本集简介
双语字幕
仅展示文本字幕，不包含中文音频；想边听边看，请使用 Bayt 播客 App。
所以董事会把这事交给了CEO。
So the board goes to the CEO.
董事会怎么说的？
What does the board say?
我们需要更多AI。
We need more AI.
那CEO怎么回应？
And what does the CEO say?
哦，行吧。
Oh, okay.
我就请个顾问来搞点AI吧。
I'll get, like, a consultant to do more AI.
然后他们搞了个集中化的项目，没人搞得懂是怎么运作的。
And then they have some centralized project that nobody knows how it works.
他们连运营都没对齐，这些东西肯定要黄。
They haven't aligned their operations, and those things will fail.
最搞笑的是，有人觉得代码写得越多，就越不需要工程师，但事实恰恰相反——现在系统比以前更复杂了，升级时、系统宕机时、出安全问题时，你得花更多力气去排查：这问题到底怎么修？
The funniest concept that the more code we write, the less we would need engineers would be the opposite because now your systems are even more complex than before, which means that you're gonna be running into even more challenges of when you need to do a system upgrade or when there's downtime and you have to figure out, like, well, how do I fix that problem or when there's a security incident.
所以啊，这方面的工作我们才刚起步。
So yeah, we're just getting started with the jobs on this front.
他们在集成这块迟早会撞墙。
They're going to hit a wall at integration.
AI和智能代理都解决不了一个根本问题：任何超过一千人、或者成立十年以上的公司，都是一堆乱七八糟的系统堆在一起，等着人去整合。
The thing that's not different about AI and that agents don't fix, that nothing fix, is that any enterprise of a thousand people or more or that's older than ten years is just a mass of stuff that's sitting there waiting to be integrated.
你不能光说‘它会自动集成’就完事了。
And you can't just say it's going to integrate.
AI其实根本帮不上忙去整合这些东西。
AI actually does it help to integrate anything.
AI的发展速度确实快得吓人。
AI feels like it's moving fast.
对很多公司来说，真正的变革才刚刚开始。
And for many companies, the real transformation is just getting started.
硅谷能实现的东西，和大公司实际部署的，差距越来越大。
There's a growing gap between what's possible in Silicon Valley and what's being deployed inside large organizations.
工程师们已经在用智能代理和新流程发版了，而大企业才刚开始把这些能力往更复杂的系统和真实场景里落地。
Engineers are already shipping with agents and new workflows, while enterprises are beginning to adapt those capabilities to more complex systems and real world use cases.
这恰恰是个绝佳的机会。
That creates a moment of opportunity.
工具越来越强，公司也在慢慢学会怎么把它们融入现有的工作流、数据系统和决策流程。
The tools are getting more powerful, and companies are learning how to integrate them into existing workflows, data systems, and decision making processes.
与此同时，更深层的变革正在发生。
At the same time, there's a deeper shift underway.
AI 不只是又一层软件。
AI isn't just another layer of software.
它越来越像一个全新的用户，逼着企业重新思考系统、权限和流程该怎么设计。
It's starting to act more like a new kind of user, one that pushes companies to rethink how systems, permissions, and workflows are designed.
本期节目中，a16z 的董事会合伙人 Steven Sinofsky、Box 的 CEO Aaron Levie，以及 a16z 的普通合伙人 Martin Casado，一起聊了聊企业当前哪些做法有效，未来转型又会往哪儿走。
In this episode, Steven Sinofsky, board partner at a16z, Aaron Levie, CEO of Box, and Martin Casado, general partner at a16z, discuss what's working in enterprise today and where the transformation is heading.
嘿。
Hey.
我们现在正在实时关注局势，特别兴奋能聊聊一堆AI相关的话题。
We are here monitoring the situation live, and we're very excited to talk about a bunch of AI stuff.
今天在场的有我们三位。
And we have three of us are here today.
我、史蒂文·西诺夫斯基，还有马丁·卡萨多，他正挥手打招呼呢。
There's me, Steven Sinofsky, and Martin Casado, who will wave and say, hi.
我是马丁。
I'm Martin.
大家好。
And Hello.
嗨，马丁。
Hi, Martin.
还有艾伦·莱维，今天他正在打理发型，我们都很期待看到效果。
And Aaron Levie, is is working on the elevation of his hair today, so we're excited about that.
越来越垂直了。
Just keeps getting more vertical.
我还以为能控制一下，结果没戏。
And I thought I could kinda tame it, but it didn't work.
这是只是令牌的问题，还是参数数量
And is that just a token issue or a parameter number of
太多了？
parameters issue?
参数太多了。
It's too many parameters.
参数实在太多了。
Too many parameters.
我这边正好相反。
Okay, I have the same thing but in reverse.
那你这个，嘿，听我说，你用的是精简模型吧。
Okay, so You, hey, listen, you have a distilled model.
你本地跑的对吧？
There you I run local.
这周事儿特别多，但我们想稍微捋一捋，聊聊现在大方向上到底在往哪儿走。
So we had a lot of there's been a busy week of things, but we're we want to bubble it up a little bit and just start talking about where where things are are heading.
我就不多说了，亚伦，你先说吧——你这周见了这么多客户，肯定有不少新发现，哪儿最让你兴奋就从哪儿讲起。
But I'll I'll let I just kick it to you, Aaron, and you start where you are the most excited this moment because you have visited a ton of customers this week and have learned a lot.
你在X上发了不少东西，但我觉得你可能是最接地气的CEO了，天天跟企业客户泡在一起，这正是我们仨最关注的。
You've shared a lot on X, but I think you're the most in the trenches CEO who is really talking to customers every single day in the enterprise, which is what the three of us tend to look at the most.
是的。
Yeah.
我觉得。
I think
我现在的工作，就是把现实带到硅谷，再把硅谷的幻想拉回现实，尽量两边对上。
my it feels like my job these days is just bring reality to the valley and then bring the valley to reality as as much as possible.
但现在这俩世界之间的差距，简直有点离谱。
And it's it is a kind of a crazy divide that that that exists at the moment.
咱们先退一步说。
Let's ask a couple step back.
我觉得这特别有意思，到底是什么原因呢？
I actually think it's a super interesting what what is it?
这个差距是由什么造成的？
What's the gap caused by?
这到底指的是什么？
What's what is that?
对。
Yeah.
我觉得这个差距，马丁，你肯定也看到了，主要是硅谷和工程岗位的工作方式，跟世界上其他地方的差异造成的。
Well, think the gap Martin, I'm sure you see this, but I think the gap is caused by the styles of work that exist Silicon Valley and in engineering roles versus sort of the rest of the world.
我们之前也聊过几次，虽然说法不一样，但工程师的技术能力真的高得离谱。
So and we've talked about this a couple times in in different forms, but but, you know, the the technical aptitude of an engineer is just like insanely high.
他们对互联网上发生的事儿，了解得简直太深入了。
The level of wired in this to what's going on in the Internet is insanely high.
自己能用顺手的工具、自己做决定，这种自由度简直高得离谱。
The the ability to use your own tools and make your own choices is insanely high.
一旦你选的系统出问题，你马上就能调试，很快就能让它为你所用。
And when things go wrong with the systems that you choose, you can just, like, quickly debug them and then make them sort of work for you.
而且很明显，现在的模型在写代码方面特别强，工作成果也容易验证。
And then, obviously, you have all the benefits of just the models are really good at code and and the work is verifiable.
所以，你有五到十个因素，让智能体在企业或初创公司的工程环境中特别管用——但这些在其他知识型工作里基本看不到，两者之间差了整整一个世界。
So you have, like, you know, five or 10 things that make agents work in an enterprise context for engineering or at least even a startup context for engineering that that tend to be there tends to be a gulf between the way you work that way in engineering and the rest of of sort of knowledge work.
所以我经常在想，怎么把我们看到的代码智能体、能操作电脑的智能体那些厉害的地方，打包复制到企业里？那边流程完全不同，用户技术没那么强，数据支离破碎，系统还全是老古董。
And so and so a lot of what I see is trying to figure out how do we kind of, you know, bottle up all of the greatness that is, you know, what we are seeing from coding agents, what we're seeing from agents that can use computers, to how do you bring that into the enterprise where the workflows are quite different, the users are less technical, the data is much more fragmented, the systems are much more legacy.
所以，这根本就是一道鸿沟。
So that tends to be the divide.
这可不是什么‘政府 vs 企业’那种典型的沟通错位。
So it's not even that we're talking past each other in another classic government versus industry.
这纯粹就是工作方式和技术栈的根本差异。
It's just literally like there is just a pure workflow and technology set divide.
所以，这种变革要从硅谷和全球科技初创公司蔓延到整个知识型工作领域，还得好几年。
And that's why it's going to be a number of years for this sort of diffusion to roll from what we're seeing in Silicon Valley, what we're seeing as tech startups all around the world into the rest of knowledge work.
马丁，接着你刚才说的，你在大公司有丰富的经验。
Martin, just to build on that, you have a ton of experience in big companies.
不过，另一个问题是规模——硅谷初创公司的运作规模，跟其他地方根本不在一个量级。
I one of the other issues, though, is scale and the the way the difference in scale that Silicon Valley operates at at the startup level versus everyone else.
是的。
Yeah.
我觉得这些长期趋势，就像当年互联网兴起一样，其实都是从个人开始的。
I I also think that I mean, these secular trends, like the Internet was like this, actually start with individuals.
大公司做决定往往是自上而下的，而这个趋势却是增长最快的之一。
And big companies tend to make decisions centrally, and this is one of the fastest growing secular trends.
所以，大公司里肯定有不少人已经在偷偷用了，只是他们自己也不好意思说。
So, like, there's probably a lot of individuals in big companies that are doing it where, like Yes.
但公司本身根本不知道该怎么看待这件事。
The big companies themselves don't know even how to think about it.
所以当你听到那些数据，比如MIT说大公司里95%的AI项目都失败了，
And so when you hear stats, like, oh, like, MIT had the stat, like, 95% of AI efforts in big companies fail.
这明显不对劲啊，我敢打赌人人都在用ChatGPT用得挺溜的。
Like, that's clearly silly because I am sure everybody is using ChatGPT very effectively.
他们真正该说的是，你知道的，反正就是那么回事。
What what they really should be saying is, you know, whatever.
我是真听进去了。
Like, I listen.
我也参加过不少董事会。
I sit in these boards too.
董事会就去找CEO说：
So the board goes to the CEO.
他们到底说了啥？
What does the board say?
我们需要更多AI。
We need more AI.
那CEO怎么说？
And what does the CEO said?
哦，行吧。
Oh, okay.
我就请个顾问来搞点AI项目。
I will get, like, a consultant to do more AI.
然后搞个中央项目，结果谁都不懂怎么运作。
And then they have some centralized project that nobody knows how it works.
他们连基本的运营都没对齐，这种项目肯定要黄。
They haven't aligned their operations, and those things will fail.
所以啊，说到‘扩展’，我们通常想到的是系统规模或者人数这些事。
And so I don't you know, when we say scale, often we we think about things like system scale or number of peoples.
但我觉得真正的趋势是发展得特别好，这从那些公司的数据就能看出来。
I think the secular trend is scaling wonderfully, which is being reflected in the numbers of these companies.
可公司自己根本不知道怎么改那些老掉牙的流程——数据、合规、运营这些，都僵化了十几年了。
But organizations don't know how to adjust these kind of age old processes that have been worked on for a decade around data and governance and operations and compliance, etcetera.
我觉得现在Aaron他们公司就正好卡在这个节点上，一边是行业大趋势在往前推，另一边公司的决策层又跟不上。
That's kind of right now where I think air like, Aaron is right between the secular trend and the the organizational decision body.
我们一直都在密切关注这个情况，这几个月总算看到AI在企业里有了一些实际落地的进展，但步子迈得特别保守。还有个情况想补充下，现在业内对AI的质疑声很多：董事会一个劲催着上AI，可之前不少CEO主导的AI项目都砸了，给公司留下了阴影，所以大家都得先跨过这个坎，才有胆量再试第二次。
And this is something that we actually track very closely because we're starting to see now, I would say in the last few months, finally some real kind of inroads into the enterprise, but it's it's it's tepid because and and the last thing I'll say of this, one of the reasons is there's a lot of skepticism because the board wants AI, CEO AI failures have created some amount of bruising, which is, you know, you know, requiring these companies to get past it in order to do kind of the second go at it.
