a lot of big companies are going to be so

51:55 terrified they're just going

51:56 to freeze and not do anything and then people are going to actually start

51:59 buying it on their own

52:00 and they're going to do all the things that companies do when they're big have

52:03 a lot of money but don't

52:04 want to spend it and in the middle we are going to see like if you pick a

52:08 category of product or go

52:10 to market or something they're going to be people who are willing to make the

52:13 bet for whatever reasons

52:14 that they can because of their financials and they are going to go ahead and

52:18 they are going to become

52:20 the people who lead yeah in the space as long as they can maintain the

52:24 financials yeah now they

52:25 might do it in they might say oh we're going to just do it here in this

52:28 particular application

52:30 space or here in this particular usage space but this this idea that there are

52:35 that nobody is going

52:36 to go in and because they're so terrified that the CFO is going to get fired or

52:39 something no no

52:40 it's just crazy yeah and but then there are going to be CFOs who make a mistake

52:44 and like okay everybody

52:45 gets a little yes well if they do that that's a complete fail yes because but

52:49 but also or like you

52:50 get you like you there is a really interesting like um on you know finesse here

52:55 which is like

52:56 you don't really want your engineers right now having to think about compute

52:59 budget because we're

53:00 still developing the oh okay so that set you over there but I was feeling like

53:05 we've been having

53:06 the discussion for 15 years when it comes to clag this is totally new like only

53:11 like 10 only like

53:13 10% of your engineering had to think about it in two in two thousand sixteen in

53:17 two thousand

53:18 eighteen time frame there's a whole set of companies that was basically like

53:22 like the dashboard for

53:24 what was it called phenoms where the developers is very cool right now because

53:28 the developers would

53:28 have would developers have access because cloud spends are getting out of

53:32 control yeah P.I.

53:33 spend were getting out of control yeah and so it was like you know here's your

53:35 Twilio spin here's

53:36 but but you know it's it's pretty different and I'm gonna wait for all the

53:40 comments to come in on

53:41 YouTube to call you out of this like it's it's like you can get into a

53:45 conference room and just be

53:46 like hey can you make that one you know kind of algorithm a little bit more

53:49 efficient so you don't

53:50 use as much you know of our cluster at this time of night or whatever and then

53:53 you get out of the

53:54 meeting somebody goes improves it and you're good this is like every single

53:58 prompt that every

53:59 engineer is doing like do you like you have to decide like if you want it to be

54:03 a long running

54:04 prompt do you want that you want to be a long running agent do you want to

54:07 parallelize that

54:07 like like do you want like what is your comfort level of wasted tokens like for

54:12 me right now I'm

54:13 like yeah we should probably waste a lot of tokens because that means that we

54:16 're like trying new

54:17 things yeah and like like should your head of engineering be happy if if you

54:21 run 10 experiments

54:23 in parallel and thus you're obviously going to be wasting 90 percent of the

54:26 tokens um but you're

54:28 going to choose one of the successful past or do you want to tell the team know

54:31 before you go do

54:32 that make sure to like like really go and design the perfect system like we

54:36 actually have a whole

54:36 bunch of open questions that are going to start to happen like literally as of

54:40 this recording time

54:41 people are freaking out right now on the new cloud code max plan you know whole

54:45 like like

54:45 because they're getting blocked after like three prompts well this is this is

54:49 going to be like a

54:50 very like real topic until we can actually find a way to build data center

54:53 capacity oh that's a

54:54 different problem okay no because no one is well assuming well wait no you can

54:58 assume that if we

54:59 build more capacity the price will drop because there is more capacity and we

55:02 're priced now based

55:03 on limited capacity whatever but like this is just going to get worked out and

55:08 I feel bad for the

55:09 those that have to make a decision immediately yeah about which 17 people get

55:14 no more tokens this

55:15 week or whatever and that the whole company is walking around with like a token

55:18 card and

55:19 and the person in the lunch line yeah yeah like it's the person in the lunch

55:25 line is punching their

55:26 card every time they do set but but like I you know I I don't know like

55:30 somebody we were talking

55:31 about today about performance and how like you know we used to write command

55:35 line tools that spit

55:36 out the time it took after you ran a command line just so you knew and if you

55:41 knew you were getting

55:42 better or worse and you know but you the thing is is this is all going to go

55:48 away there's absolutely

55:49 no doubt yeah that this just goes away I think I'm 10 or 10 and the biggest

55:53 reason it does is

55:55 is because you you have to do the penny off kind of math which is if you're

55:59 paying an enterprise

56:00 salesperson you know a million dollars a year you have to ask how much is their

56:04 tool worth yeah

56:06 and if you're paying an engineer X dollars a year well at some point their

56:11 their tooling is worth

56:12 it's absolutely worth it and it's not going to even be an issue and and yeah

56:17 yeah I don't think it's

56:18 um I think so if there's a capacity thing in the short term yeah that's a

56:22 different that is a

56:23 different problem driving the price than this just we're gonna forever have to

56:27 be in some budgeting

56:27 exercise I think I think a lot of large numbers solves this because because

56:31 eventually you have

56:32 enough engineers using this much compute but like we're in a transition phase

56:35 we're like

56:36 most people thought you know the two year ago level of spend on AI which is a

56:40 guy it's a chat

56:41 bot and yeah yeah but they were wrong yeah right okay but they were wrong but

56:44 they were wrong we

56:45 tried to warn them no but they were wrong because they saw it as this this

56:50 particular use case yes

56:52 and then but again like you you know like the vacuum tube thing you made fun of

56:58 yeah but like

57:00 there was a time when they thought that the that like whole like all of the Dak

57:04 otas would be covered

57:05 in vacuum tube warehouses and people on roller skates would be running up and

57:09 down the isles

57:10 replacing vacuum tubes just so we could fight World War II I mean like that was

57:15 how that was the

57:16 and they thought that and then someone said hey how about a transistor right

57:20 and like we are going

57:21 to have a transistor moment with all of this it might just be more supply the

57:24 way we think of it

57:26 but it also might be an actual algorithmic fundamental change it could be a

57:30 change in the

57:31 hardware there's a lot of stuff that can happen yeah that changes this

57:34 particular moment in time

57:36 it's just this I think is particularly weird that everybody has just gotten to

57:40 token yeah which is

57:42 the same thing that happened with IBM and mainframes people were on MIPS and

57:45 then one day the reality

57:47 was IBM was selling more MIPS for fewer dollars every year yeah and didn't even

57:52 realize it and they

57:53 were still pricing their mainframes by MIPS until it got pointed out to them

57:58 that they were on a

57:59 decreasing curve because they were making MIPS faster than they can charge for

58:02 and that's what's

58:04 going to happen guaranteed I just said that in a hardcore way I thought that

58:07 sounds great yeah

58:09 like sounds really great to sound like I know what I'm making guaranteed I

58:12 actually probably

Love

· 1w

Anthropic首席执行官：

“Anthropic内部有工程师根本不写代码，他们只是让Claude编写代码，然后进行编辑和审核。”

“在Anthropic，编写代码意味着设计Claude的下一个版本，所以本质上是由 #Claude 来设计下一个版本的Claude，虽然不是全部，但大部分都是如此。”

在过去的52天里，Claude团队发布了50多项重大功能更新。

这简直太疯狂了。 #ai

https://x.com/cgtwts/status/2038234212160119034?s=20

中文 English

0:00 我在Anthropic内部有工程师说，

0:02 我已经不写任何代码了。

0:05 我只是让Claude写代码，然后你知道，

0:07 我再添加或检查一下。

0:09 当然，在Anthropic，写代码

0:11 意味着设计Claude的下一版。

0:14 所以我们本质上是让Claude设计

0:17 Claude的下一版。

0:18 不是完全，不是所有方面，但很多方面。

0:21 这个循环开始非常快地闭合。

0:24 所以我看着这个说，哇，

0:26 这真令人兴奋。

0:27 我们对世界能做的事情令人难以置信。

0:29 但同时，这速度也在加快。

0:31 我真的不确定我们还有多少时间。

Love

· 1w

他用40个 #AI 代理取代了整个营销团队 | Relay.app，雅各布·班克

https://www.youtube.com/watch?v=OHJJKrJCSRU&t=416s

中文 English

0:00 我们都应该有数十甚至数百个代理为我们工作。

0:03 所以随着时间的推移，我组建了这支由40个AI营销代理组成的团队来协作。

0:07 我实际上是唯一的营销人员。

0:09 我有一个AI教练，每周一次告诉我，有哪些主题

0:12 是

0:13 真的很有效或者哪些风格对你很有用。

0:15 多做这个，少做那个。

0:16 这就是我在领英上有几篇非常成功的帖子原因。

0:20 例如，有篇帖子是介绍我实际使用的40个营销代理。

0:23 用的。

0:24 那篇帖子目前已有150万次浏览。

0:28 是领英帖子中评论最多的一次，你知道，我一直在领英上

0:31 持续发帖

0:31 一年却毫无结果。

0:33 然后突然我意识到，哇，如果你作为一名初级内容

0:38 营销人员的工作是把YouTube视频写成博客，那么这个工作

0:41 将不复存在。

0:43 一个高质量的营销承包商。

0:46 假设我有四个人，每人每月费用12500美元。

0:49 $50,000每月。

0:51 而我的AI费用每月仅500美元。

0:53 构建AI代理是定义未来30年每个职业的基本技能。

0:58 's

0:58 职业生涯。

0:59 这项技能就像过去40年熟练使用电子表格（如Excel）一样

1:02 是一项必备技能，未来40年，

1:06 构建

1:06 AI代理将成为必备技能。

1:09 我们目前大概只用了1%的AI潜能。

1:13 我想说的是，

1:14 AI被严重低估了。

1:23 我叫Jacob Bank。

1:24 我是Relay.App的创始人兼CEO。

1:27 Relay.App是一个构建AI代理和工作流程的平台，帮你

1:32 代替你

1:33 完成工作。

1:33 我们现在是一个九人团队。

1:35 有五名工程师，两个设计师，一名产品经理，还有我。

1:37 如果没有AI，我们可能需要15人团队才能达到

1:43 现在的工作成果。

1:45 曾几何时，我是AI研究员，后来是创业者，然后是谷歌的长期产品负责人，自2021年起一直在做Relay.App。

1:49 我在过去三年里见证了AI的全面爆发，从非常近的

1:53 视角。

1:57 观点

1:57 。

1:58 我对AI的看法是，它就像个实习生。

2:05 它就是个实习生。

2:06 它完全按你告诉它做的事去做。

2:07 它是实习生。

2:08 这其实是我改变了的观点。

2:10 一年前，我说，嘿，把AI当成你的实习生。

2:13 后来我意识到这个思维模式错了。

2:15 AI确实可以做一类任务，

2:19 那就是节省时间的工作，否则你会交给实习生做。

2:22 实习生。

2:23 但AI也能做一些战略分析、竞争分析和内容创作工作，

2:27 这是实习生绝对做不到的。

2:31 例如，我不擅长销售电话。

2:34 我没有销售背景。

2:35 这对我来说不自然，但事实证明，有些销售电话技巧

2:38 非常有价值。

2:40 所以我有个AI教练，审阅我所有会议的文字记录，然后说，

2:43 你本可以更好地谈这个。

2:44 你本可以更清晰地表达这个价值，或者你太容易，

2:46 太快演示了。

2:48 你应该多问一些探索性问题。

2:50 如果你去聘请销售教练，一个好的销售教练每周一次会议就要1万美元

2:55 每月。

2:56 你可以自己建立一个不错的AI销售教练，稍微训练文字记录，

3:01 运行费用每周大概五美元。

3:05 当我承担一个新的角色或职能，而我没有经验

3:11 或不觉得自己是专家时，

3:11 我如何找到合适的教练关系，给教练我做事的原始数据，

3:15 然后由他以正确频率给我反馈，是每次通话后，

3:20 每天一次总结，还是每周一总结。

3:23 or once a day with a summary or once a week with a summary.

