克劳德代码负责人：编程问题解决之后会发生什么 | 鲍里斯·切尔尼

而且性能提升简直惊人。

And the gains have just been insane.

就像，在过去一年里，自从我们推出Clawd Code以来，我们大概……我不确定具体数字。

Like, adanthropic, over the last year, since we introduced quad code, we probably I don't know the exact number.

我们大概把工程团队规模扩大了四倍左右。

We probably like four x the engineering team or something like this.

但就拉取请求而言，每位工程师的生产力提升了200%。

But productivity per engineer has increased 200% in terms of pull requests.

对于任何真正在这个领域工作、致力于开发效率的人来说，这个数字简直令人难以置信。

This number is just crazy for anyone that actually works in the space and works on dev productivity.

因为在我之前的职业生涯中，我曾在Meta工作，我的职责之一就是负责公司的代码质量。

Because back in a previous life, was at Meta, and one of my responsibilities was code quality for the company.

这涵盖了我们的所有代码库。

This is all of our code bases.

这是我的职责范围，比如Facebook、Instagram、WhatsApp等等所有这些产品。

Was my responsibility, like Facebook, Instagram, WhatsApp, all this stuff.

这其中很大一部分都关乎生产力，因为如果你让代码质量更高，工程师的效率就会提升。

A lot of that was about productivity, because if you make the code higher quality, then engineers are more productive.

我们过去看到的情况是，在有数百名工程师参与的项目中，一年内生产力只能提升几个百分点，大概就是这样。

Things that we saw is in a year with hundreds of engineers working on it, you would see a gain of a few percentage points of productivity, something like this.

所以现在看到这种高达数百个百分点的提升，简直太疯狂了。

So nowadays seeing these gains of just hundreds of percentage points, it's just absolutely insane.

更疯狂的是，这一切竟然已经变得如此习以为常。

What's also insane is just how normalized this has all been.

我们当然听到这些说法，说是AI在改变我们的一切。

We hear these numbers, of course, AI is doing this to us.

软件开发、产品构建，乃至整个科技世界所发生的变革规模，是前所未有的。

It's just it's so unprecedented, the amount of change that is happening to software development, to building products, to just this the world of tech.

人们很容易对此习以为常，但重要的是要意识到，这真的太不可思议了。

It's just, like, so easy to get used to it, but it's important to recognize this is crazy.

我得时不时提醒自己这一点。

This is something I have to remind myself once in a while.

这背后其实有一些负面影响，我们本可以讨论很多问题，但我觉得其中一点个人层面的困扰是，模型变化得太快了，我有时会陷入旧有的思维方式。

There's sort of a downside of this because the model changes so There's actually many downsides that we could talk about, but I think one of them on a personal level is the model changes so often that I sometimes get stuck in this old way of thinking about it.

我甚至发现，团队里的新人，甚至是刚入职的毕业生，做事情的方式反而比我更偏向AGI导向。

And I even find that new people on the team or even new grads that join do stuff in a more like AGI forward way than I do.

所以，比如几个月前，我就遇到过一个内存泄漏的问题。

So, like, sometimes, for example, I I I had this case, like, a couple months ago where there was a memory leak.

这其实就是，你知道的，代码的内存使用量不断上升，最终导致程序崩溃。

And so, like, what this is is, you know, like quad code, the memory usage is going up, and at some point it crashes.

这是一种非常常见的工程问题，每个工程师都调试过成千上万次。

This is, a very common kind of engineering problem that every engineer has debugged a thousand times.

传统上，你的方式是获取一个堆快照，把它导入专门的调试器，然后慢慢分析发生了什么，使用这些专业工具来排查问题。

And traditionally, the way that you do it is you take a heap snapshot, you put it into a special debugger, you kinda figure out what's going on, you know, use these special tools to see what's happening.

我当时就在这么做，一边查看这些调用链，一边试图弄清楚问题出在哪里。

And I was doing this, and I was kinda like looking through these traces and trying to figure out what was going on.

而团队里那位新来的工程师，直接让Quad Co去处理了。

And the engineer that was newer on the team just had Quad Co do it.

然后他就说：嘿，Quad。

And it was like, hey, Quad.

好像有内存泄漏。

It seems like there's a leak.

你能找出来吗？

Can you figure it out?

于是Quad Co做了和我完全相同的事情。

And so like Quad Co did exactly the same thing that I was doing.

它获取了堆快照。

It it took the heap snapshot.

它为自己写了一个小工具，以便自己分析。

It wrote a little tool for itself so it can analyze it itself.

这是一个即时程序。

It was a just in time program.

它比我还快地找到了问题并提交了拉取请求。

And it found the issue and put up a pull request faster than I could.

所以对于那些长期使用这个模型的人来说，你仍然需要把自己拉回到当下，不要被困在旧的模型里，因为现在已经不是SONNET 3.5了。

So something where for those of us that have been using the model for a long time, you still have to transport yourself to the current moment and not get stuck back in an old model because it's not SONNET 3.5 anymore.

新模型完全、完全不一样了。

The new models are just completely, completely different.

这种思维转变也非常不同。

And just this mindset shift is very different.

我听说你为团队制定了非常具体的原则，当新人加入时，你会带他们逐一了解这些原则。

I hear you have these very specific principles that you've codified for your team that when people join you, you kind of walk them through them.

我相信其中一条是：与其自己做，不如让Claude来做。

I believe one of them is what's better than doing something, having Claude do it.

这感觉就像你刚才描述的内存泄漏问题一样，几乎忘了那个原则——等等，让我看看Claude能不能帮我解决这个问题。

And it feels like that's exactly what you described with this memory leak is just like, almost forgot that principle of like, okay, let me see if Claude can solve this for me.