我觉得我们现在就正好面临这样的局面。
And so I think this is exactly where we are.
对。
Yeah.
我完全同意你说的，所以先把共识理清楚才有用，毕竟共识最容易消散得无影无踪
I I a 100% agree with that, which is it's good to start with agreements because we we know how quickly that fade
那剩下的时间咱们就要好好辩论了
We'll disagree the rest of
一整期节目都得杠到底
the show.
可不是嘛
Exactly.
没错。
Exactly.
这大概是我们唯一能达成一致的地方了。
That's the only time we're gonna agree.
我再补充一个关于共识的点。
I I think maybe one more point on the board for agreements.
也许你们几位也会同意这一点。
Maybe maybe you guys would agree.
还有一点挺有意思的动态。
There there's also this very interesting dynamic.
跟前面说的那些比起来，这算是个小问题。
I I would say this is a minor one relative to everything else.
大概只占问题的5%吧。
It's probably the 5% of the problem.
说实话，聊真正的核心问题才更带劲，不过顺带提一下，有个挺有意思的现象——过去你去找工程团队，咱们可以回溯一下历史。
I I think it'd be more fun to talk about the the real problem, but but there's a fun kind of as an aside, there's a fun dynamic where, you go to an engineering team classically for the past Steven, you can take us back in history on this one.
拖延项目最简单的方法之一，就是纠结架构，比如用什么语言、走哪条架构路线。
And one of the easiest ways to stall a project was just getting the architecture, the fights what language to use, what architecture path to go down.
这种争论往往要拖上好几个月，团队得慢慢磨合才能搞定。
That could take months and months to kind of work through as your teams work through that.
因为AI发展太快了，各大实验室你追我赶，隔三差五就出新招，但对怎么部署智能体、它们到底该怎么工作，根本没统一标准。
Because of the pace of change in AI, you actually have this incredible dynamic where the labs are obviously leapfrogging each other so frequently, but with not the exact same paradigm of how you should deploy agents and how they will work.
智能体是跑在电脑里的？
And Is the agent harness in the computer?
还是跑在电脑外面？
Is it outside the computer?
你是在云上跑它？
Do you run it in your cloud?
它是托管服务？
Is it hosted?
它能调用哪些工具？
What tools does it have access to?
这些技术现在还远不是能随便互换的。
This is not a point where these are completely fungible technologies.
这就让人特别纠结，现实中的企业架构团队都会想：我到底该押注哪个方向？之前在AI上踩过一两次坑，选错了路，结果现在整个技术栈都过时了，谁还敢轻易下注？
So that actually creates a bit of paralysis because now as an enterprise architecture team in the real world, you're like, man, what horse do I want to get behind and which architecture path do I want to get behind because I've been burned by doing the wrong thing in AI maybe three or four years ago, and I went down at some path that now is deprecated or not the right strategy anymore.
技术更新这么快，反而拖慢了它在关键业务流程里的落地，因为大家光是做决定就卡住了。
So to some extent, the speed of our change in tech actually reduces the ability for the tech to get diffused into the really, really important workflows because now you have a lot of paralysis in just making decisions.
不过我觉得这也没啥，毕竟现在大家的基础设施、系统和数据还有大量升级空间呢。
So I actually think it's kind of fine because there's still so much upgrade work people need to do in their infrastructure and their systems and their data.
但有个挺有意思的现象：我跟CIO和他们的AI团队聊天时，就会问：
But this is kind of an interesting dynamic where I'll go have conversations with CIOs and their AI teams, and I'll say, hey.
你们现在用什么做聊天系统，或者核心的智能体编排工具？
What what are you using for your chat system or your, you know, core agent orchestration?
他们往往说：是啊。
And they'll say, yeah.
我们正纠结于两三个不同的技术路线呢。
We're in the middle of a debate between these two or three paradigms.
对。
Right.
而且
And
你去跟几乎所有客户聊，都会听到这种顾虑：到底该跟谁绑定？投入多少到一条路上才合适？
it's and and you and you hear that across almost every single customer because there is a little bit of a nervousness of like, who do you get in bed with and how much do you sort of fully lock yourself into one particular path?
但我们也知道，如果你不选一条路，就得一直搞这种双轨并行，架构上也特别费劲。
And we also know that if you don't lock yourself into a path, it's always you're building for this sort of duality, which is you know, also takes a lot of work architecturally.
哦，抱歉。
Oh, sorry.
我真的不好意思。
I'm actually sorry.
不好意思插一句，Steve。
I I I hate to jump in, Steve.
Aaron说得完全对，现在产品公司里就有个特别典型的例子在上演。
Just like there's a like so Aaron is totally correct, and there's a there's a very specific instance of this playing out in product companies right now.
我来告诉你到底怎么回事。
And I'll tell you what it is.
大概半年前，软件产品公司都觉得，集成AI就是把AI真正嵌进产品里。
So so so software product companies, you know, circa six months ago, they viewed integrating AI was like you're actually integrating it into the product.
对吧？
Right?
大家都忙着加个聊天功能啊，或者类似的东西。
So everybody was, like, adding, like, whatever this chat feature or, like, the you know?
那时候感觉就像是融合或者混合模式。
And so it's kind of like this fusion or this hybrid model.
但现在我们看到的不一样了，别把AI当软件了。
What we're seeing instead is instead of viewing AI as software Yes.
干脆就把AI当个用户来看。
Like, just view it as a user.
所以呢，把你的产品改成命令行工具，然后让AI当个代理去用它。
And so instead, like, take your product, make it a CLI tool, and then have the AI be an agent that actually uses it.
所以你不是把两者融合在一起。
So you're not fusing the two.
你只是让AI用起来更顺手。
You're just making it more useful for AI.
这可是个非常重大的架构和思维转变。
This is a very, very significant architectural and mental shift.
对吧？
Right?
我们一开始是纯产品，但后来也不太清楚最终该做成啥样。
And so we started as pure product, and then we didn't quite know what the end thing looked like.
于是搞了个AI和软件的混合体，结果根本行不通。
So we created this, you know, this, you know, AI software hybrid that hasn't worked.
现在我们转向了智能体模式，说白了就是智能体想干嘛都行。
And now we're kind of going to the agentic model, which basically means the agent is gonna be whatever.
它可能是Cloud Code啥的，而我的产品只需要能被它调用就行，这才是真正的使用方式。
It's gonna be Cloud Code or whatever, and then my product now just should be something that can consumed by that, and, like, that's the actual modality.
但你知道，这一年内你已经得把软件重做两次了。
But, you know, within a year now, you've had to rearchitect your software twice.
所以我觉得，不管看行业里哪儿，大家都在纠结：到底最终形态该是什么样。
And so I think no matter but many places that you look in the industry is having this dilemma of actually trying to figure out what the final form looks like.
史蒂文，你应该还记得吧。
And, Steven, you will remember.
还记得那些云的混合版本吗？
Remember all the hybrid versions of cloud?
对啊。
Like, yeah.
记得那些远程桌面啊，一堆乱七八糟的东西吗？
Remember, like, you know, like, remote desktop and all these things?
我觉得我们现在就是在快进，直接冲向最终形态。
Like, I think we're kind of, like, speed running that evolution to the final form.
没错。
Right.
而且我觉得硅谷的人真的很难理解，大公司为什么要提前敲定未来的发展方向——在他们看来这简直太蠢了。
And I and I think that that people in Silicon Valley don't quite appreciate when a big company says, well, we have to map out our bet that we're gonna make because, like, that just seems stupid.
毕竟要是你的职场经历就是在初创公司接连待过两段两年的经历，从种子轮到A轮再到被收购，一路都顺风顺水的话。
And, you know, if you have if your job history is, you know, five, two year stints at startups that went from seed to series A to acquire or something.
那你根本学不到什么深层的东西。
Yeah, you didn't learn anything.
你也根本不会有机会去思考怎么挑选一套合适的应付账款系统这种事，这套系统得
Well, you you never you you don't your frame of reference is not, two: you know, picking an accounts payable system that's
要
going to
最后想想，这40年下来是有实际代价的。
last think 40 there's consequences.
对。
Yeah.
其实我今天准备了好多可视化的辅助材料。
I actually I have like all these visual aids today.
所以在硅谷，最顶级的工程师就是那种越低调越好、越不显山露水越好。
So here's like the ultimate the ultimate engineer if you're in Silicon Valley is Lower lower lower lower lower lower lower.
对。
Yeah.
没错。
Yeah.
就是吉尔福伊尔那种人。
Is Gilfoyle.
吉尔福伊尔就是那种：我谁也不想搭理。
And and Gilfoyle is like, I don't want to talk to anyone.
没错。
Yes.
我就自己写代码，你们该干嘛干嘛去。
And I will just write the code and you go do your thing.
关键是，企业里有些人说：我就用这个模型干我的活，但他们迟早会在集成这一步撞上南墙。
And the thing is, is that you have people in enterprises that are saying, I'm going to use the model and do my thing, but they're only they're going to hit a wall at integration.
而AI和智能代理都解决不了的那个根本问题，就是任何超过一千人、或者成立十年以上的公司，本质上都是一堆乱七八糟、等着被整合的烂摊子。
And this the thing that's not different about AI and that agents don't fix, that nothing fix, is that any enterprise of a thousand people or more or that's older than ten years is just a massive stuff that's sitting there waiting to be integrated.
你不能光说一句‘它会自动整合’就完事了。
And and you can't just say it's going to integrate.
AI其实根本帮不上忙去整合任何东西。
AI actually doesn't help to integrate anything.
所以就算你改了所有东西，人们还是会觉得：‘哦，要是做成智能代理，它就能自己去操作了。’
And so even if you change everything, the people say, Oh no, if you make it an agent, then it can just go ahead and and and be a user.
但如果你真当过用户，比如打过客服电话，你就知道，只要系统搞不定，立马就被转给另一个真人。
But if you're a user, like if you've ever called customer service for something, like literally you get bounced to another human if the system that you're talking to Doesn't doesn't have work.
对方会说：‘那你找经理吧。’
They're like, Well, that's a manager.
或者：‘不对，你说的是付款，不是预订。’
Or No, you're talking about payment, not reservations.
所以我觉得最让人兴奋的是，现在我们终于有实证了——这技术，真没人不喜欢。
So like, what I think is so exciting is that now we have proof of this technology that everybody likes it.
我的意思是，你看那些不喜欢AI的人，总在说律师事务所里出了什么事——因为有人发现了AI胡编乱造，搞得法律案子都毁了，诸如此类。
I mean, you see all the people who don't like AI are saying, look at what's happening in law firms because people are seeing hallucinations and it's ruining legal cases and all this.
而问题出在，真正用得顺手AI的，其实是那些25岁的初级律师，他们早就用了一年了。
And and the reason that's happening is because the 25 year old associate is the one using AI successfully already and had been using it for a year.
其实吧，史蒂夫，情况比这还糟一点，事情是这样的。
Well, Steve, it's actually Steve, it's actually a little worse than that, where it is.
现在公司为了推动大家用AI，居然按调用次数算绩效。
Right now, companies are incentivizing people to use AI by counting tokens.
是啊，确实如此。
Oh, yeah.
对啊。
Yeah.
没错。
Yeah.
对吧？
Right?
所以我不打算说那个人的名字。
And so I'm not gonna say the name of the guy.
我昨天跟一个在那些大公司工作的人聊了，那些公司特别爱这么干，他说他和同事们都让AI代理干些没用的事，纯粹就是为了凑数，真不是开玩笑。
I spoke to someone yesterday who worked for one of these large companies that famously does this, and he's like, me and my coworkers have agents do useless tasks just so that we can no joke.
真的假的。
No.
太离谱了。
No.
可不是嘛。
No.
完全没错。
Totally.
你衡量什么，就会得到什么。
Well, I you get whatever you measure.
所以啊。
So Yeah.
是的
Yeah.
没错
That's right.
这简直就是你说的这种情况的极端版，史蒂文。
So, like, it's like the extreme form of what you're saying, Steven.
对啊
Oh, yeah.
没错
Yeah.