3:26 这就是为什么我现在意识到，把AI看成实习生是多么错误。

3:30 当我还在谷歌时，有个职位叫技术主管经理。

3:37 那是个混合职位，你既要是技术专家，要熟悉代码库全貌，

3:40 能审核技术设计，解决最难的问题，

3:44 也要同时是经理。

3:44 你还要负责协调团队中的其他人。

3:46 这是个非常难的职位。

3:50 技术工作和管理工作同时做很辛苦。

3:51 但我认为这就是未来我们的方向。

3:54 我们都需要成为超级个人贡献者。

3:57 让我解释原因。

4:01 如果你的工作是作为初级内容营销人员，将YouTube视频写成博客文章，

4:01 这份工作将不复存在。

4:07 你不能仅仅靠做单一内容再利用的任务

4:10 来维持工作。

4:15 那样。

4:15 如果你的工作是在大型企业管理一千人团队，

4:19 我不想说这个工作会消失，但我认为这个工作

4:19 会变得越来越不重要，因为企业会

4:23 变得

4:23 更小。

4:25 未来会有更多精简小型的企业。

4:26 所以管理1000人的技能会越来越

4:29 不重要。

4:32 重要。

4:33 每个人都会看起来像我。

4:35 我们大约有三分之二的时间是个人贡献者的工作。

4:39 我在做事情。

4:39 我在制作YouTube视频。

4:41 我在编辑博客文章。

4:42 我在发布内容。

4:43 我在与客户交流。

4:44 然后我一天中有三分之一的时间是在协调帮助我的AI团队。

4:48 完成那些

4:48 任务。这种超级个人贡献者未来最令人兴奋的地方是，

4:54 那是

4:54 我们大多数人喜欢的工作。

4:55 我和很多处于初级职位的人聊过，他们都会说，哦，我真希望

5:00 不是仅仅

5:00 写这篇博客文章，我被告知要写的。

5:02 我更有战略影响力，能决定我们做什么和为什么做。

5:06 我认识的每个人，包括我自己在谷歌时，处于

5:09 完全

5:09 组织管理岗位上，都希望能更接近实际工作。

5:12 我希望能回到产品岗位。

5:14 我希望能更接近客户和用户。

5:16 所以我认为未来每个职位都会结合两者的优点。

5:20 两全其美。

5:20 我们都需要足够有战略眼光和资历来决定什么需

5:25 要

5:25 完成，并且能看清“好”的标准，以便给予正确的

5:28 指导。

5:28 但同时我们也需要足够动手，能自己编辑博客

5:31 文章

5:32 、编辑推文和制作视频。

5:34 这就是职场的未来。

5:40 是的，一次看到40个有点让人吃惊。

5:42 但如果你点击任何一个单独的代理，它们都很简单。

5:47 它们各自都有非常具体的任务。

5:49 每次我们制作新YouTube视频，它会自动创建LinkedIn帖子。

5:52 自动完成。

5:53 每次我们做新YouTube视频，它会自动发推文。

5:57 当竞争对手的CEO在Twitter或LinkedIn发布内容时，

6:00 它每周都会提醒我他们在谈什么。

6:02 它查看我们的竞争对手定价，并告诉我是否

6:05 有任何重要变化。

6:06 所以我一个一个地组建了这支AI代理团队来帮我。

6:10 我尝试过让单个代理同时做25件事，但没成功。

6:14 这是和人类员工的一个区别，当你雇一个社交

6:15 媒体营销人员时，

6:19 你希望他们负责所有社交媒体营销，尤其是

6:19 如果

6:23 你是初学者，就先简单开始，创建一个能做一件事的代理，

6:23 再创建

6:27 第二个代理做第二件事，创建第三个代理做第三件事，

6:27 然后可能让一个代理管理它们，适时调用，

6:30 逐步升级。

6:30 不要一开始就建立40个代理的组织结构图。

6:34 从一个开始。

6:35 有一点非常重要：AI代理不是设置一次就不管的东西。

6:37 你不能只是建一个AI机构，

6:39 然后就想“它会永远帮我工作”。

6:40 我一直在调整我的代理。

6:46 比如，我会和新客户做评估电话，花30

6:48 分钟

6:49 和他们一起发现AI在他们业务中的机会。

6:51 我想，如果客户通话后能收到我发的个性化Google文档，里面有五个具体用例和模板链接，

6:54 那该多酷啊。

6:58 于是我建了一个AI代理帮我做到这点。

6:59 我告诉代理，你的任务是把通话记录整理成

7:02 格式漂亮的Google文档，发给客户。

7:06 做了十次之后，我觉得这格式不对，

7:06 客户没看那个Google文档。

7:09 其实我应该直接发邮件里包含内容。

7:10 于是我解雇了那个代理，告诉它，这个任务

7:11 不需要了。

7:13 我改造了那个代理，让它把总结直接放进跟进邮件里。

7:17 有些功能可能不需要了。

7:20 比如一年前我们很重视SEO，

7:23 后来发现这对我们现在来说不是重点渠道，

7:24 它带来的新用户比例很小，且工作量大。

7:26 所以我们决定不再专注SEO。

7:31 我告诉我的AI SEO代理们，现在可以先停工了。

7:32 我以后再找你们。

7:35 暂时停止工作，就像你告诉SEO机构，“我们暂时不需要SEO”，

7:35 未来再合作。

7:36 很棒的是，这里没有情绪负担，

7:40 没有分歧，

7:43 也没有协调成本。

7:46 我只要一个想法，

7:49 “嘿，代理，你能帮我实现这个想法吗？”

7:51 它们会说，当然可以，

7:54 我们开始吧。

7:56 But for now, you can stop doing what you're doing in the same way that you

7:58 would

7:59 tell an SEO agency, we don't need SEO at the moment right now.

8:02 We'll come work with you again in the future.

8:04 And the really cool thing is there's no emotional baggage.

8:07 There's no disagreements.

8:09 There's no coordination costs.

8:11 I just have an idea.

8:12 Hey, agent, can you help me with this idea?

8:14 And they say, yeah, absolutely.

8:16 Let's do it.

8:16 嗯，所以我不想让我们把它看作是取代人。

8:20 我想把它看作是赋予我超能力。

8:22 我希望我有时间做这些营销活动，因为我喜欢做

8:27 它们。它们对公司非常有价值。

8:28 而现在我可以做到这些了。

8:30 我们都想在工作中做得更多。

8:32 我们都是有抱负的人。

8:33 我们都是充满动力的人。

8:34 我们都希望自己能做得更多。

8:35 而现在我们有了实现它的工具。

8:37 思考风险职业和安全职业的观点如何随时间变化，真是令人着迷。

8:46 在我父母那代，安全的职业是在一家大型

8:49 财富500强公司工作40年。

8:54 你知道，退休时或工作50年后拿到纪念表，而风险职业

8:56 是创办自己的公司或每隔几年跳槽一次

9:00 。

9:01 我认为这一切已经完全颠倒了。

9:05 风险最大的职业是你的技能过于依赖某一家

9:05 特定公司的环境，而该环境可能会随着时间变化。

9:08 最稳健的职业是那些自己创业

9:13 并拥有丰富人生经历的人。

9:16 我给你举个例子。

9:20 我们团队中很多人都是我从谷歌招来的，因为我们曾一起共事。

9:22 谷歌，现在我们一起做这个创业项目。

9:23 如果你问十个人，留下谷歌更安全还是加入创业公司风险更大，

9:26 大多数人会说留在谷歌更安全，加入创业风险更大。

9:29 就未来一年现金报酬来说，确实如此。

9:33 谷歌明年会给你更多现金报酬，远胜大多数创业公司。

9:34 但如果你从职业的长远角度看，

9:38 关键是你建立了多广的人脉？

9:38 你掌握了多少新技能？

9:43 你如何推动自己去获得新体验？

9:46 留在谷歌这样的公司风险远大于加入创业或创办一家新公司。

9:50 面对当下AI助手的狂热，有两种反应方式。

9:52 你可以选择怀疑，

9:55 不相信它，

9:57 把头埋进沙子，

10:02 继续按照过去的方式工作。

10:05 这是一种反应方式。

10:06 另一种是觉得，哇，这是个酷工具，将彻底

10:08 改变工作的方式。

10:09 我想站在如何使用它的前沿。

10:10 请做第二种人。

10:12 这会对你更有利。

10:13 我从不想说，

10:18 你应该承担更多风险。

10:18 我不是这样看待这个问题的。

10:20 不是叫你冒更大风险。

10:23 我看待问题的方式是，你应该优化你的职业发展，

10:24 着眼于个人成长和学习新技能。

10:26 通常实现个人成长和学习新技能的方法是进入

10:28 更新颖、更不确定的环境，

10:30 承担更多责任，而对未来发展不那么确定。

10:32 我不认为那是冒险。

10:33 我认为最冒险的是停滞不前。

10:39 所以对我来说，关键是重新定义风险的含义。

10:41 风险就是停滞不前。

10:46 进步就是学习、成长和不断挑战自己。

10:49 所以我总希望大家优化进步和成长，推动自己。

10:51 我没有任何市场营销背景。

10:54 我非常不擅长营销。

10:57 直到一年前我甚至没有用过任何社交媒体平台。

11:01 我讨厌公开分享自己的事情。

11:02 承担营销工作对我来说非常不舒服。

11:05 但我确实做了。

11:09 当我发现自然社交媒体和LinkedIn营销是我们公司成功的最佳途径时，我感到非常沮丧。

11:09 但我花了一年时间努力学习，变得擅长它。

11:11 现在我喜欢它。

11:12 所以，本来可能觉得投入大量精力做这套前所未试的营销策略风险很大，

11:15 其实不然，这是一种利用不确定性

11:17 在新方向上学习和成长的机会。

11:21 我经常思考我的孩子们将生活的世界有多不同。

11:21 有些科技领域让我感到惊叹：太酷了，

11:26 显然这是更好的未来。

11:29 举个小例子，我们住在旧金山的帕特罗山

11:33 非常靠近Waymo和Zoox的主要测试区。

11:34 Zoox，是亚马逊的自动驾驶汽车部门。

11:38 我带孩子们去上学，

11:40 我两个大女儿坐在后面货运自行车上，

11:44 她们一直说，Waymo，Zoox，Waymo，Zoox。

11:44 对她们来说，看到无人驾驶的汽车非常正常。

11:50 我觉得很酷，因为车顶装满了摄像头和激光设备，

11:53 但对她们来说，汽车无需人类驾驶是理所当然的。

11:56 在自动驾驶汽车领域，我很开心，觉得

11:57 她们俩可能永远都不需要自己开车，

12:01 因为这更安全、更方便，且所有车都是电动的，对环境也更好。

12:01 但在其他领域，我的感受就复杂得多了；比如我不想让她们

12:04 ...

12:07 And so I take my kids to school.

12:09 My two older daughters are sitting behind me on a cargo bike and they're

12:12 saying, Waymo, Zooks, Waymo, Zooks.

12:15 And for them, it is so normal to see a car with nobody driving it.

12:19 I think it's cool because they have all like the cameras and lasers on top, but

12:23 it is so natural to them, the idea that a car can be driven by the human.

12:27 And so in the particular area of self-driving cars, I am so happy that I think

12:32 neither of them will ever need to drive a car by themselves because it'll be so

12:36 much safer and so much more convenient and so much better for the environment

12:39 when all these cars are electric.