对
Yeah.
有些人明明是假装在忙，产出一堆可能根本没用的东西，就因为他们在用这些
People that are, like, have been fake productive and producing a lot of, you could say, perhaps problematic artifacts just because When they're using these
互联网一出现，公司突然都需要网站了
the internet happened, all of a sudden, company needed websites.
就在不久之前，有个特别典型的例子：每个内部团队都搞了个团队网站，还专门请外包公司来写HTML、做网页。
And so like a very famous moment in time was not too long ago when every internal team had like a team website and they went out and they got like a vendor to write HTML and to create their site.
后来呢，就出现了一个团队。
And then there was a team.
但说真的，在大公司里搞什么团队网站，简直蠢到家了——因为团队过不了六个月就重组一次。
But like, there's nothing dumber than having a team website at a large company because a team gets reorganized like six months later.
所以那时候公司里到处都是成千上万的死网站，这事儿当时就这么叫的。
And so companies were just filled like these with thousands of these dead web is what the expression was.
不过我觉得……你先说。
But I I think but but Go ahead.
你先说。
Go ahead.
不用了。
No.
不用了。
No.
不过我们得好好聊聊你说的那个对接的问题，因为我觉得硅谷确实得认清现实，得看看真实世界里要实现全场景身份识别，这条路还要走多久、最终会是什么样。
But we should we should drill into your integration point because I do think this is something for, you know, sort of some reality to settle in in in the valley on on the real world's sort of journey to fully being identified and what that's gonna take and what that's gonna look like.
还有你刚才说的，根据你要对接的不同岗位的人员，流程要转手给不同的人——智能代理目前也根本绕不开同样的问题，因为哪怕经手的是不同的真人，这些人本身的权限设置就是不一样的。
And and your your point about being passed to the different human, you know, based on the role that you needed to interact with, you know, agents basically don't have any there's no real exception yet for the agent having the same problem because you basically you know, as you pass through a different human, it's it's a different set of access controls that that that that human has.
要是智能代理能跳过任何一道这些流程，立马就会产生安全隐患。所以必须走完这些核验步骤，才不会误拿到不该碰的信息，这也是核验的意义所在。
And if an agent can sort of bypass any of those steps, then then that's how you instantly get the security risks that they that like, you you need to kind of pass through those steps so that way you don't accidentally, you know, get to the wrong piece of information and the verification.
所以要让智能代理能接入所有这些系统正常运转，还有大量的开发工作要做。
And so there's a lot that that you need to kind of build out for agents to be able to go and and work with all these systems.
这个问题我们之前聊过，大部分遗留系统的权限管理都做不到统一权威。
And and we've talked about this, but, like, most legacy environments don't have the most authoritative access control.
所以你作为员工总得一次次去问：“萨莉，那个我没权限的文件能分享给我吗？”
So you're always as a human going and saying, hey, Sally, can you share that thing with me that I don't have access to?
要么就是找鲍勃：“你们数据系统里存的这个指标数是多少来着？”
Or hey, Bob, what's the number inside your data system for this question?
如果AI代理拿到的权限和你完全一样，那它也会处处碰壁，根本没法把流程走完。
So if agents just get the exact same permissions that you had, then they'll just run into these walls everywhere, and they won't be able to complete the process.
但机器人可不像人，不会知道该去找Sally聊聊，或者去问Bob这个问题，结果就直接卡住了。
Unlike a human, they're not going to know to go talk to Sally or ask the question of Bob, so they're going to just be kind of stuck.
所以会出现大量机器人根本访问不到正确数据的情况。
So what's going to happen is you're going have a lot of agents that don't have access to the right data.
它们只能在那些根本不是权威数据源的系统里打转。
Kind of working through systems that are not the real sources of truth for the information.
拿到的数字全是错的。
They're getting the wrong number.
拿到的文档也是错的。
They're getting the wrong document.
这正是现在企业真正要面对的难题。
So this is the real work that enterprises have to go through right now.
好消息是，如果你是初创公司，现在反而是个好时机——你从一开始就清楚这些坑，能从设计上避开这些问题。
The good news is that it's actually a great time, again, if you're a startup, because you get to know all the problems right out of the gate, so you can design your organizations to try and avoid this.
但对大公司来说，真正头疼的是：我该怎么升级我的系统？
But for big companies, there's real work that goes into how do I upgrade my systems?
我该怎么升级我的技术环境？
How do I modernize my technology environment?
怎么才能让这些智能体拿到对的数据、正确的文档和足够的背景信息，好让他们干活儿？
How do I make sure that agents will have access to the right data, the right documents, the right context to be able to do their work?
这事儿才是接下来要解决的核心问题。
That's sort of the work ahead.
当时有个大新闻，说是OpenAI跟Codex合作，还跟埃森哲、德勤这些大系统集成商联手了，网上有些人就冷嘲热讽，我特别感兴趣，因为这其实暴露了科技圈和外界之间的巨大认知鸿沟。
And there was this kind of headline of OpenAI working with in Codex, working with Accenture, Deloitte, all the major system integrators, and there were some kind of you know, snarky comments online around it that that I was fascinated by because it sort of showed how how maybe, you know, great that divide is from the rest of the world versus those in tech.
在我看来，这简直是再明显不过的官宣了，
Because to me, was like the most obvious announcement of all time, which
没错。
is Right.
对。
Right.
大型企业要想让这些智能体真正跑起来，光有技术还不够，还得搞变革管理、系统落地、技术整合，这一套下来可不简单。
A large enterprise is gonna have to go through the the change management, the systems implementation, the integration of technology for these agents to be able to go and work.
所以呢，有人觉得挺讽刺的，说咱们还得靠人来部署那些将来要取代人的智能代理。
And so there was this sort of like, you know, people thought it was somewhat ironic that, oh, we need people to implement the agents that are going go automate the people.
其实呢，这正是关键所在。
And it's like, no, that's exactly how it
就是这样。
works.
没错。
Right.
你真得花大量精力做很多工作，才能让智能代理真正帮上忙，实现自动化。
You actually do need to do lots and lots of work to be able to be in a position where agents can actually go and help you do any of the automation.
所以啊，这类工作会持续几十年，一直都有人做。
So so that and that's gonna be there's gonna be businesses that are doing that for decades.
这对新一代公司，还有那些愿意投身其中的老牌企业来说，都是巨大的机会。
Like, it's gonna be an incredible opportunity for this kind of next generation set of firms as well as existing ones that that lean into that.
我提个观点啊。
Let me throw this out there.
首先，我觉得人们不该为这些失败欢呼，因为它们肯定会失败——就像马丁说的，很多都是自上而下的硬性命令，挑公司里最棘手的问题，觉得‘AI能搞定它’。
Well, first, I think the other thing that people shouldn't celebrate when those fail, because they will fail, because they're as Martin was describing, a lot of them are going to be these sort of top down mandates where they pick like the most acute problem in the company and think, Oh, AI is going to go solve that.
然后IT部门的人肯定要崩溃了：‘天啊，这事儿又来了。’
And the IT people are going to be like, Oh God, that's
对，千万别这么干。
Yeah, the don't do that.
没错。
Yeah.
这简直是世界上最糟的入手方式。
That's the worst system to try to do that.
但CEO、CFO，或者更可能的是客服人员，会特别执着于解决这个问题。
But the CEO or CFO or whatever is going to be obsessed with solving, or the most likely the customer service person will be obsessed.
不过，如果我是给初创公司提建议，想切入企业市场，我肯定会建议：别只想着做产品，第一步就锁定所有无界面的SaaS工具，因为根本没人会用那些传统方案；你要做的，是把你能提供的价值结构化。
But I do think if I were if I were advising a startup specifically in order to to sort of enter the enterprise space in that way, definitely would be thinking about not just like building a company that step one, I only work with all the headless SaaS software that's out there because there just won't be any like the but the thing you can do is structure the value that you offer.
而且这一点也适用于你在公司里该做什么——这其实是个关键的分叉点。
And also, this applies to what you go do in a company is it's really a fork.
这个分歧在于，这个智能体是负责搜集信息然后呈现给人类，
And the fork is, is this an agent that is seeking information and presenting it to to some human?
还是说它要主动去执行任务、做点什么？
Or is this an agent that's supposed to go act and do something like this?
它是去获取信息，还是去动手操作？
Is it acquiring or is it doing?
因为事实上，互联网的发展就是这么过来的。
Because if if it's it turns out that's how what happened with the Internet.
互联网真正变得有价值，是从第一步——让人能访问到信息开始的。
The Internet got very, very valuable when the first step was just providing access to things to people.
没错。
Yes.
突然间，所有那些提供信息整合服务的网站就火了，比如‘我需要按部门查看报销单’，或者‘我想看看我们两家并购公司现在的库存情况’，网页一下子就成了整合数据的枢纽。
And and like all of a sudden, all the sites that were like that literally did integration like, Hey, I need expense reports but viewed by department, or I need to see our current inventory status across like the two companies we've acquired, all of a sudden the web became the integration point.
所以我觉得，如果你能率先出现，说‘我们可以用智能体来搞清楚公司里正在发生什么’，尤其是因为你在这儿，Aaron，像这种跨文件学习的能力，以前根本不可能做到，现在却变得容易多了。
And so I do think that that if you just show up first and just say, Hey, we can actually use agents to learn stuff about what's going on in a company, and in particular, because you're here, Aaron, like learning across the files becomes way more possible than it ever was before.
事实上，AI可能是公司内部搜索第一次真正能立刻产生价值的时候，因为互联网根本不是为提供这类结果而设计的。
In fact, AI might be the first time that inside a company search can actually provide immediate value because the web just wasn't structured to deliver those results.
然后你就会想到，一旦能把这些全都整合起来，就可以加个代理，比如带个‘通过’或‘拒绝’按钮之类的功能。
And and then you start to think, once you could bring them all together, then you can add, like, an agent that has an approve button or a reject button or something like that.
让我最后说说我的不同看法。
Let let me let me just try and provide finally, the point where I get to disagree.
所以……
So
哎呀，糟了。
let Uh-oh.
这下麻烦了。
We're in trouble now.
不会。
No.
真不会。
No.
不对。
No.
不对。
No.
你还可以被邀请回来。
You can get invited back.
你已经被邀请回来了。
You're invited back.
太好了。
So good.
不对。
No.
不对。
No.
我觉得这个观点很有道理，但并不是唯一正确的看法。
I I think this is a very legit view, but it's not the only view.
而且考虑到AI，我觉得这也不是唯一有说服力的观点。
And in light of AI, it's I think it's not the only kind of compelling view.
那我来试着重新说一遍。
So here's the so the let me let me just try and rephrase.
现在主流的看法是，AI就是一种软件。
So the the current view is we've got, like, AI is software.
它的运作方式不太一样。
It it works in a different way.
我们现有的系统已经有一套了，现在得把这种新型软件整合进去，让它能访问数据。
We have a current set of systems, and we have to integrate this new type of software with our existing system so that it can get access to data.
它能做事，但得安全地做。
It can do things, but in a safe way.
对吧？
Right?
所以，这就是为什么我觉得这根本不是关于软件系统进化的问题。
So here's to the kind of end to end argument of why this isn't about about evolving software systems.
端到端的观点是，这些大语言模型是非确定性的。
The end to end argument is these LLMs are nondeterministic.
它们很聪明。
They are smart.
它们能应对那些复杂情况中的长尾问题，而这些恰恰也是人类擅长的。
They deal with the long tail of complexity, and it turns out those are all things humans do too.
我们花了四十年时间，设计界面、流程和系统，就是为了应对人类的混乱和不完美。
And we've spent forty years building interfaces, processes, and design to deal with messy humans.
我们知道该怎么授权，也有一套访问控制机制。
And, you know, we know who to access and we have access control.
所以如果你把智能体当成真人来对待——就像雇了个员工，给它独立的邮箱，让它像人一样访问文档、登录系统、提出需求——那它就能直接利用我们为人类设计的整套流程，而不是为软件设计的那一套。
And so if you have the mindset that an agent is more like a human and you hire the agent, you give it its own email address, it can access documents like humans can, it can log in, it can request the things that it needs, then it will be drafting on all of the process that we've put in place for humans, not for software.