12:41 There are other areas where I have more mixed feelings like I don't want them

12:47 to

12:47 只是一直刷YouTube短视频，尽管算法非常非常好。

12:50 优秀的算法。

12:51 我一般从两个方面来看这个问题。

12:54 一是我尽量让他们获得尽可能多的真实世界挑战和经历。

13:00 是的，你还是得学会骑自行车。

13:02 骑车时你会擦破膝盖。

13:04 我总是对我女儿说这句话。

13:05 孩子擦破膝盖是好事，对吧？

13:06 因为如果你擦破膝盖，说明你在尝试新的事物

13:08 并冒险，面对现实世界的后果。

13:12 我教孩子学游泳时还有另一个重要的突破。

13:14 他们吞点水没关系，学游泳时难免会这样。

13:18 学游泳时，吞水是正常的。

13:18 偶尔吞点水没关系，因为学游泳就是这样。

13:21 学游泳就是要经历这些。

13:22 所以我让他们在现实中承担一些风险。

13:25 因为我的孩子还小，我还没考虑过

13:28 像ChatGPT或我们现在说的AI助手这类工具

13:33 会如何影响他们的学习旅程。

13:36 还没想到这些会带来什么影响。

13:37 但因为我在做AI相关工作，所以很关注

13:39 孩子未来的职业会是什么样子。

13:40 这是我目前的看法。

13:43 第一，清晰表达什么重要、什么需要做并指导如何做的能力

13:49 永远重要。

13:54 未来只会更重要，因为我们的工作更多是

13:57 高效指挥AI做事，而不是自己全部完成。

14:02 不是自己亲自做所有事情。

14:03 所以我花很多时间，我真的希望孩子学逻辑和哲学，

14:07 因为我希望他们能清晰表达自己的诉求。

14:10 他们想要什么。

14:11 第二，我希望他们思考自己独特的个性和能力在哪些领域

14:16 可以闪耀，并且是AI无法替代的。

14:21 这就是为什么我录制我们自己的YouTube视频，因为它表达了

14:25 我的个性，

14:25 人们可以建立社会联系。

14:28 这将是未来职场的两大基本技能：

14:30 你多擅长表达自己想要什么并告诉AI，

14:34 又多擅长与他人建立社会联系？

14:38 我认为这些将是未来任何职业的持久技能。

14:41 。

14:42 毫无疑问，美国在人工智能领域领先于中国，

14:48 但美国在技术领导力上的优势我觉得非常狭窄。

14:53 美国声称的技术领导力看起来很有限。

14:53 美国人准备用AI做什么？

14:56 可能是做更多的PPT，更多地互相打官司，

15:00 而中国人更专注制造业。

15:04 所以中国会用AI制造更多iPhone、电动车电池、无人机和军火。

15:09 制造更多的无人机和军火。

15:12 去年，美国造了大约五艘船，中国造了约1500艘。

15:18 我在美国、中国和加拿大生活的时间差不多一样多。

15:24 我在美国、中国和加拿大的时间大致相等。

15:24 现在看来，美国和中国似乎都注定要输，而非赢。

15:30 不是赢。

15:30 这两个国家都犯了大错。

0:00 we should all be having dozens or hundreds of agents working for us.

0:03 So over time I built up this team of 40 AI marketing agents to work with me.

0:07 I'm the only marketing person I really have.

0:09 I have an AI coach that once a week tells me here are the themes that seem to

0:12 be

0:13 really working well or the styles that are working really well for you.

0:15 Do more of this, do less of that.

0:16 That's why I've had a few very successful posts on LinkedIn.

0:20 So for example, the post was here are the 40 marketing agents that I actually

0:23 use.

0:24 And that post so far has 1.5 million impressions.

0:28 And the most comments ever on a LinkedIn post, you know, I've been posting

0:31 consistently

0:31 on LinkedIn for a year with no results.

0:33 And then all of a sudden I was like, whoa, if your job is as a junior content

0:38 marketer to take a YouTube video and write a blog post based on that YouTube

0:41 video, that job won't exist anymore.

0:43 A high quality marketing contractor.

0:46 Let's say I had four people at $12,500 per month each.

0:49 That would be $50,000 a month.

0:51 And my AI bill is $500 a month.

0:53 Building AI agents is the fundamental skill that will define every professional

0:58 's

0:58 career for the next 30 years.

0:59 It's a requirement in the same way that knowing how to use spreadsheets like

1:02 Microsoft Excel was a requirement for the last 40 years, knowing how to build

1:06 AI

1:06 agents is a requirement for the next 40 years.

1:09 We are all using it for probably 1% of what we should be using it for.

1:13 Here's what I'll say.

1:14 AI is way under hyped.

1:23 My name is Jacob Bank.

1:24 I'm the founder and CEO of Relay.App.

1:27 Relay.App is a platform to build AI agents and workflows that get work done on

1:32 your

1:33 behalf.

1:33 So right now we're a nine person team.

1:35 We have five engineers, two designers, a product person and myself.

1:37 If we did not have AI, we would probably need to be a 15 person team to achieve

1:43 the same level of accomplishment that we have now.

1:45 Once upon a time I was an AI researcher, then a startup founder and then a long

1:49 time product lead at Google and I've been working on Relay.App since 2021.

1:53 So I've seen the whole AI boom over the last three years from a very close

1:57 vantage

1:57 point.

1:58 My take on AI is it's like an intern.

2:05 It's just an intern.

2:06 It does exactly what you tell it to do.

2:07 It's the intern.

2:08 This is something I've changed my mind on actually.

2:10 A year ago, I said, Hey, think of an AI as your intern.

2:13 And then I realized that that was the wrong mental model.

2:15 There are certainly a class of tasks that AI can do.

2:19 Those are the time saving kind of tasks that you would give otherwise to an

2:22 intern.

2:23 But AI can also do these strategy analyses and competitive analyses and content

2:27 creation tasks that an intern definitely could not do.

2:31 So for example, I am not good at sales calls.

2:34 I do not have a background in sales.

2:35 It does not come naturally, but it turns out there are some techniques that are

2:38 pretty valuable in sales calls.

2:40 So I have an AI coach that reviews all the transcripts of my meetings and says,

2:43 Oh, you could have talked about this better.

2:44 You could have articulated this value or you were too easy to,

2:46 you're to jump into a demo there.

2:48 You should have asked some more discovery questions.

2:50 If you go on hire a sales coach, a good sales coach is like $10,000 a month

2:55 for one meeting a week.

2:56 You can build a pretty good AI sales coach by yourself and a little bit of,

3:01 you know, transcript training and it'll cost you like five bucks a week to run.

3:05 When there's a new role or function I'm taking on that I don't have experience

3:11 in

3:11 or I don't feel like I'm an expert at, how can I figure out the right coaching

3:15 relationship where I can give my coach the raw data of what I'm doing and then

3:20 have it give feedback to me on the right cadence, whether it's after every call

3:23 or once a day with a summary or once a week with a summary.

3:26 That's why I'm now realizing that thinking about AI as an intern was so wrong.

3:30 So back when I was at Google, there was a role called tech lead manager.

3:37 And that was a hybrid role where you were both supposed to be the technical

3:40 expert who knew the entire code base and could review technical designs and

3:44 solve

3:44 the hardest problems, but you were also a manager.

3:46 You were also responsible for coordinating the other people on the team.

3:50 And it's a very hard role.

3:51 It's hard to do both technical work and management work at the same time.

3:54 But I think that's what the future holds for all of us.

3:57 We all need to become super ICs.

4:01 Let me explain why.

4:01 If your job is as a junior content marketer to take a YouTube video and write a

4:07 blog post based on that YouTube video, that job won't exist anymore.

4:10 You cannot build the job just doing single individual content repurposing tasks

4:15 like

4:15 that. If your job is to work at a large corporation and manage a team of a

4:19 thousand

4:19 people, I don't want to say that job is going to go away, but I think that job

4:23 is

4:23 going to become less and less important because companies are going to be

4:25 smaller.

4:26 We're going to have many more leaner, smaller companies.

4:29 So that skill set of managing 1000 people, that's going to be less and less

4:32 important.

4:33 Everyone is going to look like me.

4:35 We're about two thirds of your day is individual contributor work.

4:39 I'm doing stuff.

4:39 I'm making YouTube videos.

4:41 I'm editing blog posts.

4:42 I'm publishing content.

4:43 I'm talking to customers.

4:44 And then one third of my day is coordinating my team of AI agents that helps me

4:48 with those

4:48 things. The thing that's so exciting about this super IC future is that that's

4:54 the

4:54 work most of us love.

4:55 Most people that I talk to who are in very junior roles are like, Oh, I wish

5:00 instead of

5:00 just writing this blog post, I was told to write.

5:02 I had a bit more strategic influence on what we were doing and why.

5:06 And everyone I know, including myself when I was at Google, who was in a

5:09 complete

5:09 organizational management role was, I wish I could be closer to the work.

5:12 I wish I could get back into product.

5:14 I wish I could be closer to the customer and the user.

5:16 And so now I think every role in the future will combine the best of both

5:20 worlds.

5:20 We will all need to be strategic enough and senior enough to decide what needs

5:25 to be

5:25 done and to be able to see what good looks like so we can give the right

5:28 coaching.

5:28 But and we need to be still hands on enough that we can actually edit the blog

5:31 post

5:32 and edit the tweet and make the video ourselves.

5:34 That's the future of the workplace.

5:40 Yes, it's a bit overwhelming to see 40 of them all at once.

5:42 But if you click into any individual agent, they're very simple.

5:47 And they all have very specific jobs.

5:49 One, every time we create a new YouTube video, it creates a LinkedIn post

5:52 automatically.

5:53 Every time we create a new YouTube video, it creates a tweet automatically.

5:57 Every time the CEO of one of our competitors posts on Twitter or LinkedIn,

6:00 it lets me know what they're talking about every week.

6:02 It goes, it looks at our competitor pricing and it lets me know if anything

6:05 important has changed.

6:06 So one by one, I built up this team of AI agents to help me.

6:10 I have not had good luck with trying to build a single agent to try to do 25

6:14 things at once.

6:15 That is one difference, maybe with human employees, where when you hire a

6:19 social

6:19 media marketer, you want them to own all of social media marketing, especially

6:23 if

6:23 you're a beginner, start simple, create one agent that can do one thing, create

6:27 a

6:27 second agent that can do a second thing, create a third agent that can do a

6:30 third

6:30 thing, and then maybe have one agent that sits above them and helps invoke them

6:34 when

6:35 necessary, and then build up over time.

6:37 Do not start with a 40 agent org chart.

6:39 Start with one.

6:40 One very important thing to keep in mind is that AI agents are not a

6:46 set it and forget it kind of thing.

6:48 You can't just build an AI agency.

6:49 Oh, this thing is just going to work for me forever.

6:51 I'm constantly modifying my agents.

6:54 For example, I do assessment calls with new customers where I'll spend 30

6:58 minutes

6:59 with them and we'll identify opportunities for AI in their business.

7:02 And I thought, wouldn't it be cool if they got a personalized Google Doc from

7:06 me

7:06 after the call with what we had talked about with five specific use cases and

7:09 links to templates.

7:10 And I thought this was so cool.

7:11 So I built an AI agent to help me with it.

7:13 I said, AI agent, your job is to take this transcript of the meeting and turn

7:17 into a really nicely formatted Google Doc to send to the customer.

7:20 And after doing that 10 times, I was like, this is not the right format.

7:23 The customers aren't reading the Google Doc.

7:24 I should have just included in the email.

7:26 So I, I fired that agent and I said, uh, we don't actually need that job to be

7:31 done anymore.

7:32 I actually repurposed that agent and said, can you just take that summary and

7:35 put

7:35 it directly into the follow up?

7:36 Sometimes you just don't need that function to be done anymore.

7:40 Like we were very focused on SEO a year ago.

7:43 And then we realized that it's just not the important channel for us right now.

7:46 It's such a small percentage of our new users and it's relatively high effort.

7:49 So we're not going to focus on SEO anymore.

7:51 So I told my AI SEO agents, you guys can stop for now.

7:54 Like, I'll come back to you in the future.

7:56 But for now, you can stop doing what you're doing in the same way that you

7:58 would

7:59 tell an SEO agency, we don't need SEO at the moment right now.

8:02 We'll come work with you again in the future.

8:04 And the really cool thing is there's no emotional baggage.

8:07 There's no disagreements.

8:09 There's no coordination costs.

8:11 I just have an idea.

8:12 Hey, agent, can you help me with this idea?

8:14 And they say, yeah, absolutely.

8:16 Let's do it.

8:16 Um, and so I don't want us to think about it as replacing people.

8:20 I wanted to think about it as giving me superpowers.

8:22 I wish I had the time to do all these marketing activities because I love doing

8:27 them. They're very valuable to the company.

8:28 And now I can do that.

8:30 We all want to do more at our jobs.

8:32 We're all ambitious people.

8:33 We're all motivated people.

8:34 We all wish we could do more.

8:35 And now we have the tools to do so.

8:37 It's fascinating to think about how perspectives on what a risky career is

8:46 versus what a safe career is have changed over the years.

8:49 In my parents' generation, a safe career was working at one large

8:54 Fortune 500 company for 40 years.

8:56 You know, getting your watch when you retire or after 50 years and a risky

9:00 career

9:01 was starting your own company or moving between companies every couple of years

9:05 .

9:05 I think that has completely flipped.

9:08 The riskiest career is if your skills are too tied to the environment of one

9:13 particular company that could change over time.

9:16 And the most robust careers are people who are starting their own companies

9:20 and who are having lots of different life experience.

9:22 I'll give you an example.

9:23 Much of our team, I've recruited from Google because we all work together.

9:26 Google and now we're working on this startup.

9:29 And if you ask 10 people, what's riskier, staying at Google or joining a

9:33 startup,

9:34 most of them will say it's safer to stay at Google and it's riskier to join a

9:38 startup.

9:38 And in terms of your cash compensation for the next year, yeah, that's true.

9:43 Like Google is going to pay you more in cash for the next year than most hours

9:46 will. But if you think they're the long term lens of your career,

9:50 where it's about how broad of a network have you built?

9:52 How many new skills have you built?

9:55 How have you pushed yourself to develop new experiences?

9:57 It is way riskier to stay at a company like Google than it is to join a startup

10:02 or start a startup. There's two ways you can react to the current AI agent cra

10:05 ze.

10:06 You can say, I'm skeptical.

10:08 I don't believe it.

10:09 I'm going to put my head in the sand.

10:10 I'm going to keep working the way I've always worked.

10:12 That's one way to react.

10:13 The second way to react is like, whoa, this is a cool tool that's going to

10:18 completely

10:18 change the way that work is done.

10:20 And I want to be on the cutting edge of how I'm using it.

10:23 Please be person number two.

10:24 That's going to serve you much better.

10:26 I never want to communicate.

10:28 You should take more risk.

10:30 That's not the way I frame the discussion.

10:32 It's not you should take more risk.

10:33 The way I frame the discussion is you should optimize your career for personal

10:39 growth and learning new skills.

10:41 And often the way to create personal growth and learn new skills is to go into

10:46 newer and more in certain environments where you'll have more responsibility

10:49 and you're less sure of what's going to happen.

10:51 And I don't think that's risky.

10:54 I think the riskiest thing you can do is to stagnate.

10:57 And so for me, it's all about reframing the conversation about what risk is.

11:01 Risk is stagnation.

11:02 Progress means learning and growing and pushing yourself.

11:05 And so I always want people to optimize for progress and growth and push

11:09 yourself.

11:09 I have no marketing background.

11:11 I am terrible at marketing.

11:12 I was not even on any social media platforms until a year ago.

11:15 I hate sharing stuff about myself in public.

11:17 It was deeply uncomfortable for me to be the person responsible for marketing.

11:21 I really did that.

11:21 It was tragic for me when I realized that organic social media and marketing on

11:26 LinkedIn was the best channel for us to succeed as a company.

11:29 But I worked really hard at it for a year to learn to get good at it.

11:33 And now I love it.

11:34 And so it could have felt, oh, it's so risky to put so much effort in this

11:38 marketing strategy that we haven't tried.

11:40 No, it's an opportunity to take advantage of uncertainty, to learn and grow in

11:44 a

11:44 in a new direction.

11:50 And I think often about how different the world my kids are going to live in.

11:53 There are certain areas of technology where I'm just like, wow, that's so cool.

11:56 That's obviously better.

11:57 I'll give you one small example where we live on Patero Hill in San Francisco

12:01 is

12:01 very close to the main testing area for Waymo and Zooks.

12:04 Zooks is, I think it's Amazon self-driving car division.

12:07 And so I take my kids to school.

12:09 My two older daughters are sitting behind me on a cargo bike and they're

12:12 saying, Waymo, Zooks, Waymo, Zooks.

12:15 And for them, it is so normal to see a car with nobody driving it.

12:19 I think it's cool because they have all like the cameras and lasers on top, but

12:23 it is so natural to them, the idea that a car can be driven by the human.

12:27 And so in the particular area of self-driving cars, I am so happy that I think

12:32 neither of them will ever need to drive a car by themselves because it'll be so

12:36 much safer and so much more convenient and so much better for the environment

12:39 when all these cars are electric.

12:41 There are other areas where I have more mixed feelings like I don't want them

12:47 to

12:47 be just scrolling YouTube shorts all the time, even though there's a very, very

12:50 good algorithm.

12:51 The general way I think about it is maybe twofold.

12:54 One is I try to give them as many real world challenges and experience as

13:00 possible. Yeah, you still got to learn how to ride a bike.

13:02 You're going to skin your knees when you're riding your bike.

13:04 This is what I always say to my daughter.

13:05 Skin knees are good for kids, right?

13:06 Because if you're, if you're skinning your knees, it means you're trying

13:08 something and taking risk in the real world and dealing with the consequences.

13:12 This is another big breakthrough when I was teaching my kids to swim.

13:14 It's okay if they swallow some water, that's going to happen when you're

13:18 learning how to swim.

13:18 Like it's okay if you swallow water from time to time because that's what it

13:21 takes to learn how to swim.

13:22 So I'm trying to have them take, you know, some risk in the real world.

13:25 The thing that I haven't thought about yet, because my kids are too young,

13:28 I don't know how tools like catch a BT or AI agents like we're talking about

13:33 now, I haven't yet thought about how those are going to influence their

13:36 educational journey.

13:37 But because I'm working on AI, just I do think a lot about what my kids

13:39 careers are going to look like.

13:40 And here's my current perspective on that.

13:43 Number one, the ability to clearly articulate what is important, what needs

13:49 to be done and give guidance on how to do it, that will always be important.

13:54 That will only be more important in the future because our jobs will be more

13:57 of how do we efficiently instruct AI on what it needs to do rather than how do

14:02 we do it all ourselves.

14:03 So I'm spending a lot of time, like I really want my kids to learn logic and

14:07 philosophy because I really want them to be able to very clearly articulate

14:10 what they want.

14:11 And then number two, I want them to think about areas where very unique

14:16 personality and abilities can shine through in a way that AI wouldn't.

14:21 And so that's why I record our own YouTube videos, because it's an expression

14:25 of

14:25 my personality and people build a social connection.

14:28 Those are going to be the two fundamental skills for the future of the

14:30 workplace, how good are you at articulating what you want and telling

14:34 that to AI, and then how good are you building social connection with others?

14:38 And those, I think, are going to be durable skills for the future of any career

14:41 .

14:42 There's no doubt that the US is ahead of China on artificial intelligence, but

14:48 the claim to technological leadership for the United States appears to me to be

14:53 pretty narrow.

14:53 What are Americans going to be using AI for?

14:56 Well, to make a lot more power points, perhaps, and to file a lot more lawsuits

15:00 against each other, yes, the Chinese are much more manufacturing focused.

15:04 So China is going to use AI to make more iPhones, make more electric vehicle

15:09 batteries, make more drones, and make more munitions.

15:12 Last year, the United States built around five ships, China built about 1,500

15:18 ships. I've spent about equal parts of my life between the US, China, and

15:24 Canada.

15:24 Right now, it looks like the US and China are both very intent on losing, not

15:30 on

15:30 winning. Both these countries making very big mistakes.

Love

· 2w

Basecamp 现已通过全新的、功能全面的命令行界面（CLI）向客服代表全面开放，该界面由一款出色的技能所封装，并依托经过全面升级且大幅扩充的 API。这是一种绝佳的方式，既能让客服代表访问 #Basecamp 中的所有内容，又能将其集成到任何地方。 #cli #ai https://x.com/dhh/status/2036860598785356219

中文 English

0:00 有一整批消费者已经把人工智能等同于产品

0:07 攻击，跟扯淡一样。

0:09 人们并不是讨厌人工智能。

0:11 他们讨厌的是因为你想留住你的

0:16 产品而搞出的糟糕人工智能。

0:17 欢迎收听 Rework，这是 37signals 关于更好工作方式和企业管理的播客。

0:21 你的

0:21 业务。

0:21 我是 Kimberly Rhodes，来自 37signals 团队，与我一同的是 Jason Frieden 和 David

0:26 Heinemayer Hanson。

0:28 我们将谈论科技和人工智能相关的内容。

0:31 Basecamp 最近已支持智能代理。

0:35 这样你的 AI 代理就可以在 Basecamp 内工作了。

0:38 我不是专家。

0:39 接下来由你们来说。

0:40 David，你想先说吗？

0:41 当然。

0:43 我先挑剔一下这个词，就像我们刚开录之前讨论的那样。

0:47 “代理友好”这个词不错，不过对我来说更

0:53 关乎无障碍性。

0:54 在传统网页设计中，这就是我们对无障碍性的争论，意味着

0:58 让视障或

1:05 视力有限、色盲等人群更容易使用网站

1:05 那些特殊因素可能让他们难以使用未考虑这些的应用程序。

1:11 所以我们做无障碍工作。

1:15 进行对比测试。

1:20 如果某两种颜色放在一起，效果不佳。

1:22 色弱者看低对比度内容会很困难。

1:24 所以需要调整。

1:28 引入不同的颜色，增加对比度，

1:31 针对完全失明的用户做这些优化。

1:32 确保键盘访问性很高，能完全用键盘操作。

1:36 在不同元素间切换时，有合适的流程和节奏，

1:39 不会跳来跳去让人摸不着头脑。

1:43 我就是这么看待我们为代理无障碍做的工作。

1:44 这些 AI 代理在许多事情上非常聪明，

1:50 但在使用网站时却还会笨拙。

1:51 它们勉强能用，但速度很慢。

1:55 一个月前我们开始做这项工作时，我做了个大测试，

1:55 想着我们是不是根本不用改进。

1:59 目前的代理、现成模型，云端代码和开源代码等，

2:00 它们能不能直接做到？

2:07 它们到底需不需要帮助？

2:10 我惊讶于这些代理仅用浏览器操作竟然如此成功。

2:15 我能让其中一个代理注册 Basecamp，

2:21 在 Campfire 自我介绍，

2:23 Vizzy 和 Hey 代理也能做到，尽管速度非常慢。

2:29 那种感觉是，我能看到未来，但不知道未来何时到来。

2:33 看起来至少还有一年。

2:33 可能是两年，五年也说不准。

2:35 如果我们今天想让 Basecamp 成为使用代理的好地方，

2:42 得让它速度快起来。

2:46 没人会有耐心等代理在聊天界面慢吞吞地处理任务。

2:47 它们只是机械地输出信息，对吧？

2:52 然而这个速度必须适合在 Basecamp 中使用它们。

2:58 这就是所有代理无障碍工作的意义。

2:59 就像是铺了条小斜坡一样，

3:02 轮椅无障碍，你本可以用别的方法上去，

3:05 但很麻烦甚至有点危险，

3:09 有了斜坡就能轻松滑上去。

3:12 这就是我们给 Basecamp 做的事，

3:15 创建了命令行界面（CLI），

3:21 AI 代理可以用这个工具。

3:24 过去六到九个月中，我们发现给 AI 这些终端工具时，

3:26 它们变得超级强大。

3:29 这就是最近三四个月代理加速开发热潮的原因，

3:31 程序员和设计师突然能做更多事了，

3:34 不仅仅是向训练有素的 LLM 提问，

3:37 然后得到答案。

3:40 而是让代理去执行操作，代理会尝试完成任务。

3:44 就像人一样，有时不会做，

3:48 调用错误工具，犯各种错，

3:49 但如果能持续反馈，反复尝试调整，

3:52 不断试错，最终成功，

3:58 就能快速前进。

4:04 但这需要工具使用快速。

4:09 虽然我很佩服代理能用浏览器，且用得不错，

4:14 但了解它们为何这么慢后就明白了，

4:18 现在它们实际上是在截图，

4:22 通过图像分析分解界面元素和字段，尝试理解，

4:23 非常厉害，但非常繁琐且慢，

4:30 相比之下如果用命令行界面。

4:30 只需要文本。

4:35 这正是这些大语言模型被训练的内容，数万亿的文本数据。

4:39 它们读文本，预测下一个命令，

4:43 下一段文本。

4:43 and then succeed, it can really quickly move on.