所以在这场讨论里，我想提醒大家，我跟你们一样，也是从小在软件环境里长大的。
And so I would just encourage us as we have this discussion, like, listen, I grew up like you guys in software.
我总习惯把所有系统都当成软件来看，但说实话，这些模型根本没法跟传统软件很好地融合。
I always think of every system like software, but these models don't integrate well with software, actually.
我觉得，作为整个行业，我们逐渐明白：如果你把它们当成人来看，利用我们为人类设计的那些机制，整合起来就会容易得多。
I think it turns out and what we're learning as an industry is if you view them more like humans and you draft on the mechanisms we put in place for humans, they're much easier to integrate.
嗯，确实如此。
Well,
我特别喜欢这个观点。
that's And I think there's I love that.
我，我觉得……
I I
我觉得我们肯定都同意这一点。
think we I think we agree with that for sure.
但问题是，人类拥有一些代理没有的额外优势。
I think the issue is humans have a bunch of extra benefits that the agent doesn't have.
人类能天然地掌握大量背景信息——比如我们能轻松记住公司里各种复杂的人际关系，需要帮忙时，随便拍拍谁的肩膀就能搞定，或者问个问题就能拿到信息。
The human has a lot of context that it gets for that they get that we get for free by virtue of we can keep track of the myriad relationships that we've built in our organization and the person to tap on the shoulder when when we need something done or we need to get information.
但这些，公司里还没形成能让代理直接利用的系统化记录。
That's not documented in a company yet in a in a way that the agent can just sort of draft on.
所以，我的意思是，我想我们都同意，不能把这当成普通软件来处理。
And so so I I, like I I mean, I think we all would agree that that you you can't treat this like software.
你应该把这些代理当成使用系统和工具的人来看待，它们的优势在于能并行工作、无限扩展，但劣势是不知道该找谁帮忙。
You treat these as as people accessing systems and tools, but they are at a they're both at a a massive advantage that they can work in parallel in at, you know, at infinite scale, and they're at a disadvantage in that they don't know who to tap on the shoulder.
嘿。
Hey.
我在听。
I listen.
对。
Yeah.
亚伦，我完全支持代理入职培训。
Aaron, I am all for agent onboarding.
就是说，代理来了之后，先参加个入职说明会。
Like, you know, the agent comes and it goes to orientation.
对。
Yeah.
是的。
Yeah.
对。
Yeah.
然后CEO会来做文化宣讲，我真的没开玩笑。
Then the CEO gives it the culture discussion, and then every I'm not kidding.
不对。
No.
你可能说得对。
You're probably right.
每个人都会全力以赴。
Every guys does a 100%.
每个部门都会做自己的介绍。
Every department does their pitch.
没错。
Yes.
我们就是这么做的。
Like, this is what we do.
对。
Yes.
我的意思是，说实话，考虑到这些代理的技術特性，确实如此。
I mean, I think I I actually honestly think given given the technical nature of these agents Yeah.
而且它们这么混乱、这么难掌控，我们只能沿用之前为人类优化的那一套流程。
And how much entropy they have and kind of how unruly they are, we're gonna have to go through the processes that we've refined around humans.
没错。
Yeah.
完全正确。
100%.
人类可不就有这些特点嘛。
Humans have all of those things.
是的。
Yes.
展开剩余字幕（还有 454 条）
所以我觉得，关键不是去搞什么高大上的索引数据库，而是给它们提供像学校一样的训练环境。
And so I just you know, it's more about providing schools for them than somehow building some, fancy index database.
哦，不用了。
Oh, no.
我完全同意。
Totally agree.
说到这个，我特别喜欢这个类比，因为它其实和‘人形机器人才是最佳机器人’的说法是一回事——毕竟我们这个世界本来就是为人类设计的。
That I mean, of course, what I love about about that is you just keep going with the analogy because what that is is it's the same argument that humanoid robots will be the best kind of robot, which is we have a whole world designed for humans.
我在消费电子展上见过一个机器人进电梯，结果电梯里还另有个专门按按钮的机器人。
I like I I saw at the Consumer Electronics Show, I saw this robot go into an elevator, and then there was a button pushing robot on the elevator.
因为那个机器人太小了，跟扫地机器人差不多，根本够不着按钮。
So because the robot was a tiny little thing that, like a Roomba on the floor, it couldn't push the button.
于是，造这个机器人的公司又专门开发了一个电梯按钮自动按压装置卖给你。
So the same company that invented that robot invented a device you buy for the elevator that pushes the button.
但我当时就问了：既然这样，你们干嘛非得造个机器人去按按钮呢？
But then I asked, why did you need a device to push the button?
这事儿挺有意思的。
And it was very interesting.
他们说，电梯根本没设计成能让机器人接入的系统。
They said because the elevators don't have systems that they can hook into as a a robot.
根本没法通过Wi-Fi让电梯按按钮。
Like, there's no Wi Fi press the button in the elevator capability.
是啊。
Yeah.
这事儿连个API都没有。
So There's no API for that.
电梯压根儿就没有无头版的版本。
There's no there there is no headless version of the elevator.
对。
Yeah.
而且我
And and I
这其实特别形象地说明了我们在企业里用这些智能代理真正要解决的问题——我们目前有两种系统，一种是给人用的，一种是给软件用的，而这些代理更贴近前者。
That's actually a great metaphor for, like, the problem that I think that we're actually solving in the enterprise with these agents, which is we just you know, we have two types of systems, those for humans and those for software, and these tend to be more like humans.
所以我们应该尽量利用现有的人类系统，而不是硬往里塞改造。
So we should draft on those as much as possible rather than try to retrofit them.
不过话说回来，
Well, as Well,
上周的大新闻是，我觉得Salesforce这个动作，不知道算不算让人意外，但从反应来看，好像确实挺出人意料的。
and so so the big news last week was was I I I think, you know, Salesforce I don't know if they surprised people or not, but but, I mean, based on the reaction, it seemed like it was maybe a surprise.
他们直接搞了全头less架构，明确说想让他们的系统能被所有不同的智能代理无缝调用。
They they went full headless, and they basically said, you know, like, we wanna be used everywhere across all of our all of the different agents.
我觉得这有点像风向标，因为Salesforce怎么走，很多企业级软件通常也会跟着走。
And I see that as a a little bit of a bellwether because I think as Salesforce goes, so does a lot of of of enterprise software.
我猜接下来会有很多公司得琢磨，在这个头less的世界里，新的商业模式到底该怎么定。
And I think a lot of people are gonna try and you know, have to figure out what is the new business model in this headless world.
要不要收点小小的API使用费？
Do you charge a little bit of a small just API tax?
那代理有座位吗？
Is there a seat for the agent?
这方面显然还有很多工作要做。
So there's obviously some work to do with that.
Steven，我看到你发过一条推文，提到过这里面的复杂性。
And Steven, saw one of your tweets on some of the complications there.
但我觉得这事挺重要的，因为它说明大家终于意识到：软件以后会一直在后台跑。
But I think as a as a moment, it's a big deal because because I think it's a recognition that that, you know, software will be running in the background.
以前机器和应用早就这样了，现在轮到这些概率性、非确定性的机器用户了。
It always has for machine users and applications, and now it is for these sort of probabilistic machine users and or nondeterministic machine users.
最酷的是，我一看到这个消息，立马就想到了五到十个个人能用的场景——因为我老在做大量客户相关的智能分析工作。
And what's cool and and where I think this gets pretty exciting is, you know, as soon as I saw that announcement, like, I had, like, five to 10 personal use cases where I would need the headless version of Salesforce because I'm always doing just a tremendous amount of customer related intelligence work.
我要去开会了，得先查点资料。
I'm going into a meeting, I need some information.
我要去个城市，该约谁见面？
I'm going into a city, who should I be meeting with?
你可以想象一下，如果能在所有数据系统里跑起这些智能代理，能打开的使用场景简直太疯狂了。
So if you imagine being able to run compute in the form of agents across all of your data systems, the use cases become pretty wild around what that opens up.
我觉得这对很多软件平台来说都是全新的机会——以前你受限于平台上有多少真人用户，但现在，这些无界面的用户数量可以是人类用户的上百倍、上千倍。
So I think this gives a lot of software platforms all new use cases that they can tap into that where, again, you were normally constrained by the number of people on these platforms, but now the headless user can be, you know, a 100 or a thousand x the scale of those human users.
这确实是个让人兴奋的时刻，因为随着越来越多代理在跑、无界面模式越来越普及，这些工具能用的地方就越来越多了。
So this is a I think an exciting moment because as you have more of these agents running around and the headless software modes, you just have, you know, way more use cases for these tools.
我也觉得，关于这一点，我真的特别想说，拜托，真的拜托。
I also think so I just I think on this one Please, please.
最酷的地方在于，第一步当然就像你刚才说的，先去查信息。
What's so what's so super cool is is that, of course, the first step is doing exactly what you described, which is just looking stuff up.
真正有意思的是，把代理看作一个独立实体这个概念。
And so the most interesting thing is using this notion that an agent is just an entity.
对我来说，这明显就是一种新的授权方式。
It's it's incredibly obvious to me that it's another license.
可能授权模式不一样，但它必须有个身份。
Now, it might have a different license model, but it has to have an identity.
比如你去 Box CRM 系统里查信息，不管那是 Salesforce 还是别的，系统默认都得是个人在操作，还得有特定的权限等级。
Like when you go look something up in the Box CRM system, I don't know if it's Salesforce or not, when you use Box CRM system, it has to be a person, like with a certain amount of access rights.
作为 CEO，你可能能看很多东西，但其实很多时候，公司并不希望你在那个时间点拥有那些权限。
And you presumably, as CEO, you might have access to a bunch of stuff, but also there's a lot of ways that they actually don't want you to have the rights at the right time.
比如你能看到客户名单，但没必要让你实时掌握每个销售的业绩配额这些数据。
Like, you might be able to look and see who is on the account, but you don't need the up to date quota of those salespeople and stuff.
这些信息可能属于 H。
And that might be H.
R。
R.
属于敏感内容，你得通过别的渠道才能查看。
Sensitive, and you should probably have some other level to go see that.
往下走，代理的权限永远不能超过它所服务的那个人。
But as you go down the org, the agent is never going to have more permissions than the person who's getting it to go do something.
实际上，它就跟组织里的其他员工平级就行，不然就会出问题——人类员工一拍脑袋：‘给我调个超级智能代理，啥都知道，我权限不够的东西它全都能看。’
And in fact, it's just going to be like a peer to somebody else in an organization, because otherwise you have all of these issues where the peer where a human can just say, Oh, get me the super smart agent that knows everything that I'm not allowed to know.
从IT架构的角度来看，最有趣的是，你必须提前设计好，不能让智能代理先拿到结果，再回头去试哪些能用、哪些不能用。
And to the the IT architecture sense, what's so fascinating about that is you have to build you can't let the agent get the results and then try to figure out what works or not.
就拿马丁说的那些关于大语言模型随机性的问题来说吧，你根本没法预测它会输出什么。
Suppose of all the points that Martin made about the LLM stochastic model, which is you're not going be able to figure out.
这可不是SQL表里的记录，能简单地加个访问权限控制就完事的。
It's not like a record in a SQL table that you could just apply ACLs to.
它可能是句子中的某个词，也可能只是蹦出来的一个数字。
Actually like, it could be words in a sentence or just the number that shows up.
所以，整个‘无头架构’的讨论让我觉得‘SaaS末日’这说法比原来还蠢——本来就已经够蠢了。
And so I actually think that whole discussion about headless for me made the sasspocalypse seem even dumber than it was already, and it was already dumb.
一开始我觉得这想法挺傻的，后来一想，天呐，居然比我想的还要傻得多，因为你根本控制不住这种爆炸式增长。
So like, it was like, at first it was dumb, and then I'm like, Oh my God, it's actually much dumber than I thought it was in the first place, because you're just gonna have this explosion.
当然，也许有人会想出特别聪明的定价方案，比如前五年让智能代理只读，或者必须绑定真人使用之类的，但归根结底，它还是得算一个独立用户席位。
Now, someone might come with a very clever pricing scheme and that agents, you know, somehow cost less because maybe for the first five years they're read only or they're always tied to a person or something, but it is another seat.
这事儿没得商量。
There is no way around it.