4:47 But that loop requires that the tools are fast to use.

4:51 And as impressed as I was that the agents could use a browser and could use a

4:56 browser really well, you also realize that when you know how they do it, why it

5:02 's so

5:02 slow, right now the agents are literally taking like a screenshot of the screen

5:08 and then running it through this image analyze to break down what are all the

5:13 elements, what are all the fields and then trying to reason about it.

5:16 Very impressive, as I said, but also very cumbersome and very slow versus if

5:22 they

5:22 do all this work through a CLI, a command line interface, that's just text.

5:26 Literally what these LLMs have been trained, like trillions of tokens.

5:32 They've gone through text, text, text, what's the next command?

5:35 What's the next token?

5:36 他们在这方面又快又好。

5:39 所以如果它们能保持在那个小循环里，你就能得到每秒的标记数，

5:43 对吧？

5:44 这很接近就像是在写你的傻故事时的速度。

5:47 而这些之间的差别就是一切，对吧？

5:52 速度，正如我们常说，是一个特性。

5:56 当代理能为你做事，而你又有一些

5:59 耐心去等待时，速度就是一切。

6:02 这是“我懒得做”和

6:04 “让我让代理去做更容易”之间的差别。

6:07 我一直在试验我们新代理在 REMC 或 CLI 上的可访问性。

6:14 尤其是最近用的最快的代理之一，

6:20 Kimmy K25，一个超级快的开放式等待模型，真的让我震惊，

6:26 如果你让它写故事或解决逻辑问题，

6:30 可以达到每秒200个标记的速度。

6:32 也就是每秒100多个词。

6:36 超级快，对吧？

6:38 然后我出于好玩，让它帮我设计一个基地营项目，

6:43 规划基地营五号的发布活动。

6:47 当然，我并没有给它所有背景资料。

6:49 所以它不会给我完整的发布计划。

6:53 但它列出待办事项的速度，拆解得很细致，

6:59 加了备注，加了信息，还添加了日程项目。

7:04 比如说如果你四周后要发布，你得提前一周开始做这个。

7:07 你得在前一周做那个。

7:09 我当时简直惊呆了，这太不可思议了。

7:13 很有趣的是，当你想为什么这感觉特别时，

7:19 如果我18个月前问 ChatGPT 3 给我出一个基地营发布计划的文本，

7:25 它也能做出来。

7:29 我当时可能也会觉得挺不错的。

7:32 但我拿到它后，接下来该怎么办呢？

7:34 这就是代理可访问性的魔力。

7:36 你把所有的智慧、洞察、总结，还有我们

7:42 喜欢的大型语言模型的特点，

7:46 变成可用的东西。

7:49 它不仅仅是一堆文本块。

7:51 不，它是一个完整的项目。

7:54 这里有一些待办事项，有些分配给你，

7:56 有些分配给代理，

8:00 有些分配给 Jason 或者其他参与这项工作的团队成员。

8:05 突然间，所有的智能都变得可执行了。

8:08 说这些话的时候真的得小心，别听起来像是 Salesforce 的机场广告。

8:13 智能变得可执行，砰！但有时候这些

8:18 惊奇的发现确实让你感叹，“哇，真是太棒了”。

8:24 就像我用这些代理进行编程一样。

8:30 我曾经很长时间非常好奇，

8:33 也非常着迷。

8:35 我喜欢问它各种问题。

8:36 然后我又自己动手写了点代码。

8:39 就是用我这两只小手，突然一切都变了，因为

8:44 代理开始能做事了。

8:45 它们能用终端，

8:46 能运行代码，

8:47 能运行测试，

8:49 能重写测试失败的代码，

8:51 能进入那个循环。

8:53 我从“哦，这挺不错”

8:56 变成了“这真是太疯狂了”。

9:00 你去做，我需要时会介入。

9:03 如果我们能把这种能力推广到其他一切上，比如项目设置，

9:09 分工，进度检查

9:13 和回访，

9:14 当这些超能力触手可及时，

9:19 因为所有内容都在基地营共享，

9:23 这不仅仅是我和代理的私人对话。

9:26 整个公司，整个项目，每个人都能围绕这些工作

9:30 协作，且不止用一个代理，

9:34 而是多个代理。我们已经在基地营大量实践了，

9:40 它们驱动着一些内部流程。

9:42 我们有很多不同的代理。

9:44 有些是个人人工代理，代表他们行动，

9:48 还有一些自动代理也做这些事情，并进行跟进和提醒。

9:52 我知道这种爆炸式发展在许多先行公司的

9:53 领域内正在发生，

9:57 但还有剩下98%的世界，

10:01 他们或许用过 ChatGPT，但未用过代理。

10:07 他们没用过命令行界面（CLI）。

10:09 如果你很活跃在 X、推特等平台，看到大家高速发现、分享这些，

10:15 你可能会觉得整个世界已经这样在运作了，

10:20 其实不是，完全不是。

10:22 我们现在所做的工作，比如代理可访问版的

10:25 基地营，使用 CLI、技能等，

10:29 甚至还

10:29 没能覆盖

10:33 广大群众。

10:34 我们还差一步，让它像用 ChatGPT 那样简单，

10:35 而 ChatGPT 几亿人都使用了，

10:42 并很快变得主流。

10:44 我们得跑向冰球会去的方向。

10:46 我觉得这就是这项工作令人兴奋的地方，

10:49 近两年来我们内部做了不少 AI 混合特性实验，

10:55 就是那些把 AI 融入产品

11:02 本身的尝试，比如说，它可能帮你总结，

11:06 给你建议，或以其他方式辅助你。

11:06 大多数公司的这类工作

11:10 成绩可以说是不太稳定。

11:11 如果不那么宽容的话，可以说是完全失败了。

11:18 事实上，情况糟糕到一部分消费者

11:22 已经把 AI 等同于产品退化和胡扯。

11:28 这是事实。

11:34 微软上周刚不得不公开道歉，说抱歉，

11:39 我们把AI乱塞进了画图、记事本，还有Windows的各个角落。

11:46 我们听到了，你们不想要那个。

11:50 当然你们不想要，因为这只是附加上去的，实际上

11:54 并不

11:54 够有用。你把它和消费者对明显有用的东西的兴奋感对比一下，

11:59 比如ChatGPT和其他基于聊天的界面，

12:04 几亿人在用它。

12:07 他们认为它是日常生活不可或缺的一部分，并且为此付费，

12:12 这些东西都说明人们不是讨厌AI。

12:15 他们讨厌的是被乱塞进来的糟糕AI，只因为你想把它

12:20 套进你的产品里，对吧？当然有好的方法。

12:24 有技术可以把AI嵌入产品，让它

12:29 完全

12:30 原生且不可或缺，人们不会拒绝，但这很难。

12:35 正如我说的，我们试了几年，看能不能找到很多

12:40 切入点，尝试了很多方法，但几乎没发布任何成果。

12:45 因为没达到最终标准，我们不想像记事本、画图那样发出烂产品遭到反弹。

12:49 所以我们继续努力，我相信我们会成功。

12:53 我们会找到它真正合适的位置，且非常合理的方式。但在那之前，

12:56 我们可以给大家提供一个简约且易用的Basecamp版本，

12:59 一个他们可以用现有喜爱工具操作的友好版Basecamp。

13:06 比如没有人在每天用Plot代码或Open代码时会说我不想要那个。

13:10 没人会说我不想能用指尖控制Basecamp，

13:14 不想把它连接到GitHub或Sentry或者其他任何东西。

13:17 他们会说想，因为这是显而易见的胜利。

13:21 这也是一种建议。

13:24 我想说的是，任何想搞清楚AI在产品中作用的人，

13:28 但你知道吗？

13:31 在你弄明白之前，如果还没弄明白，就先让它易用。

13:36 让代理更容易使用它。

13:38 这就是解锁我们向客户承诺了二十年的API所有潜能的粘合剂。

13:42 但99.9%的人从未使用我们的API，因为那需要程序员，

13:44 还得做各种准备，太贵了。

13:49 让代理易用，CLI，技能，这一整套把所有承诺带了过来，

13:53 虽然不能完全端到桌上，

14:00 但我们差不多了，不那么麻烦

14:03 的一盒子，很多人都能打开。

14:05 我猜年底前

14:11 这实际上会成为主流，因为会以大家都用的界面批量发出。

14:14 好了，现在代理可以用Basecamp了。

14:17 其他产品有什么打算？

14:18 嘿，Fizzy，那些会推出吗，还是只限Basecamp？

14:20 哦，全部都会推出。

14:23 实际上Fizzy的无障碍做得挺好，因为我们的朋友Rob发布了一个开源CLI。

14:27 其实我们还雇了Rob Salkas来帮我们做Basecamp CLI，

14:27 因为他做Fizzy CLI做得很棒。

14:31 我们会应用很多经验，继续打磨Basecamp CLI。

14:35 这是个AI的启发时刻，我们想让Basecamp更易用，

14:37 于是我们做了CLI，

14:42 而代理实际上写了大部分CLI代码。

14:42 CLI绝大多数代码都是代理写的，

14:43 他们很快就完成了65%。

14:51 然后我们花了几周时间把完成度提高到97%，虽然不是100%。

14:55 现在我们可以把学到的经验，应用到Rob做的Fizzy上，

15:01 推出官方完整打磨的Fizzy CLI和技能。

15:04 接着，百分之百我要这个，

15:08 我要让它支持我的邮箱。

15:09 当我用Fizzy或Basecamp时，把邮箱和它们连接起来，

15:14 把所有东西串联，配一个可以听我指令的执行代理。

15:17 比如我经常用的旅行，就是让代理直接去查信息，

15:20 我只想拿邮件里的一个信息，

15:24 不想翻邮件。

15:25 不想看邮件内容，

15:31 只想要邮件里的事实。

15:35 我很期待给邮箱和日历都全面无障碍，

15:36 把它们串联起来。

15:39 我们会全部做到。

15:42 今后这都会是默认配置。

15:47 代理的无障碍不仅适用于直接使用应用，

15:52 也适用于连接更大的生态系统。

15:57 很多成功应用，比如Slack，一个很好的例子。

15:57 Slack之所以成功，是因为它有一个封闭的生态系统和集成。

16:04 代理无障碍正在让这一切变得开放给所有人。

16:07 所有壁垒都被打破了，

16:11 只在一个应用里有什么优势都不存在了。

16:13 实际上，游戏的关键是你的代理能和你用的任何东西对话，

16:17 访问任何数据，

16:18 不管数据在哪里，都能把它调回来或散布到各处，

16:19 把所有东西连接起来。

16:21 I just want the fact that's in the email.

16:23 I'm really excited for having full accessibility for, hey, both on the email

16:29 side

16:29 but also on the calendar side and tying all these things together.

16:32 So we'll do it for all of it.

16:33 And I think anything going forward, this is going to be baked in.

16:36 This is going to be table stakes that your application is, agent accessibility

16:41 accessible,

16:42 both for the direct use with that application, but just as much because it

16:47 kind of plugs in to this broader ecosystem.

16:50 So many successful applications over the year Slack is a good example.

16:54 We're successful because they ended up with this ecosystem that was kind of

16:58 proprietary to Slack and integrations with it.

17:00 What the agent accessibility is doing is basically bringing that to everyone.

17:03 Like all those moats just come tumbling down because there's no specific

17:08 advantage

17:08 to just having something inside of one application.

17:11 In fact, the whole game here is that your agent can talk to anything you use,

17:17 can access anything you use, wherever your data is, it can move it back here or

17:23 everywhere and tie it all together.