如果你是个SaaS公司，还想着让系统直接用别人的账号登录，那简直是疯了。
And like if you're a SaaS company, you're crazy to try to say, Oh, just use the credentials of another human.
这从一开始就是严重的安全漏洞。
Like, that's just that would be like bad security practice from the get go.
没错。
Exactly.
实际上，这种情况在很多领域都在发生。
So actually, in in fact so this is playing out in many domains.
你甚至可以说，所谓的‘无头SaaS’根本站不住脚。
You can even make the argument that, like, a a a headless SaaS doesn't make sense.
我来给你说说理由。
And and here's the argument.
我举个例子你就明白了。
The argument is let let me give you an example.
你用OpenClaw的时候，有没有想过为什么非得用一台Mac mini？
So if you use OpenClaw, do you know why you use a Mac mini with OpenClaw?
嗯，这个嘛，
Well well,
你是说两边都这样吗？
for two side, is that what you're getting at?
不是的。
It's not.
这是两回事。
It's two things.
第一，针对 iMessage。
It's number one for iMessage.
对。
Right.
没错。
Right.
是为了集成。
It's for the integration.
是的
Yeah.
对
Yeah.
没错
Yeah.
因为没有无头版本，所以你只能直接用它。
Because there is no headless version, so you're just gonna use it.
第二个就很有意思了，如果你试过用无头浏览器配合智能代理，问题在于几乎所有网站都有反爬虫机制。
And then the second one is very interesting, which is if you've tried to use headless browsers with agents, the problem is is all of the websites have anti scraping measures.
对。
Right.
没错。
Right.
所以根本用不了。
So they don't work.
所以你用 Back Mini 的原因，就是为了让它能真正用上完整的 Safari。
And so the reason you use a back mini is so it can actually use Safari proper.
但凡你打算搞无头浏览，就等于默认整个互联网都要变成无头的了，可实际上，所有这些模型训练用的数据，都是真人用真实的应用程序操作的，根本不是无头的。
So to do anything headless kind of assumes that, like, the entire Internet is gonna go headless when I think all of these models like, all of the data is humans working on the actual apps that are not headless.
反正所有数据都是这么来的。
Like, that's all of the data anyway.
我觉得这些模型最终只会擅长像现在这样正常使用各种应用，而且我们已经看到这种趋势了。
So I think these models are gonna be very good at just using apps like they are today, and we're already seeing this happen.
真正被用起来的，根本不是无头版本，而是带界面的正常版本。
And rather than the headless versions, the non headless versions are what's actually being used.
所以你甚至可以说，这其实就是 Salesforce。
So you could argue that it's just Salesforce.
不是什么无头模式。
Not a headless.
对，没错。
Like Yeah.
它会进入等待状态。
It will go to Wait.
等等。
Wait.
等等。
Wait.
等等。
Wait.
不过说清楚点，你是说智能体真的会去打开浏览器吗？
Actually, just to clarify, do you literally mean the agent goes to the browser?
对。
Yes.
哦，不是的。
Oh, no.
不好意思。
I'm sorry.
我完全站在另一边。
I'm taking the other side on that one big time.
是的。
Yeah.
不是。
No.
对。
Yeah.
但让我把这事儿说简单点，咱们才能好好聊。
But but let me just let let let me let me simplify the argument so we could actually have it.
比如说现在，如果你用 NanoClaw 或 OpenClaw 这种代理，是可以用无头浏览器的。
So so today, if you use an agent like NanoClaw or OpenClaw, you could use a headless browser.
假设我想查一下我家房子在 Zillow 上值多少钱。
Let's say I wanted to, like, look up the value of my house on Zillow.
但无头浏览器根本行不通，因为 Zillow 被人爬得太狠了，一眼就能识破无头浏览器。
The headless browser simply doesn't work because Zillow's so tired of people scraping it, so it will detect headless browsers.
没错。
Totally.
所以管用的办法是，你直接打开Safari，让它用真正的Safari浏览器。
So the thing that works is it you as you pop up Safari and it uses a proper Safari directly.
对吧？
Right?
这样一来，问题就解决了。
And so then then all of a sudden it works.
所以呢，
And so
不对。
no.
但我认为，只要是软件有好用的API，智能代理肯定优先用API，只有当遇到执行障碍的时候，才临时切到浏览器里操作。
But but but but I think I think I mean, I would just say that that that any software that has a good API, the agent would absolutely prefer to use the API, and then you and then you pop into the browser the moment that you run into some execution problem
系统出了问题的时候。
with the system.
你知道吗，Seth 是个特别厉害的资深计算机科学和软件专家。
You know, Seth Seth is a a fantastic long term computer science software guy.
不过这些模型训练用的数据和强化学习环境，都来自那些根本没有这些 API 的旧软件。
However, these models are trained on data and RL environments from existing software that didn't have those APIs.
是的。
Yeah.
而且现在，如果你真去看看这些智能体的实际使用情况，它们的行为方式更像人类，而不是传统程序。
And and and right now, if you actually look at the adoption and the use of these agents, they look far more like what a human would do than what, like, a program would do.
也许你说得对，但现实情况根本不是这样，而且从数据和互联网上所有现有控制机制的角度来看——就像史蒂文说的——这些智能体最终会和人类采取完全一样的操作。
So maybe you're right, but, a, that's not what we're seeing, and you can honestly make the end to end argument when it comes to data and all of the controls in the Internet, to Steven's point, all the existing controls, to just be like, these are gonna actually have the same actions as humans.
其实大多数 API 的访问权限，都跟用户本身的权限是一样的，这个道理很明显。
Well, the APIs most that that I mean, I think the I mean, the APIs of any software provider will follow the same access controls of of whatever the whatever the user is that that
没错。
is Right.
但它们
But they
得重新构建，我的意思是，他们得从头再来一遍。
have to rebuild mean, they have to re they have to rebuild it.
这就像你有个现成的应用，所有模型都训练自用户平时的使用方式，但……
I mean, it's like it's like you've got this existing app and all the models trained on all of the people using Well, the
说到这点，确实很有道理，但我猜随着时间推移，我们会逐渐积累出非常精准、严谨的数据集，专门用来训练模型去对接每个SaaS平台的MCP和API。
on that point, and it's a totally fair point, but I would guess over time, you're gonna have very you know, kind of accurate, rigorous datasets for, you know, for models to be trained against the MCPs of of every SaaS platform, the APIs of every SaaS platform.
现在其实已经在做了，他们已经在用我们产品文档和API说明来训练模型了。
Already, it's in the you know, they're already training against all of our documentation on our on our products and our our APIs.
但对我来说，这更多只是个效率问题——与其在界面上点来点去，不如直接加个接口。
But but I I just think to me, it's it's more of just an inefficiency of of navigating through pixels versus just, you know, you you can just you can just add
快点搞定。
it quick.
是的。
Yeah.
还有，艾伦，加了智能代理系统之后，那些中间层永远不会消失。
And and add agent systems, Aaron, is that is that layers never go away.
它们只是层层叠加而已。
They just get layered.
呃，这个嘛，这个嘛，这个嘛……
Well well well well
我支持你的观点。
well I'll support your point.
百分之五十。
50%.
哎呀，这不就明白了嘛。
Oh, well, there you go.
是的。
Yeah.
对。
Yeah.
没错。
Yeah.
我只是觉得，如果你要找文档，用我们的搜索API比一点一点点进去快多了。
I I just I just think I just think if you need to do a search for a document, our search API is gonna be a faster way to do it than, you know, you know, clicking through a 100%.
对。
Yeah.
没错。
Right.
但是
But
不过呢，为了支持一下这个观点，现在的新型编码器，电脑桌面用起来简直太夸张了。
but but the but to to half support the point, like, the new codecs, the computer use that on the desktop is is, you know, just insane.
而且史蒂文显然对这一切怎么运作了如指掌。
And I I mean, Steven obviously knows, you know, everything about how it would work.
我看到自己鼠标一点，那边又有个鼠标自己动、自己点，我真觉得我完全不懂电脑了。
I when I saw my ability to move a mouse and then this other sort of mouse moving and clicking things, I was like, I don't understand computers anymore.
对。
Right.
对。
Right.
所以，马丁，你说得对，我第一反应是用它来处理那些根本没有公开 API 的任务。
So so there is a pretty and and and to your point, Martin, my first instinct was to use it for something where I know there's no available API.
所以我确实马上用它来做了个我没法调用 API 的事儿。
So I did actually use it right away for a thing that I don't have access to the API.
随着时间推移，这个智能体得自己判断：有没有现成的 MCP 或者命令行工具能干这事儿？
And an agent over time is gonna probably have to figure out, is there an easy MCP or CLI for this action?
如果没有，我就得启动个云浏览器、云电脑，或者本地工具，平行跑一遍，然后去执行操作。
And if not, then I'm going to pop into some kind of cloud browser or cloud computer or maybe local thing that I can, you know, sort of parallel track and then and then go and execute that.
这么搞确实挺合理的。
So that does seem like a reasonable architecture.
但我还是觉得，直接用无头模式狂刷 Salesforce 的 API 才是最高效的，至少查数据是这样。
But I still think that I'm to pound the Salesforce API massively in headless mode just because that'll be an efficient way to go look
对。
Yeah.
没错。
Up Yeah.
我的意思是，你们俩其实说的是一回事，只是时间点不一样而已。
I mean, I think that you're sort of I think you're both saying the same thing, but there's just a time to but no, but there's a time dimension.
我记得互联网早期有那么个阶段，你们说到时间差的时候，我突然就想起来了——当年我们写了成千上万页的 Word 和 Excel 使用指南，要么打包在产品里，要么让用户存到硬盘里，完全离线。
I think like there was a moment at the Internet that I really was thinking about when you guys were when I was seeing the timescale difference, which was suddenly the 8,000,000 katrillion pages of how to use word in Excel that we had written over the years that we posted on the Internet, we had used to ship them with the product and people would have them on their hard drive, not connected to anything.
人家问：‘怎么弄个漂亮的图表？’
And they would say like, How do I make a nice chart?
或者别的啥问题，结果根本找不到答案。
Or whatever, and it never worked.
他们怎么也找不到自己需要的东西。
They could never find the thing that they wanted.
但互联网带来的结果是，因为大家都来找，我们不得不改进文档，同时搜索引擎也变得更懂怎么找信息，这彻底改变了我们做文档的方式。
But what happened with the Internet was the net result of everybody finding it caused us to make better documentation, but it also caused Google Search to be better at finding the information that it needed, which then completely changed the way that we thought about doing documentation.
我觉得，对于这种专门用来找东西的无头模式来说，它也会彻底改变信息的呈现方式。
And I think that with headless, especially for the kind that's just finding things, it's going to really change the way that information is exposed.
所以，Salesforce 现在展示无头 API 的方式，我敢打赌，如果你去查一下，它看起来就是个前端开发 API，裹着个命令行工具，跟那玩意儿差不离。
And so the way that Salesforce sees today of exposing a headless API is I'm almost certain if I were to go look at it, it's going to look like the developer API in front of it, behind a CLI, and it's going to look a lot like that.
但事实上，普通人用 Salesforce 的方式根本不是这样——比如你站在电梯里，等着去见客户，你真正需要知道的是啥？
But in fact, that's not at all how humans using Salesforce interact as a human trying to solve, like, I'm standing in the elevator waiting to go see a customer, what is the stuff I need to know?
这两套逻辑完全是两码事。
Like, that that mapping is completely different.
所以，这个 API 会随着时间慢慢彻底改变。
And so the that API is gonna really change as a result over time.
没错。
And Yes.
我觉得 API 肯定会变。
I I I think I think the API changes for sure.
我同意。
I I agree that.
但我觉得，跟人形机器人不一样——那种在物理世界里总会撞上各种物理限制的情况，在数字世界里根本不存在。
But I do think that unlike the humanoid, you know, kinda comparison where where sort of the physical world has interesting physics issues that that you eventually run into, the digital world doesn't.
所以啊，某天你的智能代理能同时并行运行五百次。
And so so at some point, your agent can run-in parallel, you know, 500 times.
我想做一份财富500强客户的市场地图，
And, like, I'm going into I wanna do a market map of of customers across the Fortune 500.