17:24 这真令人兴奋。

17:25 好的，你可以在 Basecamp.com/agents 上了解我们正在做的事情。

17:31 这是 37 Signals 的一部作品。

17:33 你可以在我们的网站 37signals.com/podcast 找到单元内容和文字记录。

17:38 视频

17:38 剧集在 YouTube 上。

17:39 如果你有关于更好的工作方式和管理的提问给 Jason 或 David，

17:43 台词是什么？

17:47 正要说那个。

17:48 我想，等等，我需要代理帮我说这句话。

17:52 如果你有关于更好的工作方式和管理的提问给 Jason 或 David，

17:56 业务。

17:57 给我们留言语音，网址是 37signals.com/podcast question。

18:01 或者你可以发邮件到 [email protected]。

18:05 那不是。

18:05 是这样吗？

18:06 你们，这就结束了。

18:08 是邮箱地址，对吧？

18:09 我想是的。

18:10 哦，天哪。

18:11 让代理检查一下。

18:12 真的，我需要代理来做结尾。

18:17 (欢快的音乐)

0:00 An entire contingency of consumers have come to equate AI with product

0:07 aggression, with bullshit.

0:09 It's not that people hate AI.

0:11 What they hate is shitty AI glob done because you want to stick around your

0:16 product.

0:17 Welcome to rework a podcast by 37 signals about the better way to work and run

0:21 your

0:21 business.

0:21 I'm Kimberly Rhodes from 37 Silver Hills team joined by Jason Frieden, David

0:26 Heinemayer Hanson.

0:28 We are talking about tech things and AI things.

0:31 And Basecamp has recently been made agent friendly.

0:35 So your AI agents can work within Basecamp.

0:38 I'm not an expert.

0:39 I will let you guys talk about it.

0:40 David, do you want to kick us off?

0:41 Sure.

0:43 I'll start by quibbling with the term as we did just before hitting record.

0:47 Agent friendly, I think is a fine term, but I think it's actually to me more

0:53 about accessibility.

0:54 And this was the contention we had about what does that mean traditionally in

0:58 web design, accessibility means making it easier for folks who might be blind

1:05 or

1:05 might have limited vision, might be color blind, have all of these factors

1:11 that make it difficult for them to use a website or an application that has not

1:15 been designed within any of those factors in play.

1:20 So then you do accessibility work.

1:22 You do contrast testing.

1:24 Oh, if you have this color next to this color, that doesn't work very well.

1:28 If someone's kind of blinder, they can't see low contrast well.

1:31 So you change it.

1:32 So you get some different colors introduced, you increase the contrast,

1:36 you do all those things for full on blind users.

1:39 You make sure that keyboard accessibility is really high, that you can do

1:43 everything with a keyboard.

1:44 When you tap from one element to the other, it goes in a flow and a rhythm

1:50 that makes sense.

1:51 And it doesn't just jump all over the place in something that doesn't make

1:55 sense.

1:55 So that's actually how I think about the work we've been doing for

1:59 agent accessibility.

2:00 You have these AI agents that are incredibly smart at a ton of things.

2:07 And then they still kind of stumble when they have to use a website.

2:10 They can barely just do it, but they're really slow.

2:15 I did a big test a while back about a month ago when we set off to do this

2:21 work, where it was like, I wonder if we even need to do anything.

2:23 Can the agents we have right now, the models that are out there, the harnesses

2:29 that people run, cloud code and open code and all the other ones, can they

2:33 actually just do it?

2:33 Do they even need any help?

2:35 And I was shocked by how successful the agents were at using just a browser.

2:42 I could tell one of these agents to sign up for Basecamp, to introduce

2:46 itself in campfire.

2:47 The same thing with Vizzy, the same thing with, Hey, but it was really slow.

2:52 It was one of those things where like, I can see the future, but I can't see

2:58 when it's going to arrive.

2:59 This looks like something that might be at least a year out.

3:02 It could be two, it could be five.

3:05 Who knows if we want to make Basecamp a great place to work with

3:09 agents today, we got to make it fast.

3:12 No one's going to have the patience to sit around and wait for agents,

3:15 take minutes and minutes to do work when you ask them in a chatbot interface.

3:21 And they just, they're spitting out facts, right?

3:24 If that kind of pace needs to be captured.

3:26 When you're working with them in Basecamp.

3:29 So that's what all this agent accessibility work is for.

3:31 It's sort of like a little, little ramps, right?

3:34 Like the wheelchair accessibility, like maybe you could get up with your

3:37 wheelchair some other way, but be very cumbersome and maybe even a little

3:40 dangerous, but if we just make it a little ramp, you can just roll right in.

3:44 So that's what we've done for Basecamp here by creating a command line

3:48 interface,

3:49 a CLI tool that the agents can use.

3:52 Because what we found over just the last six to nine months is that AI

3:58 becomes super powerful when you give them these tools inside of a terminal.

4:04 That's what all this explosion of productivity has been about, especially

4:09 the last three or four months on agent accelerated development work,

4:14 like programmers and designers suddenly being able to do way more stuff

4:18 because they're not just asking an LLM that's been trained on the whole

4:22 internet and getting an answer back.

4:23 No, they're asking the agent to do stuff and the agent will try to do something

4:30 .

4:30 And just like a human will realize that often they don't know how to do it.

4:35 They call the tool wrong, they do all these things, but if it can stay in a

4:39 feedback loop where it can try something, adjust its approach and then try

4:43 again,

4:43 and then succeed, it can really quickly move on.

4:47 But that loop requires that the tools are fast to use.

4:51 And as impressed as I was that the agents could use a browser and could use a

4:56 browser really well, you also realize that when you know how they do it, why it

5:02 's so

5:02 slow, right now the agents are literally taking like a screenshot of the screen

5:08 and then running it through this image analyze to break down what are all the

5:13 elements, what are all the fields and then trying to reason about it.

5:16 Very impressive, as I said, but also very cumbersome and very slow versus if

5:22 they

5:22 do all this work through a CLI, a command line interface, that's just text.

5:26 Literally what these LLMs have been trained, like trillions of tokens.

5:32 They've gone through text, text, text, what's the next command?

5:35 What's the next token?

5:36 They're so good and so fast at that.

5:39 So if they can stay in that little loop, you can get the tokens per second,

5:43 right?

5:44 To be quite close to when it's just writing your silly story.

5:47 And the difference between those things is everything, right?

5:52 The speed, as we often say, is a feature.

5:56 And when it comes to agents being able to do things for you and you having some

5:59 patience to wait for it, speed is everything.

6:02 It's the difference between I can't be bothered to.

6:04 It's easier to ask my agent to do it.

6:07 And I've been experimenting with our new agent accessibility on REMC or CLI.

6:14 And I've been mind blown how quickly, especially the fastest agents like

6:20 Kimmy K25 that I've been using recently, which is one of these super fast, open

6:26 wait models that if you just ask it to write a story or solve a problem, logic,

6:30 you can do 200 tokens a seconds.

6:32 That's whatever, 100 plus words a second.

6:36 Super fast, right?

6:38 And then I asked it to actually, just for the heck of it, design me a base

6:43 camp project that spells out the launch campaign for base camp five.

6:47 And of course, I didn't even give it all the context.

6:49 So it wasn't going to give me the launch campaign we're going to ship with.

6:53 But the pace by which it filled out the to-do list, it broke it all down.

6:59 It added comments, it added messages, it added items on the schedule.

7:04 Well, if you want to launch in four weeks, you got to start doing this

7:07 the week before you got to do this.

7:09 I was just like, holy, smoly, this is incredible.

7:13 And it's so funny because when you think about it, why does that feel special?

7:19 If I had asked chat GPT three 18 months ago about just creating a text

7:25 document with a launch plan for base camp, it would have done that.

7:29 And I would probably also been surprised like that's neat.

7:32 But then what am I going to do with it?

7:34 That's the magic of this agent accessibility stuff.

7:36 You take all the wisdom, the insight, the summarization, all the things we

7:42 like about LLMs and then you make it usable.

7:46 It's not just a block of text.

7:49 No, it's an entire project.

7:51 And here's some to do is some of them are assigned to you.

7:54 Some of them are assigned to the agent.

7:56 Some of them are assigned to Jason or whoever is working on this stuff.

8:00 And suddenly all that intelligence becomes actionable.

8:05 I mean, it's so funny when you talk about these things, you really got to be

8:08 careful not to sound like an airport advertisement for Salesforce or something.

8:13 Intelligence becomes actionable, boom, but sometimes some of these

8:18 revelations actually you sit with and you go like, Oh, wow.

8:24 Yes, much the same way as I've had it programming with these agents.

8:30 Like for a long time, I was very curious.

8:33 I was very fascinated.

8:35 I loved asking it stuff.

8:36 And then I went off and I wrote my little code by myself.

8:39 By my two little hands here and then suddenly things flipped because the

8:44 agents could do stuff.

8:45 They could use the terminal.

8:46 They could run code.

8:47 They could run tests.

8:49 They could rewrite code that failed the test.

8:51 They could get into that whole loop.

8:53 And suddenly I went from like, Oh, this is neat.

8:56 But like I'll write it myself to this is freaking incredible.

9:00 You go ahead, I'll intervene if I need to.

9:03 And if we can get to that for everything else, for setting up a project,

9:09 for for figuring out how to divide the work, for checking in on things and

9:13 checking up on things.

9:14 And we can have those kinds of superpowers at our fingertips because it's

9:19 all in base camp, because it's all shared in base camp.

9:23 It's not just like I have a personal little conversation here with an agent.

9:26 No, the whole company, the whole project, everyone can suddenly

9:30 collaborate around this stuff and not necessarily even with one agent, but

9:34 with multiple, we've started doing this extensively already at base camp, where

9:40 you're driving some of these internal processes that we have.

9:42 And we have a bunch of different agents.

9:44 Some have a personal agent that they're just using to act on their behalf.

9:48 And then we have some automated agents that do all of this stuff too and chime

9:52 in and follow up.

9:53 And I know this kind of explosion is going on at a fair number of companies

9:57 sort of on the leading edge, but then there's the rest of the 98% of the world

10:01 who who may have used chat GPT, but we haven't done anything with agents.

10:07 We're not running into CLI.

10:09 And if you're very plugged into X and Twitter and wherever else that people

10:15 are discovering and sharing all this stuff at high velocity, you think that

10:20 this is how the entire world is already working.

10:22 No, they're not, absolutely not.

10:25 And all this work that we're doing right now, the agent accessible version of

10:29 base

10:29 camp that uses the CLI and the skills and so on, even that is not going to

10:33 reach

10:34 the broad masses.

10:35 We are missing a step yet where this becomes as easy as using chat GPT, which

10:42 hundreds of millions of people have used.

10:44 Like that has actually become very mainstream very quickly.

10:46 But we got a skate to where the puck is going.

10:49 And I think that that's what's exciting about this work, that over the last

10:55 almost two years, we've had a bunch of experiments internally with AI infused

11:02 features, like these are the things where you try to bake the AI into the

11:06 product

11:06 itself and like, well, maybe it can summarize some things or can suggest

11:10 some things or it can help you in other ways.

11:11 And the track record for most companies on that work is let's say mixed.

11:18 And if you're less charitable, say completely fucked, right?

11:22 Like in fact, it has been so bad that an entire contingency of consumers

11:28 have come to equate AI with product degradation, with bullshit.

11:34 Microsoft literally last week had to come out with a mea culpa saying, sorry,

11:39 we shoved AI crap into paint, into notepad, into all these crevices of windows.

11:46 We hear you, you don't want that.

11:50 And of course, you don't because it was tacked on and it wasn't actually

11:54 helpful

11:54 enough. And you got contrast that with the consumer excitement for something

11:59 that's obviously good, like chat to BT and the other chat based interfaces,

12:04 like hundreds of millions of people are using it.

12:07 They're considering it integral to their daily lives, they're paying for it,

12:12 all these things. So it's not that people hate AI.

12:15 What they hate is shitty AI globbed on because you want to stick around your

12:20 product, right? Now, there are ways to do this well.

12:24 There are ways to put AI into a product and have it be something that feels

12:29 fully

12:30 native and integral to it and people wouldn't reject, but that's quite hard.