这个代理能一下子铺开干这活儿，我一个人在浏览器里可做不到。
That agent can can fan out and do that work in a way that I can't as a person in a browser.
没错。
Right.
所以说，某种程度上，智能代理能让你突破人类常规工作流程的限制。
So so so to some extent, agents get to let you, you know, sort of bend the laws of of normal, you know, human based workflows.
所以呢，我觉得API最终可能会演变，但不是朝着终端用户产品那个方向，而是更偏向于智能代理自身的任务流程——它到底想干点啥。
And and so then just like that's why And I think that means the APIs maybe eventually evolve, but not obviously in the direction of the end user product, but maybe more toward an agentic sort of set of workflows of what is that agent looking to do.
不过马丁，我觉得他肯定会插一句：等等，你这根本没说新东西。
Well, but Martin, I think would jump in and just say, wait, you didn't describe anything new.
你只是描述了个架构罢了。
You described an architectural no.
但你提到的是当今软件的一个架构问题——它的API和性能瓶颈取决于你能打多少字，这恰恰就是我想说的重点：我们的帮助系统是按一张CD能装多少内容来设计的，根本不知道用户真正想干什么，也没有任何上下文。
But you described an architectural problem with today's software, which is it's API and performance gate was based on how much I can type, which which is sort of the point I was making, which was our help system was designed on how much we could ship on one CD and had no data about what it is that people were trying to do and no context.
问题本身没变，我还是得
It didn't change, like, the problem, which is I needed
做个图表。
to make a chart.
对。
Yes.
没错。
Yeah.
正是如此。
Exactly.
举个实际例子，我们推出过一个Box智能代理，功能强得多，其中一个能力是能扫描你整个Box环境里的内容，但它不像人类搜索那样，只能输一次关键词，等一堆结果出来，再一个个翻。
The well, so, like, one one real example of this, we we launched a Box agent that gives you know, that has, you know, much more capabilities built into it, and one of the capabilities is it searches across your whole Box environment, but it doesn't it doesn't have the same limitations of a human based search where you you type in one query, you get back a set of results, you look through them.
它会同时发起多个查询，自动展开搜索。
You know, it fans out, does multiple queries.
它能瞬间浏览数百个结果，并自行重新排序。
It can look through hundreds of results instantly and do its own re ranking of that.
所以你看，你肯定不希望被人类那种操作方式拖慢节奏——就像 humanoid 机器人，你大概觉得它还是得走进电梯。
And so that's just like, you know, again, you wouldn't want to be rate limited by the same process that a human went through, which is which is where the humanoid robot is, you know, you're kind of willing to be like, okay, the humanoid is still going to walk into the elevator.
它还是得去按按钮，但其实，在智能代理的世界里，你根本不会这么想。
It's still going to press the button when actually, you know, in age agent world, you're like, no.
不会。
No.
我就希望你直接去按我要去的楼层。
I just want you to go and instantly press the the the floor that I'm going to.
对。
Yeah.
没错。
Yeah.
但咱们得说清楚一点。
But we should just be very clear.
顺便说一句，我非常同意。
Like, by the way, I I very much agree.
但我们需要分清楚，会不会有人专门建一个只给AI用、不给人类用的索引系统？
But we need to make a distinction between, like, would you ever build an indexing that's only for AI and not for a human?
我觉得这一点就没那么明显了。
And I think that's less obvious.
对。
Yeah.
很明显，自动化能带来性能上的提升。
So clearly, there's, like, performance gains based on automation.
咱们的架构得跟上这些需求。
We've gotta involve our architectures for those.
但如果你找到了一种特别棒的文档索引方式，却不让人类用，我觉得——
But if you find a great way to index documents and you don't expose it to a human, I think Yeah.
没错。
You got it.
是的。
Yeah.
没错。
Exactly.
完全正确。
A 100%.
其实呢，这恰恰印证了史蒂文关于文档的那个互联网类比。
Well, this is I mean, it's actually and I think this kind of probably reinforces some of Steven's kind of Internet analogy on documentation.
我们在开发下一组智能体的时候，一开始是给了它们现有的工具集。
There is this really interesting thing where it started out where we as we've been building our next set of agents, we first gave it the current set of tools.
然后观察它们是怎么用这些工具的。
We saw how it used those.
后来我们才发现，其实还有更好的方法，智能体可以做得更高效。
And then eventually we realized, oh, there's actually an even better way that the agent could do it.
于是我们就优化了底层的架构。
So we improve the underlying scaffolding.
没错。
Right.
对了，这其实也会让最终用户受益。
And oh, by the way, that will actually help the end user also.
这样你就能反过来推动整个技术平台的升级，真正提升所有用户的体验。
So it does let you sort of contribute back into the mothership of of techno of technology improvement that does, you know, sort of lift all the boats of of your users.
我有个问题想问一下。
Let me let me ask this.
你刚才说的时候，我突然想到一件事。
It occurred to me as you were saying it.
一想到这个，我就紧张了：现在我们每天有上万人在用我们的SaaS系统，一切运行得挺好。
Like, sort of got all tense when the idea became, no, like, oh, we have 10,000 people hitting our SaaS system today, and we've got it all working, and it's all great.
但马上又要新增一万个新用户——也就是每个员工对应的智能代理，它们对系统的调用频率会飙升500倍。
But now we're going to have 10,000 new people, which are the agents for each of those 10,000 employees, and they're they're actually hitting it 500 times as much.
那我们的SaaS系统肯定得崩。
Okay, so that SaaS product will collapse.
所以这首先是架构问题，根本没按这么大的量来设计。
So like, that's the first order, because it wasn't architected for that volume.
我们以前在所有商业智能工具上都见过这情况。
Like, we saw this with all the BI tools.
当初BI工具刚出来的时候，大家突然开始盯着SAP的数据，每晚都要全量快照，重新切片分析，搞出五百种不同的视角。
Like when all the BI tools came out, all of a sudden they were looking at the SAP data and trying to snapshot it and absorb the whole thing every night for a new kind of set of slices and dice that your view across all 500.
结果做ERP的人就说：我们可不干这事儿。
And like all the people making ERP were like, Well, we don't do that.
他们只好自己从头搭一套，毕竟只有他们最懂这些数据。
And so they had to go build all of this themselves because they had the knowledge of the data.
他们的API压根不是为这种查询设计的，根本扛不住。
Their API just couldn't was not designed for that kind of lookup.
所以我现在想抛出来讨论的是：到底该咋办？
So my my sort of thing to throw out there and fight about is what is it?
公司里头，这种变革到底该怎么管理？
What does the change management look like in a company?
你总不能放一个代理，让它对系统发起的请求是人类的500倍吧。
Because you can't let loose an agent that hits the system at 500x the humans.
这可不是什么令牌问题。
And it's not a token thing.
而是真真切切地发现：我们的网络带宽和吞吐量，根本扛不住任何一个客户500倍的负载。
It's an actual like, wow, we don't have the network bandwidth and the the throughput to handle 500 x for any one of our customers.
那接下来怎么办？
So what happens?
我这儿有个有点激进的观点，你们帮我看看是不是跑题跑太远了。
So I've so I've got a provocative adjacency, which you guys can tell me if I'm doing too much on a tangent here.
但我的观点是：多几个代理，真不一定算得上是架构上的大变革。
But but here's my provocative adjacency, which is I don't know if having more agents is that big of an architectural shift.
我觉得吧，这事儿其实我们都懂。
I just feel like we understand, like, whatever.
如果是只读数据，直接缓存就行了。
If it's read only data, you cache it.
你懂的。
You know?
说白了，所有状态问题都出在可变的、全局共享的状态上。
Like, all the state issues are around mutable mutable globally shared state.
我们早就清楚这些限制了。
We understand the limits of those.
我们都知道怎么设计系统来避开这些问题——当年我们搞互联网的时候不就解决过一箩筐吗？
We know how to architect around We had to tackle all of those things when we went to the Internet.
如果你连这都搞不定，那只能说明你系统设计太烂了，活该崩，不如去搞个不烂的系统。
And so if you built your system out to handle it, like you suck at building the system, you deserve to go down and just go build a system that doesn't suck.
我觉得这其实就是计算机科学里的基本常识。
And, like, I just feel like this is kind of standard computer science.
不过呢，我觉得智能体确实给企业带来了一些技术上必须面对的新问题。
However, I do think agents do introduce something that organizations technically have to deal with.
我打个代码上的比方吧，就是——
And I'll let me just give the analogy in code, which is I
我觉得，史蒂文，这其实就是我们说的‘蹭流量’
think, Steven, this is what we call mugging on
不，不是这样的。
I No.
不对。
No.
根本不是。
No.
完全不对。
No.
这其实挺棒的，我刚才只是在调侃一下。
It's a great I I mocked.
我只是觉得这事儿本身存疑。
I just it was in question
我不知道他刚才干了啥，但我特别期待他是怎么神乎其技地把问题给解决了。
I have no idea what he just did, but I I'm I'm just looking forward to how he magically made the problem go away.
那你继续说吧。
But go ahead.
不行。
No.
不行。
No.
不行。
No.
不行。
No.
不行。
No.
不行。
No.
问题就摆在那里。
The problem is there.
我只是觉得，咱们都知道怎么从10个用户做到
I just think, like, we know how to go from 10 users to
一千个用户。
a thousand users.
其实就是把那些不靠谱的家伙全筛掉了。
We just got rid of the stupid people.
现在剩下的全是聪明人。
So now everybody is smart.
不对。
No.
不对。
No.
行吧。
Okay.
那我给你举个编程方面的例子。
So so let me give you an example for coding.
所以我觉得，这里其实标志着工作方式发生了一个转变。
So this is where I actually think there's a shift in how work gets done.
当你用AI写代码时，代码质量会越来越差，而且是明显地变差。
So when you code with AI, your code kinda gets worse over time pretty materially.
结果你每解决一个问题，就顺便引入了两个新问题。
And so it's almost like you're introducing as many problems as you are solutions.
我觉得我们到现在还没搞清楚该怎么应对这种情况。
And I don't think we've actually figured out how to manage that.
这么说你能明白吗？
Does this make sense?
现在整个世界都是这样。
It's in the whole world right now.
对。
Yeah.
我意思是，你知道的，
I I you know,
我的意思是，这个问题其实挺合理的：你用AI确实效率高了，但你是不是制造的问题比解决的还多？
I mean, like, this this kind of reasonable question is, you know, if you're using AI, yes, you're productive, but are you creating more problems than you've actually solved for solutions?
我觉得现在有个很关键的问题还没搞清楚，就是用智能体去操作现有系统做开发时，我们到底有没有办法控制住这种不断膨胀的混乱？
And I do think that there's this actually, you know, open question when it comes to using agents on existing systems for creating things, which is like like, do we know how to wrap the growing set of entropy around that?
从我身边看，很多公司用AI写代码都挺吃力的——当然，我跟不少AI编程公司关系都很近。
And I would say anecdotally, watching companies struggle with AI coding, which of course, listen, I'm very close to many AI coding companies.
我本人对这事儿特别看好。
I'm clearly very bullish on it.
但说实话，我们目前还不知道该怎么妥善处理这个问题。
I don't think we know how to do that yet.
至于在系统里用智能体做短期任务，我觉得用现有方法还能应付。
And so the agents on a a system, I think we can tackle those with known techniques.
但要是用智能体搞长期、跨组织的复杂项目，就像系统刚上线三天后，整个环境早就乱成一锅粥了——
Using agents for long running things organizationally where, like, you know, you know, the the universe is kind of as clean as it was three days after the new started.
这个我真的不确定我们有没有办法搞定。
I'm not actually quite sure we know how to
根本做不到。
do that at all.
我特别喜欢这一点，因为它回到了我们一开始讨论的核心：规模化和非规模化之间的区别。
Well, I I love that point because that gets back to where we started, which is the difference between scale and not scale.
没错。
Yes.
难怪大公司的人都觉得：这玩意儿绝对不可能进我们公司。
And why it's perfectly rational for big company people to be like, No frickin' way is this coming into our company.
因为大公司里，任何一个部门、任何一个产品，随时都可能崩盘。
Because a big company is about to the wheels are gonna come off a big company or a division in a big company or a product in a big company at any minute.