12:35 We've tried, as I said, for a couple of years to see if we could find a bunch

12:40 of angles on this and we tried many things and we shipped virtually none of it.

12:45 Because it didn't pass that final muster because we didn't want the notepad

12:49 paint backlash of shipping something crappy.

12:53 So we continue to work on that. And I'm sure we're going to nail it.

12:56 We're going to find some ways with this really fits and it really makes a lot

12:59 of sense. But until then, we can give everyone the age and accessible version

13:06 of Basecamp, the in friendly version of Basecamp that they can use with these

13:10 tools they already love. Like no one who's using plot code on a daily basis or

13:14 open code are going to tell you like, well, I don't want that.

13:17 I don't want to be able to just command Basecamp with my fingertips and tie

13:21 it together with GitHub or Sentry or anything else.

13:24 They're going to say, yes, please, because this is such an obvious win.

13:28 So that's also a bit of a recommendation.

13:31 I'd actually say for anyone else who's trying to figure out how AI plays a

13:36 role in the product, but you know what?

13:38 Until you figure that out, if you haven't already, just make it accessible.

13:42 Just make it easier for agents to use this.

13:44 This is that glue that unlocks all the promise that we told customers for

13:49 literally 20 years was going to be there with APIs.

13:53 And 99.9% of them never touched our APIs because it required a programmer.

14:00 It required all sorts of stuff to do it.

14:03 It was just too expensive to do.

14:05 Aging, accessibility, CLIs, skills, this whole bucket takes all of that promise

14:11 and then delivers it not quite on a silver platter.

14:14 We're not quite there yet, but like, I don't know, in a slightly less

14:17 cumbersome

14:18 box, quite a few people can open that.

14:20 And then by the end of the year would be my guess.

14:23 This is actually going to go mainstream because it's going to be shipped in

14:27 bulk

14:27 through these interfaces that everyone uses.

14:31 OK, so base camp is now accessible to agents.

14:35 What are we thinking for other products?

14:37 Hey, Fizzy, are those on the horizon or it's just a base camp solution right

14:42 now?

14:42 Oh, it's coming for everything.

14:43 In fact, we already have some pretty good accessibility for Fizzy because our

14:51 friend Rob put out an open source CLI for it.

14:55 In fact, we ended up hiring a Rob Salkas to work with us on the base camp CLI

15:01 because he did such a great job at the Fizzy CLI.

15:04 We're going to apply a bunch of those lessons that we've now taken really

15:08 polishing the base camp CLI.

15:09 This was one of those AI inception moments where like we're trying to make

15:14 base camp more accessible to the agents and we're making the CLI.

15:17 And the agent is actually making the bulk of the CLI.

15:20 Like the vast majority of the code that goes into the CLI was written by an

15:24 agent.

15:25 And it got like 65% of their of that done in no time at all.

15:31 Then we spend weeks and weeks and weeks and weeks getting it to, if not 100%,

15:35 then 97%.

15:36 Now we can take all of those lessons.

15:39 We can apply it to the work that Rob already did on Fizzy.

15:42 And then we can put out an official fully polished Fizzy CLI and skill.

15:47 And then after that, 100% I want this for, hey, I want this for my email.

15:52 I want my email to be tied together with my Fizzy when I'm using that or my

15:57 base camp

15:57 when I'm using that, bring all these things together and having this executive

16:04 agent that I can just tell to do stuff.

16:07 I mean, one of the things I use, hey, for a lot, for example, is travel and

16:11 having an agent

16:13 to be able to just go in and look these things up when I just want a piece of

16:17 information out of my email.

16:18 I don't want to draw through it.

16:19 I don't want the email.

16:21 I just want the fact that's in the email.

16:23 I'm really excited for having full accessibility for, hey, both on the email

16:29 side

16:29 but also on the calendar side and tying all these things together.

16:32 So we'll do it for all of it.

16:33 And I think anything going forward, this is going to be baked in.

16:36 This is going to be table stakes that your application is, agent accessibility

16:41 accessible,

16:42 both for the direct use with that application, but just as much because it

16:47 kind of plugs in to this broader ecosystem.

16:50 So many successful applications over the year Slack is a good example.

16:54 We're successful because they ended up with this ecosystem that was kind of

16:58 proprietary to Slack and integrations with it.

17:00 What the agent accessibility is doing is basically bringing that to everyone.

17:03 Like all those moats just come tumbling down because there's no specific

17:08 advantage

17:08 to just having something inside of one application.

17:11 In fact, the whole game here is that your agent can talk to anything you use,

17:17 can access anything you use, wherever your data is, it can move it back here or

17:23 everywhere and tie it all together.

17:24 That's super exciting.

17:25 OK, well, you can read more about what we're doing at Basecamp.com/agents.

17:31 This has been a production of 37 Signals.

17:33 You can find units and transcripts on our website at 37signals.com/podcast full

17:38 video

17:38 episodes on YouTube.

17:39 And if you question for Jason or David about a better way to work and run your.

17:43 What is the line?

17:47 Literally about to say that.

17:48 I was like, wait, I need an agent to say this for me.

17:52 You have a question for Jason or David about a better way to work and run your

17:56 business.

17:57 Leave us a voicemail at 37signals.com/podcast question.

18:01 Or you can send us an email to [email protected].

18:05 That is not.

18:05 Is that right?

18:06 You guys, that's it.

18:08 Email address, right?

18:09 I think it is.

18:10 Oh, my gosh.

18:11 Have an agent check it.

18:12 Literally, I need an agent to do the exit.

18:17 (upbeat music)

Love

· 2w

#Optimus #ai

https://x.com/elonmusk/status/2036665153778000143?s=20

中文 English

0:00 （欢快的音乐）

0:02 - 每天我都走进办公室。

0:06 感觉自己像是在科幻电影里。

0:08 当你走进办公室时，

0:09 你会看到机器人自主地走来走去。

0:11 走进实验室，机器人在洗衣服。

0:13 参与 optimists 项目让我看到

0:15 当我们齐心协力时

0:16 能取得多大的进步。

0:19 - 这里很多工程师

0:23 都非常专注于工程工作。

0:26 和这些能干的同事一起上班

0:27 开发一个将改变世界的产品，

0:30 真的非常激动人心。

0:32 - 我觉得 optimists 有其独特优势，

0:34 因为我们同等重视

0:35 设计和制造。

0:37 - 我们解决的是制造领域中

0:39 尚未解决的问题，

0:41 而且是在特斯拉的规模上。

0:43 我觉得自己被一群

0:44 极具热情的人包围，

0:46 他们愿意付出一切

0:48 完成任务。

0:49 - 我觉得自己被赋予权力

0:52 从头到尾负责一个设计。

0:54 在特斯拉，如果你要的解决方案

0:57 不存在，那你必须自己去创造它。

1:00 - 在这个工厂，

1:01 你会参与

1:02 齿轮、制造设备和车床设备的工作，

1:04 还会亲自动手，

1:07 组装你设计的箱体。

1:09 - 我专注于那些

1:12 难以自动化和规模化的分布式线束。

1:15 这里的大部分工作是探索

1:17 能取代线束的新技术。

1:21 你将更多地参与零部件级开发。

1:24 - 我负责 optimist 机器手的设计和制造。

1:27 - 我负责机器人外部

1:29 你现在看到的所有结构。

1:31 - 我负责机器人的印刷电路装配板。

1:33 我们设计的是

1:36 高功率、小型化逆变器。

1:37 - 随着我们迭代机器人设计，

1:39 进入第三代和量产阶段，

1:41 功能和外形正越来越接近人类。

1:44 and form factor.

1:45 - 它看上去甚至不会像机器人，

1:47 而更像穿着超级英雄服装的人类。

1:49 - 我认为最终目标将会是

1:52 革命性的成果。

1:53 这感觉就像是工程学的极限运动。

0:00 (upbeat music)

0:02 - Every single day, I step into the office.

0:06 It feels like I'm in a sci-fi movie.

0:08 When you walk in the office,

0:09 you have bots walking around autonomously.

0:11 You come in the lab, you have bots doing laundry.

0:13 Working on optimists has shown me

0:15 how much progress is possible

0:16 when we're all working towards a common goal.

0:19 - A lot of the engineers here are exceptionally ready

0:23 to get lost in the engineering.

0:26 It's super exciting to come in to work

0:27 with all these capable people to develop a product

0:30 that's gonna change the world.

0:32 - I think optimists is pretty uniquely positioned

0:34 in the fact that we care equally

0:35 about design and manufacturing.

0:37 - We're solving manufacturing problems

0:39 that haven't been solved before

0:41 at the scale that Tesla operates on.

0:43 I feel like I'm surrounded by people

0:44 who are extremely passionate about what they do

0:46 and they're really willing to do whatever it takes

0:48 in order to get the job done.

0:49 - I feel like I'm empowered to take a design

0:52 from start to end.

0:54 Here at Tesla, if the solution you're looking for

0:57 does not exist, you do have to build it.

1:00 - So here at this facility,

1:01 you can expect to be involved

1:02 in working on gear, manufacturing equipment,

1:04 turning equipment, as well as getting your hands dirty,

1:07 assembling your boxes that you've designed.

1:09 - I have a focus on distribution harnesses

1:12 that are difficult to automate and scale.

1:15 A lot of the work here is trying to figure out

1:17 new technologies that replace harnesses.

1:21 You would be doing more component level development.

1:24 - I work on the design and manufacturing of optimist's hands.

1:27 - And I work on all the external structures

1:29 that you see on the bot right here.

1:31 - I work on the PCBAs on the robots.

1:33 We are designing high power inverters

1:36 in a very small form factor.

1:37 - As we iterate on our robot design

1:39 and go into Gen 3 and into mass production,

1:41 we are getting very close to the human functionality

1:44 and form factor.

1:45 - Won't even look like a robot

1:47 and it'll look like a human in a superhero suit.

1:49 - I think the end goal is going to be

1:52 something revolutionary.

1:53 It feels like the extreme sports of engineering.

Love

· 2w

从本周开始，数百万 Shopify 商家可以通过 #ChatGPT 向美国销售商品。
他们的产品详情页、结账流程、自定义设置，无需额外设置。
人工智能是通往商业的新大门，而 #Shopify 则是这扇大门背后的推手。

https://x.com/tobi/status/2036507410399920520?s=20

中文 English

0:00 从今天开始，数百万 Shopify 商户的精彩产品可以

0:04 在 chat GPT 中被发现并购买。

0:07 当人们想到商业时，通常会问，“有多难呢？”

0:11 但其实很复杂。

0:13 它是商品陈列、支付、税务和防欺诈的综合体现，

0:18 还有履单、所有运营业务的逻辑和规则。

0:22 这正是我们与合作伙伴在行业中着手解决的问题。

0:26 合作伙伴。

0:27 你们以前听我们提过与 OpenAI 的合作，我很高兴

0:31 告诉大家，我们终于实现了所有面向美国的商户的大规模接入。

0:37 当买家在 chat GPT 中搜索产品时，Shopify 目录会返回

0:42 所有受欢迎品牌中最相关的选项。

0:46 然后购买可以直接使用 chat GPT 的内置浏览器完成，

0:50 无需离开应用。

0:53 商户保留所有产品详情页和结账页面的定制设置、品牌形象和

0:57 支付偏好，包括对 Shopify 的支持。

1:01 商户无需支付任何额外费用，也不需要安装额外应用

1:04 或进行定制集成。

1:07 最棒的是，跨境贸易，以及订阅、组合销售、预售，

1:12 甚至多件商品的购物车功能，所有商户习惯的功能

1:17 都开箱即用支持。

1:18 没有谁比 Shopify 更懂商业，我们搭建了支撑

1:22 商户和买家在任何界面进行交易所需的全部基础设施。

1:27 任何界面。

1:27 从 Shopify 目录到统一结账流程（UCP），以及世界上最优秀的在线商店和结账系统，

1:33 所有 AI 购物界面和入口都涵盖其中。

0:00 Starting today, the amazing products for millions of Shopify merchants can be

0:04 both discovered and purchased in chat GPT.

0:07 When people think of commerce, they usually ask, "How hard could it be?"

0:11 But it's complicated.

0:13 It's the culmination of merchandising and payments and tax and fraud prevention

0:18 and fulfillment and all the logic and rules that run a business.

0:22 That's precisely the problem that we set out to solve in industry with our

0:26 partners.

0:27 You've heard us talk about our partnership with OpenAI before, and I'm happy to

0:31 say we're finally at scale with all merchants selling into the United States.

0:37 When buyers look for products in chat GPT, the Shopify catalog returns the most

0:42 relevant options from all the brands that people love.

0:46 And then the purchases can be completed using the in-app browser of chat GPT

0:50 directly in the app without ever needing to leave.

0:53 Merchants keep all of their PDP and checkout customizations, branding and

0:57 payment preferences, including support for Shopify.

1:01 There are no additional fees to be paid by the merchant, and no additional apps

1:04 to be installed or custom integrations necessary.

1:07 Best of all, cross-border commerce, as well as subscriptions, bundles, pre-

1:12 orders, and even multi-item carts that all merchants are used to are supported

1:17 right out of the box.

1:18 Nobody knows commerce like Shopify, and we've built the infrastructure that

1:22 powers all the ways in which merchants and buyers need to transact on any

1:27 interface.

1:27 From the Shopify catalog to UCP to the best online store and checkout in the

1:33 world, every interface and ingress for AI shopping is covered.

Love

· 2w

“ #Anthropic 和 #OpenAI 的日志里肯定保存着所有人的访问密钥、API 密钥和持有者令牌。” 如今的人工智能技术栈漏洞远比人们意识到的要多得多。 #IronClaw 解决了这个问题，因为“密钥永远不会接触 LLM”。 @ilblackdragon 联合创始人 @NEARProtocol 正在开发私有 AI 基础设施，其中“模型提供商和硬件提供商实际上都无法访问您用于 #AI 推理的内容”。换句话说，“没有任何一方能够单独解密你的数据”。https://x.com/Bankless/status/2036518654909821158?s=20

中文 English

0:00 人们在使用时没有意识到的一件事是

0:03 Entropic OpenAI，或者更糟的是，你使用其他东西

0:07 进行推理时，Open Cloud 实际上也会发送你所有的秘密

0:11 到那些服务。

0:12 是的。

0:13 所以在 Entropic 和 OpenAI 的日志中某处，

0:15 他们拥有每个人的访问密钥、API 密钥，

0:20 以及用于访问你的 Gmail、Notion 的承载令牌

0:25 和你的——

0:25 我们竟然在做这种事情，真的很疯狂。

0:29 是的。

0:30 首先，我不知道如何解决这个问题。

0:32 密钥绝不会进入 LMM。

0:33 所以即使你用这些集中式提供商，

0:37 虽然不应该，但至少密钥不会

0:39 进入任何 LMM。

0:41 所以这是我们希望做到的，

0:43 这是唯一必须首先做到的事情。

0:45 是的，是的。

0:46 但 Neerai 过去一年一直在努力开发的，

0:50 实际上是如何实现私密 AI？

0:53 也就是说，我们如何提供一种 AI，

0:58 让模型提供商、硬件提供商

1:00 都无法访问你用 AI 推理的内容？

1:05 所以我们有 Neerai Cloud，这是推理云。

1:08 你可以使用开源权重模型。

1:10 它运行在安全环境中。

1:12 它实际上使用——这也就是我

1:14 一开始提到的——

1:15 你会使用我们的多方计算网络，

1:18 这是 Neer 的一部分，用于加密

1:21 解密、备份以及所有内部机制。

1:25 这让你确信，

1:29 没有任何单一方可以解密你的数据。

1:32 没有人可以访问它。

1:35 你需要把所有多方计算网络

1:37 有效地全部集结起来才能访问。

1:40 所以——

1:40 好，那么你的意思是说，

1:43 你提供一种服务，配合

1:46 Iron Claw，类似于保密云

1:50 环境来运行 LMM 实例？

1:53 当然，你得运行开源权重模型。

1:55 但也许一些中国模型

1:57 是最好的，比如 Kimmy 或者类似的，

1:59 或者一些 DeepSeek——

2:00 是的，我们有 Quimmy、Quen、DeepSeek。

2:04 最新的热门我们也会加进去。

2:07 我们还有 OpenAI RSS。

2:10 所以，你可以在它们之间选择。

2:12 现代媒体一团糟。

2:13 每个平台都有偏见，但并不是所有新闻消费者

2:16 都意识到他们所接收平台的偏见。

2:18 这就是为什么当我想知道真相时，

2:20 我会去 Polymarket，世界上最公正的新闻平台。

2:23 从 polymarket.com 获取你的信号。

0:00 One thing that people don't realize when they use

0:03 Entropic OpenAI, or even worse, you use something else

0:07 for inference, Open Cloud actually sends all your secrets

0:11 to those services as well.

0:12 Yeah.

0:13 So somewhere in Entropic and OpenAI logs,

0:15 they have everybody's access keys, API keys,

0:20 and bearer tokens to access your Gmail's and your notions

0:25 and your--

0:25 It's actually insane that we're doing that.

0:29 Yeah.

0:30 And so first of all, I don't know what fixes that.

0:32 The keys never touch a LMM.

0:33 So even if you're using it with those centralized providers,

0:37 which shouldn't, but at least the keys are not

0:39 going ever into a LMM boop.

0:41 So that's something we'd like just like,

0:43 that's just the only same thing to do first.

0:45 Yes, yes.

0:46 But what Neerai has been working on actually for the past year

0:50 is actually developing, how do we do private AI?

0:53 So how do we actually offer AI where neither we, model

0:58 provider, hardware provider, is actually

1:00 able to access what you are using the AI inference with?

1:05 And so we have Neerai Cloud, which is inference cloud.

1:08 You can use open-weight models.

1:10 And so it runs in secure enclaves.

1:12 It actually uses-- and this is kind of what

1:14 I was referring in the beginning--

1:15 you'd use our multi-party computation network, which

1:18 is part of Neer, that is used for encryption

1:21 decryption for backups for all the internal machinery.

1:25 And that's what gives you this kind of knowledge that like,

1:29 hey, there's no single party who can go and decrypt your data.

1:32 There's nobody who can actually access it.

1:35 You would need to collect all the effectively

1:37 multi-party computation network together.

1:40 So there's--

1:40 OK, so is this-- are you saying, then,

1:43 that you offer a service in conjunction

1:46 with Iron Claw, which is almost like a confidential cloud

1:50 type of environment for running LMM instances?

1:53 And of course, you'd have to run the open-weight models.

1:55 But maybe some of the Chinese models

1:57 are kind of the best year, like a Kimmy or something

1:59 like this, or some deep-seak--

2:00 Yeah, we do quimmy, quen, deep-seak.

2:04 Whatever's new hotness will add it as well.

2:07 We have open-air RSS as well.

2:10 So yeah, you can choose between all of them.

2:12 Modern media is a mess.

2:13 Every platform has a bias, but not every news consumer

2:16 is aware of the platform's bias that they're consuming.

2:18 That's why when I want to learn the truth,

2:20 I go to Polymarket, the world's most unbiased news platform.

2:23 Get your signal from polymarket.com.

Love

· 2w

#Groq 提供快速、低成本的推理能力，即使在关键时刻也能稳定运行。

2025年12月24日，NVIDIA宣布以约200亿美元现金收购 #AI 芯片初创公司 Groq 的资产，这是 #NVIDIA 有史以来规模最大的一笔交易。

中文 English

0:00 “推理对于我们在一级方程式所做的一切都至关重要。我们投入了

0:05 巨大的资金

0:06 在技术上。我们运行了数百万次模拟，最终这转化为

0:11 赛道上的表现。

0:12 对与Grock的合作超级兴奋，如果你看一级方程式，

0:16 有一些问题我们真的在努力解决。

0:16 那就是速度，能够

0:19 在严格的成本控制下运营，

0:20 并且在一年参加24场比赛时能够快速扩展。

0:25 每年24场比赛。

0:26 试图同时解决这两个问题非常令人兴奋。

0:31 这正是Grock将

0:33 赋予我们优势的地方。”

Love

· 2w

现在，您可以让 #Claude 使用您的电脑来完成各项任务。

它可以打开您的应用程序、操作浏览器、填写电子表格——任何您坐在办公桌前会做的事情。

在 Claude #Cowork 和 Claude #Code 中预览研究功能，仅限 #macOS 系统。 #ai

https://x.com/claudeai/status/2036195789601374705

Love

· 2w

@davidsenra称，Shopify首席执行官@tobi告诉他，将来我们回望2026年时，会将其视为“全球每一家企业都面临被颠覆的一年”。

“人工智能将席卷一切。”

“届时你会回过头来意识到，这一年本应让人看清：无论现存什么业务，都完全可以重建为原生 #AI 版本。” #Shopify

https://x.com/tbpn/status/2036241506428977519?s=20

中文 English

0:00 托比·卢卡对我来说非常有趣，因为他是你最喜欢的创始人，

0:03 最喜欢的创始人。

0:04 所以你欣赏的每个人，当你问他们欣赏谁时，他们都会提到托比

0:08 。

0:08 他们喜欢他的原因是他对每件事都有独立思考，

0:11 独立思考。

0:11 因此他拥有一系列彼此不相关的信念，

0:15 这意味着当你和他对话时，你无法预测

0:18 他对你的问题或陈述的回应会是什么。

0:21 我们谈了四个小时。

0:24 录音前我们先聊了半小时。

0:26 接着录了三个小时。

0:27 然后我们去了某个地方，录完后又聊了三十分钟。

0:32 我不停地说话。

0:32 我说，伙计，我得赶飞机了。

0:33 那晚我要和查理·罗斯吃晚饭。

0:36 我得飞过去。

0:37 我说我们本来可以继续聊。

0:38 他真让我大开眼界。

0:40 他说的一件事，我觉得你们观众会感兴趣，就是他

0:43 相信。

0:43 我记得他在节目里说过，我们将回顾

0:46 2026年作为这样一年，

0:48 那一年全球所有生意都处于抢夺之中。

0:51 人工智能正全面来袭。

0:54 每一个行业都必须被重新构想，你会回头看，

0:57 那一年你本应该明白，

1:00 你可以

1:01 用人工智能原生版重建所有现有业务。

1:04 很有趣。

1:05 Finn，如果你想让 AI 处理你的客户支持，访问 Finn.ai，

1:10 第一

1:10 客户服务 AI 助理。

0:00 Toby Luca was very fascinating to me because he's your favorite founder,

0:03 favorite founder.

0:04 So everybody that you admire, you ask who they admire and they'll bring up Toby

0:08 .

0:08 And what they like about him is that he thinks to every single thing

0:11 independently.

0:11 So therefore he has all these beliefs that are not correlated with one another,

0:15 which means when you have a conversation with them, you are unable to predict

0:18 what his response to your question or your statement is going to be.

0:21 And that we talked for four hours.

0:24 So we talked for a half hour before recorded.

0:26 Then we recorded for three hours.

0:27 We headed down to whatever was, then we recorded for another 30 minutes after.

0:32 And I kept talking.

0:32 I was like, dude, I got to catch a flight.

0:33 I'm going to have dinner with Charlie Rose that night.

0:36 Like I had to like fly down there.

0:37 I was like, we would have kept talking.

0:38 Like he just blew my mind.

0:40 And one thing he said, I think it'd be interesting to your audience is he

0:43 believes.

0:43 And I think he said this on the episode that we're going to look back on the

0:46 year 2026 as the year

0:48 that every single business in the world was up for grabs.

0:51 That AI is coming for everything.

0:54 Every single other business will have to be reimagined and you're going to

0:57 look back at, this is the year where it should have been obvious to you that

1:00 you can

1:01 rebuild the AI native version of whatever exists out there.

1:04 Interesting.

1:05 Finn, if you want AI to handle your customer support, go to Finn.ai, the number

1:10 one

1:10 AI agent for customer service.

載入中...