比如你我，都当过大公司的高管。
Like if you're a Marti, we were both giant company executives.
说实话，我们每天早上醒来第一件事就是担心：今天会不会出大事？
Like literally, we woke up every morning thinking, Oh, the wheels are coming off today.
这事儿怕是要完了。
This is the end of it.
我每天五点前都觉得自己要被炒了，本来以为项目已经晚了三个月，结果现在拖了九个月，这日子太正常了。
I'm getting fired by the 05:00, and whatever started, I left yesterday thinking we were three months late, and we're now nine months late, and that's a typical day.
但之所以没出事，是因为我们到处设了限制，没错。
But the reason that that doesn't happen is because you put constraints all over the place, Exactly.
这正好解释了为什么吉尔福尔没法在大公司待下去——他觉得自己啥都懂；也说明那些靠灵光一现、凭感觉写代码的人为啥觉得‘没问题’，因为他们从来没经历过那种约束：必须防止整个系统彻底崩盘。
Which is exactly why Guilfoyle can't work at a big company, because he thinks he knows, and it's also why all the one shotting, vibe coding kind of people have no problem saying it's fine because they've never had to live in an environment where the constraint was to prevent the whole thing from imploding.
我觉得这一点特别关键，史蒂夫。听起来可能有点跑题，但它跟这个直接相关：我觉得底层技术其实是在满足一些人类需求。
I feel this is so critical, Steve, that you're like So again, this is gonna sound like a little tangential, but it feeds into this, which is I feel like core technologies kind of cater to, like, a some human needs.
比如互联网满足了连接的需求，社交网络满足了虚荣心。
Like, the Internet catered to, like, connectivity, and social networking kinda catered to vanity.
而我觉得，AI满足的是我们对高效生产的渴望。
And I feel like AI caters to our need to be productive.
所以我们总觉得干了这么多事就是在高效工作，但其实可能只是堆了一大堆额外的活儿出来。
So I feel like we feel like we're being very productive when we do all of these things, but we may actually be creating, like, mounds of extra work to do.
而且呃，嗯，
And Well, well,
亚伦，你现在正在部署AI吧。
Aaron, you're deploying AI right now.
比如Box，已经全面投入了。
Like, Box is all in.
给我们讲讲吧，是出问题了，还是挺顺利的？
So tell us tell share a story, like, of the wheels coming off or not coming off.
是的。
Yeah.
嗯，我觉得我们可能处在更务实的那一端。
Well well, I I think I think we're probably in the more pragmatic part of the continuum.
所以我们不会说AI让工程团队的生产力提升了十倍。
So so which is why we don't claim that it's a 10 x productivity gain to our engineering team.
不会，因为我们系统里设置了大量防护机制，自动施加了各种限制。
It's like, no, no, because we have a lot of guardrails in place that create these constraints automatically in our system.
我们依然非常依赖代码审查。
There's we we still rely heavily on on code reviews.
我们还是很依赖安全审查。
We still rely heavily on security reviews.
所以你们 coding 的时候，是拿石头和凿子敲吗？
So are you are you guys coding with, like, a rock and a chisel and stuff?
听起来好像我们还在用黑板似的，但其实不是。
Know it just sounds like We have chalkboards and, like but no.
我们刚上线了一个新功能，我当时就说了：上吧。
But, like, I I we had this new feature that we that we launched, and I was like, go.
上吧。
Go.
上吧。
Go.
上吧。
Go.
上吧。
Go.
这个功能的80%到90%都是AI帮着写的。
And AI built probably 80 to 90% of the feature.
真正拖慢发布进度的，是我们得走一遍完整的安全审查，毕竟不能让咱们开发的东西不小心被注入恶意代码。
And the the thing that slowed down the release of it was we have to do a full security review because we can't let there be any, you know, accidental code injection into the thing that we created.
很多地方你确实能冲得飞快，但总被流程里别的环节卡住，限制住了。
So so there's a lot of stuff where you kind of go super fast, but then you get still rate limited or constrained by some other part of the process.
我觉得这挺正常的，直到我们真正解决这些瓶颈——比如安全审查、代码审查，或者上线流程本身这些环节。
And I I think that's sort of, you know, relatively natural until we figure out then that other part of the process, security review being one or the actual code review being one or just even your pipeline for for, you know, getting things into production being another one.
我们现在正在大改整个产品开发的流程，但我觉得这不会带来五到十倍的提升。
So we we're doing quite a little you know, quite a quite a bit of retooling of the whole product development life cycle, but I I don't think that it's a five to 10 x gain.
我觉得整体上能提升两到三倍就不错了。
I do think it's a two to three x gain maybe across the board.
你还是被审查和检查的速度卡着，再快也快不起来。
You are still rate limited by how quickly can you review this stuff and and check on the work.
不过马丁提到的这点，确实是现在企业里最大的一个待解问题。
I do think that that Martin's sort of pointing at, though, a thing that is the the big open topic across enterprises.
某种程度上，工程师会最先遇到这个问题，我们会慢慢找到平衡点。
And and, you know, to some extent, and engineers will will face it first, and and we'll find the right equilibrium.
真正的难点，还是在其他知识型工作上。
The harder part still remains in the rest of knowledge work.
是的。
Yeah.
所以如果你做会计，现在还说不准什么时候能完全放手，让AI去搞完整个审计——毕竟这事儿还没法全交给机器。
This is why if you're in accounting, you know, we don't quite yet know when you can take your hands off the wheel, you know, doing a full accounting audit, you know, because of AI.
但你可以让AI去翻海量数据，找出异常点，提醒会计团队：嘿，这个得深入查一查。
What what you can do is have the AI go and and, like, comb through unlimited amounts of data to find anomalies that maybe are that would alert your accounting team to, oh, we actually have to go dig into this.
这太棒了，因为这相当于新增了一层洞察力，而不是取代那些必须逐笔核对、确保每个数字都精准的繁琐工作。
That's awesome because that's only net new level of visibility versus the part of the accounting process where you're doing a fine tooth comb on making sure every single number is is accurate.
那种精细核对，现在还是得靠人。
That's probably still humans right now.
所以关键在于，你到底在哪儿能真正提升效率？
So I think the key is where do you find the productivity gains?
我觉得，如果你是CEO、董事会或者管理层，你其实正在努力搞清楚现状，但又被硅谷那边各种说法搞得一头雾水。
And I do think that if you are a CEO or a board of directors or a management team, you're kind of trying to figure out and you're also getting confused because Silicon Valley is telling you all the things.
所以你得琢磨清楚，哪儿的生产力提升最明显，哪儿能真正拿到收益，同时把风险降到最低。
And so you have to sort of figure out where is where is the productivity most potent, where I actually can get the gain, I can get the success with less of the downside.
我觉得整个行业现在都在摸索着往前走。
And I think as an industry, we're all sort of figuring it out.
不过顺便说一句，正因如此，我对就业前景依然超级乐观——我觉得大家对AI取代人类的判断完全搞反了，你还是得有人在流程里兜底，这点跑不掉。
But by the way, this is actually why I remain unbelievably optimistic on jobs because I don't think you like, I just think we've gotten it wrong on on thinking, you know, all the places where you're going to remove humans from this because you still need a human in that, you know, somewhere in the loop.
也许现在人的角色更宏观了，不像一年前那样，每个环节都得亲力亲为。
Maybe the abstraction is a little bit higher, and you don't need the human in the loop at every at every single stage that you needed a year ago.
但你还是得有人来启动流程、复核结果、把AI生成的东西整合进去。
But but you do need a human sort of kicking off the process, reviewing the process, and incorporating whatever the work was.
这样一来，这些组织里依然有大量机会和岗位留下来。
And so that creates just still a tremendous amount of opportunity and jobs across these organizations.
哎，我得插一句，我今天带了一大堆视觉辅助材料，就想让这话题更生动点。
Oh, let me I have to jump in because I I have I have a whole bunch of, like, visual aids I brought today to make it exciting.
关于MTS直播那件事，好多人都留言支持你，我可不能就这么放过这个机会，毕竟我们平时老抱怨没人同意你。
We got you got a bunch of comments on the MTS Live thing about people agreeing with you, so I don't want to let that slide because we complain about not agreeing with you.
比如这儿——
Like here-
我完全同意。
I totally agree.
你提到的这个，是八十年代一本叫《工作的终结》的书。
To your point, this was a book in the '80s called The End of Work.
对。
Yep.
其实不好意思，应该是九十年代出的。
I so actually, sorry, it was in the '90s.
那本书出版的时候，正好赶在互联网爆发前六个月，书里说技术革命根本没用，生产力一点没提升，以后也不会有新工作了，因为经济已经停滞了。
It came out like six months before the internet hit, and the whole thesis was the technology revolution was a complete bust, and we got no gains in productivity, but now there's going to be no more jobs because the economy is stagnant.
作者自称是个未来学家。
And this was a guy, he called himself a futurist.
所以啊，这个AI时代特别有意思的一点是，很多事你第一次听说的时候觉得荒唐，可回头一想，发现它其实更荒唐。
And so the whole notion that One of the neat things about this whole AI moment is the number of things that when you hear them the first time, you think they're stupid, and then you go back and think about it, you're like, Oh my God, it's way stupider.
说AI会直接消灭工作，这想法老得跟你说会计一样古老了。
And this idea that AI just gets rid of jobs, it's as ancient as like you talk about the accountant.
以前大家就以为电脑一出来，会计就得失业。
Like, one of the things people thought was that computers would get rid of accountants.
对啊。
Yes.
这可是IBM在1965年主打的宣传点。
And that was like IBM's pitch in like 1965.
但结果呢？天啊，完全不是那么回事。
But what it actually did was like, oh my god.
现在会计的工作内容根本不是整天加加减减了，反而能干更多高级活儿。
We could do so much more with accounting now that they're not, like, literally just adding numbers all day.
没错。
Yes.
我觉得，当你看生成信息、整合信息这些事的时候，AI其实是个加速器。
And I think when you look at, like, just the notion of, like, creating information, synthesizing and all that, like A.
对。
I.
它能帮那些懂行的人干得更快，公司现在会更想要这种能创造信息的人。
Is is an accelerant for that, for a person who knows what they're doing, and companies are suddenly going to want more of those people creating more of that information.
更别说了，如果AI在源源不断地产出有价值的信息，
Not to mention the fact that if A.
对。
I.
那肯定会有更多人需要去消费这些信息，用来做点什么。
Is creating valuable information and there's more of it, then more people will need to consume it to do something.
公司的本质，不就是根据信息来做决策吗？
And the idea the essence of a company is acting on information.
所以说什么信息会泛滥、用不完，这根本说不通。
And so this idea that information is just going to get produced easily and be in surplus and not used makes no sense at all.
因为你知道，在非结构化信息的世界里，问题不在于能不能生成，而在于怎么有效利用——这才是现在的关键瓶颈。
Because as you know, like in the unstructured information world, the problem is that you can make it, but the consumption of it effectively is the And guiding that's the gating factor now.
我跟我们审计委员会的主席聊过，他是个注册会计师，他说自己刚入行那会儿，整个流程简直太原始了，我都懒得复述，真没法想象没这些现代技术之前世界是怎么运转的。他跟我讲会计师的工作流程，我就说：对啊。
I had a conversation with one of our board members who's chair of our audit committee, so he's a CPA, and he was telling us kind of early in his career, I can't even retell it because it felt so manual, so I don't I don't even know how the world worked I don't know how the world worked before all of the modern technology, but he was explaining the CPA's process, and I was like, yeah.
那简直是我见过最手工、最繁琐的事了。
It it seemed like the most manual thing of all time.
不过，史蒂文，我觉得这跟你书里说的完全一致——那时候能做的事其实特别少，流程特别简单。
But but and and, Steven, I think this is right out of your book is, like, it was actually quite simple in in sort of the amount of things you could do.
是啊。
Yeah.
因为那时候整个系统还没数字化，全靠手工操作。
Because of of how undigitized and relatively manual the whole thing was.
结果计算机一进来，反而让流程变得更复杂、更全面，也因此催生了更多岗位，就是因为这种人为增加的复杂性。
And computers actually only made it more complicated, more comprehensive, and thus created even more jobs because because of that complexity that that we introduced.
你就能看到，这种现象在很多工作领域都一模一样——我们现在有能力，也愿意把事情搞得更复杂了。
And and you can just sort of see how easy this is to show up in so many areas of work is like, we can just now we can afford to make things more complex.
所以如果你把事情搞得越来越复杂，最终还是会碰到新的瓶颈——谁能理解得了这种复杂性？
And so if if you make things more complex, then actually you eventually still run into now new constraints of who can understand that complexity.
说真的，我觉得最搞笑的就是那种说法，以为写代码越多，就越不需要工程师，其实完全相反。现在系统的复杂度比以前高多了，一旦要升级系统、出故障了，或者遇到安全事件，你根本不知道怎么下手解决，挑战只会越来越多。
And and and and so, like, you know, it's like to me, it's like the funniest concept that the more code we write, the less we would need engineers would be the opposite because because now your systems are even more complex than before, which means that you're gonna be running into even more challenges of when you need to do a system upgrade or when there's downtime and you have to figure out, like, what well, how do I fix that problem or when there's a security incident?
所以没错。
And so yeah.
我的意思是，这方面的就业机会，我们才刚刚开始呢。
I mean, this is this is this is like we're just getting started with the jobs on this front.
没错。
Right.
而且你听好了。
And all listen.
我们其实已经进入这个阶段好几年了，你甚至可以去看看相关数据。
We're we're a few years actually into this, and you can actually look a bit at the data too.
对吧？
Right?
那现在哪些公司招人招得最猛？
Like, what are the companies that are hiring the fastest?
就是那些原生AI公司，疯狂招人。
Like the AI native companies, and they're hiring like crazy.
而且别忘了，以前还有一种说法，说AI写代码会把基础设施给干掉。
But not only that, I remember there was this early prognostication which is AI writing code will get rid of infrastructure.
说基础设施会变得像商品一样廉价，可这说法真挺奇怪的，毕竟现在写的代码比以往任何时候都多。
It's gonna commoditize infrastructure, which is this kind of very strange prediction given the fact that there's more software than ever before been written.
我本人在好几家基础设施公司当董事，有些之前一直没起色，现在全都在爆发，就是因为软件太多了。
And sitting on the board of a bunch of infrastructure companies, some that have been flat for a while, they're all doing fantastic because there's so much software.
现在市面上的软件数量简直翻了天。
There's so much more software out there now.
所以你听好了。
And so listen.
要是真看企业那边的实际数据，结论就是：软件越来越多。
If you look at the data on the ground from the companies, it's more software.
AI原生公司招人招得最猛，这让我很清楚，我们现在正处于扩张期。
The AI native companies are hiring the the the most, and so it's very clear to me that we're in an expansion phase.
至少关于这一点，我想最后补充一点：硅谷的人往往眼光太窄，总觉得工程师就是去谷歌、或者你随便说个科技公司、创业公司上班，这才叫工程师工作。
And maybe just my my only final point on this one at least is I think people we have a little bit of a myopic view in Silicon Valley on thinking that engineering jobs are you go to work at Google or name your tech company and start up, and that's an engineering job.
我们都被这种思维惯住了，毕竟我们就在这个圈子里嘛。
We're so wired into that because that's obviously the ecosystem that we're all part of.
可你别忘了，约翰迪尔在搞自动驾驶拖拉机，卡特彼勒在搞AI系统，礼来制药在拼命研发新药。
And then you sort of forget, well, like John Deere is trying to make automated tractors and Caterpillar is trying to have AI systems and Eli Lilly is trying to design even more pharmaceutical therapeutics.
这种公司，你随便数都能数出五千家。
You can go through 5,000 other companies.
接下来，这些公司会涌现出新一代工程师，他们用Cloud Code、Codex、Cursor这些工具，自动化自己的业务，为自己的工作流程和系统开发更多软件。
They're going to now have the next set of engineers that are going to use Cloud Code and Codex and Cursor to be able to automate even more of their businesses and be able to design and develop even more software for their workflows and their systems.
你不一定非得去搞社交网络，优化什么推荐算法。
And so it just might be that you don't go and work on a social network and improve the social network algorithm.
你也可以去约翰迪尔，优化那个智能农业的算法。
You go work at John Deere and you improve the, you know, intelligent farming algorithm.
我们就得这么想，对吧？
And and that and and we just have to you know?
我的意思是，这其实就像马克·安德森十五年前说的那样——软件会吞噬世界。
I mean, this is sort of like like completely you know, like Marc Andreessen predicted this, you know, fifteen years ago of, like, software is gonna eat the world.
这意味着每个人都会用上大量软件，而现在大家终于有了能力去做到这一点。但你还是需要专家或者准专家来指导AI代理该做什么、检查它的成果、管理它搭建的系统。
And what that means, though, is that everybody's gonna have lots of software, And this gives everybody the ability to finally have lots of software, but you still need then an expert or semi expert to be actually going and prompting the the the, you know, the agent on what to do, reviewing its work, and managing the system that it builds.
所以那些说别去学编程、别搞软件工程的预测，我觉得全都会被打脸。
So so all of the, you know, predictions on don't go into coding and don't go into software engineering, I think will be proven quite quite wrong.
我觉得，我的意思是，那
I think I mean That
说得太到位了。
was so good.
说得真好，亚伦。
That was super good, Aaron.
我觉得这一切的核心在于，现在太多人讨厌科技，对世界抱有僵化的看法。
And I think that that the the the base case of all of this is just that there's too many people out there right now that don't like technology and have a static view of the world.
所以当他们想象AI会做什么时，一听到‘自动化’，就以为会夺走他们的工作。
So when they look to whatever it is that they think AI is going to do and people hear automation, they just assume it's going to take things away.
哦，原来如此。
Oh, I see.
哦，你是在看我的Mac摄像头啊。
Oh, you're looking at my Mac camera.
是的。
Yeah.
对。
Yeah.
怪不得呢。
Oh, that's why.
你挺讲究的嘛。
Oh, you're fancy.
你挺讲究的嘛。
You're fancy.
好的。
Okay.
所以呢。
So
这可是《时代》杂志啊。
this is like Time Magazine.
那时候每个高中生都看这个。
Every kid in high school read it.
1981年的事儿了。
1981.
但那时候大家觉得电脑就是用来把公司里的纸张流程自动化。
But the whole view of what computers would do would be they would automate the paper in a company.
所以第一代计算机的思路，就是把纸质表格直接搬到屏幕上，再打印出来，让一切变得更方便。
And so the idea, like the whole first generation of computing, was literally taking paper forms and turning them into something on a screen, then printing them out and then making it all easier.
再往后一看，就像你刚才说的，亚伦，以前还真有律师从来不打字的年代。
And you fast forward and it's all of these things that you just said, Aaron, like, you know, there was an era when lawyers didn't type.
所以当时他们雇了法律助理来处理这些事。
And so what happened was they just they they had people who were legal assistants.
他们叫这些人‘法律文员’，所有打字的工作都由他们来做。
They called them paralegals, and they did all the typing.
后来，一些哈佛的学生把一台电脑带进了教室。
And then like some students at Harvard, they brought a computer into the classroom.
所以我把它放低一点，
And so this is I'm lowering it so
你们才能看清楚。
you guys can see.
对，对，没错。
Yeah, yeah, yeah.
他们在八十年代初就把这台原始的笔记本电脑带进了教室。
So they brought that's an original laptop in there in the early 1980s.
他们把这台电脑带进教室，结果被赶了出去，但那时候上法学院，你得用手写论文，写在本子上，教授还得一篇一篇地读。
They brought this computer into the classroom, and then they got thrown out for using But they were literally they used to do law school, you'd write the essays in longhand in a book, and then the professor would have to read them.
现在当然都是打字了，还能直接查所有引用文献，但说实话，现在没人会找一个合同不开启‘修订模式’的律师。
And now, of course, you just type them and you have access to the database of all the citations, but that's exactly like nobody deals with a lawyer who isn't in track changes with your contract.
对。
Right.
我查过，现在律师的人数比三十年前多太多了，你随便找一个律师，都是靠电脑干活的。
Last I checked, there are way more lawyers today than there were thirty years ago, and they Every all are human lawyer you talk to is a computerized lawyer.
他们的引用都来自网络，法律简报里的资料也是上网找的，简报本身也是打出来的。
Their citations come from the internet, their information in the brief comes, and they type the brief.
我觉得我们又陷入那种狭隘的思维了。我本人超爱科技，真希望真是这样，但我觉得我们太想当然了，以为每个人的工作就是坐在微软Word里敲文档。
And I think we, you know, kind of going back to the myopic approach, I think we maybe over like, I mean, as a big lover of technology, I I wish this was true, but I think we just over assume that, like, everybody's job is is just they're just inside of Microsoft Word, and they're just typing a a Word document.
其实对律师来说，大部分时候我是在帮他们出策略，或者一起分析一个特别复杂的情况。
It's like like I mean, most of the time with lawyers, I'm, like, you know, strategizing something or or they're, like, working through a a complex analysis of a situation.
我当然可以去问AI建议，但那样反而更可能让我最后还是得打电话给律师，问：‘你觉得我们这事儿该怎么处理？’
And it's not like, I could go to an AI for for advice, but and but that would probably only increase the chance that I go and then call a lawyer to say, hey, what do you think about this situation that we're dealing with?
所以很多工作都不是简单地坐在电脑前把活全干了，背后有大量复杂的背景和判断。
And so a lot of these jobs just have a lot of context that aren't sitting just literally on the computer doing all the work.
他们确实得时不时接地气，这是工作的一部分。
They do have to kind of touch grass as a part of the job.
所以AI呢，确实能帮着自动化内容生成和信息审核，但最终还得落地到真实的生产价值里。
And so then AI, yeah, AI will help automate the creation and production of the content and the review of the information, but then it still has to be incorporated into the real world of real value production.
我觉得咱们现在是直播，本来该四点结束的。
I feel like we're live and we're supposed to end at four.
我打算直接说：咱们在直播，现在四点了。
What I'm going do is just say we're live and it's four.
那意思就是咱们直接收工，灯啊什么的慢慢暗下去吧。
And I guess that means we just stop and some lights fade or something.
咱们谁都没干过这事儿。
None of us have done this before.
根本没人知道接下来该干嘛。
We don't know what's supposed to happen.
有个人冲我挥手微笑，说：对，你说得对。
Someone is waving at me and smiling, saying, Yes, I think you're right.
笑的意思是别说了？
The smile means stop talking?
行吧。
Okay.
好嘞。
All right.
很高兴见到大家。
Well, it's great to see everybody.
大家再见！
Bye, everyone.
感谢收听本期的a16z播客。
Thanks for listening to this episode of the a16z Podcast.
如果你喜欢这期节目，别忘了点赞、评论、订阅，给个评分或写个评价，也分享给你的朋友和家人。
If you liked this episode, be sure to like, comment, subscribe, leave us a rating or a review, and share it with your friends and family.
想看更多节目，去YouTube、Apple Podcast和Spotify吧。
For more episodes, go to YouTube, Apple Podcast, and Spotify.
关注我们在X平台的账号@a16z，订阅我们的Substack频道：a16z.substack.com。
Follow us on X at a sixteen z, and subscribe to our Substack at a 16z.substack.com.
再次感谢收听，咱们下期节目见。
Thanks again for listening, and I'll see you in the next episode.
以上内容仅供教育参考，不构成任何投资或金融产品的买卖或持有建议。
This information is for educational purposes only and is not a recommendation to buy, hold, or sell any investment or financial product.
本播客由第三方制作，可能包含付费推广内容、其他公司提及以及与a16z无关的个人。
This podcast has been produced by a third party and may include paid promotional advertisements, other company references, and individuals unaffiliated with a16z.
这些广告、公司和个人均未经Ah Capital Management LLC、a16z或其任何关联方认可。
Such advertisements, companies, and individuals are not endorsed by Ah Capital Management LLC, a16z, or any of its affiliates.
所有信息均基于发布时的可靠来源，但a16z不保证其绝对准确。
Information is from sources deemed reliable on the date of publication, but a16z does not guarantee its accuracy.
关于 Bayt 播客
Bayt 提供中文+原文双语音频和字幕，帮助你打破语言障碍，轻松听懂全球优质播客。