当你对每件事都略微资金不足时，就会发生一件有趣的事，因为人们被迫去标准化流程。

There's this interesting thing that happens also when you underfund everything a little bit because then people are forced to codify.

这是我们经常看到的现象。

And this is something that we see.

所以你知道，有时候在工作中，我们只安排一名工程师负责一个项目，而他们之所以能快速交付，是因为他们本身就想尽快完成，这是一种内在的动力。

So, you know, for work where sometimes we just put, like, one engineer on a project, and the way that they're able to ship really quickly because they wanna ship quickly, this is like an intrinsic motivation that comes from within.

只是单纯地想把事情做好。

It's just wanting to do a good job.

如果你有一个好点子，你就特别想把它实现出来。

One if you have a good idea, you just really wanna get it out there.

没有人需要强迫你去做这件事。

No one has to force you to do that.

这是你自己想做的。

That comes from you.

如果你有Claude，你就可以用它来自动化很多工作，这是我们反复看到的情况。

And and so if you have claud, you can just use that to automate a lot of work, and that that's kind of what we see over and over.

所以我认为其中一个原则就是稍微减少对项目的投入。

So I think that's kind of like one principle is underfunding things a little bit.

我认为另一个原则就是鼓励人们加快速度。

I think another principle is just encouraging people to go faster.

所以如果你今天能做某件事，就应该今天做完。

So if you can do something today, you should just do it today.

这是我们团队非常非常鼓励的做法。

And this is something we really, really encourage on the team.

早期这非常重要，因为那时候只有我一个人。

Early on, was really important because it was just me.

所以我们唯一的竞争优势就是速度。

So our only advantage was speed.

这是我们能在如此拥挤的编码市场中推出有竞争力产品的唯一方式。

That's the only way that we could ship a product that would compete in this very crowded coding market.

但如今，这仍然是我们团队坚持的一项原则。

But nowadays it's still very much a principle we have on the team.

如果你想更快地推进，一个非常好的方法就是让Claude承担更多工作。

And if you want to go faster, a really good way to do that is to just have Claude do more stuff.

因此，这非常有力地推动了这种做法。

So it just very much encourages that.

这种资金不足的想法非常有趣，因为通常人们觉得人工智能会让你不需要那么多员工，不需要那么多工程师。

This idea of underfunding, it's so interesting because in general, there's this feeling like AI is going to allow you to not have as many employees, not have as many engineers.

所以不仅仅是你能更高效，你所说的其实是，如果你资金不足，反而会做得更好。

And so it's not only you can be more productive, what you're saying is that you will actually do better if you underfund.

这不仅仅是人工智能能让你更快，而是如果你有更少的人在做一件事，你会从人工智能工具中获得更多的收益。

It's not just that AI can make you faster, it's you will get more out of the AI tooling if you have fewer people working on something.

是的。

Yeah.

如果你雇用了优秀的工程师，他们会自己找到解决方法，尤其是当你赋予他们这种自主权时。

If you hire great engineers, they'll figure out how to do it, especially if you empower them to do it.

这实际上是我经常跟CTO以及各种公司谈论的话题。

This is something I actually talk a lot about with CTOs and all sorts of companies.

我的建议通常是，不要试图优化。

My advice generally is don't try to optimize.

不要在一开始就想方设法削减成本。

Don't try to cost cut at the beginning.

首先，给工程师尽可能多的令牌。

Start by just giving engineers as many tokens as possible.

现在你开始看到像Anthropic这样的公司，每个人都可以使用大量令牌。

And now you're starting to see companies like, at Anthropic, everyone can use a lot of tokens.

我们开始看到一些公司把这当作福利。

We're starting to see this come up as a perk at some companies.

比如，你加入后就能获得无限令牌。

Like if you join, you get unlimited tokens.

我非常鼓励这一点，因为它让人们能够自由尝试那些原本被认为太疯狂的想法。

This is a thing I very much encourage because it makes people free to try these ideas that would have been too crazy.

如果某个想法可行，那你再想办法扩展它。

Then if there's an idea that works, then you can figure out how to scale it.

优化和成本控制的时机应该在之后，比如看看是否可以用Haiku或Sana代替Opus之类的模型。

That's the point to optimize and to cost cut, figure out maybe you can do it with Haiku or with Sana instead of Opus or whatever.

但在初期，你只需要大量投入令牌，看看这个想法是否可行，并给予工程师这样的自由。

But at the beginning, you just want to throw a lot of tokens at it and see if the idea works and give engineers the freedom to do that.

所以这里的建议是，使用这些模型时，对令牌的使用要放宽一些，别太在意成本。

So the advice here is, just be be loose with your tokens with the the cost on on using these models.

听到这些建议的人可能会想：‘当然了，他是在Anthropic工作。’

People hearing this may be like, of course, he works at Anthropic.

他当然希望我们尽可能多地使用令牌。

He'd want us to use as many tokens as possible.

但你在这里说的是，最有意思、最具创新性的想法，往往来自于有人大胆尝试极限，看看能实现什么。

But what you're saying here is the most interesting innovative ideas will come out of someone just kind of taking it to the max and seeing what's possible.

是的。

Yeah.

我认为现实情况是，在小规模下，你不会产生巨额账单或类似的情况。

And I think the reality is, at small scale, you're not going to get a giant bill or anything like this.

如果只是个别工程师在做实验，令牌成本相对于他们的薪资或公司运营的其他成本来说，仍然相对较低。

If it's an individual engineer experimenting, the token cost is still probably relatively low relative to their salary or other costs of running the business.

所以这实际上并不是一笔巨大的开销。

So it's actually not a huge cost.

当事情规模扩大时，比如他们开发出一个很棒的东西，但消耗了海量的令牌，成本变得非常高，这时就是你该优化的时候了。

As the thing scales up, so let's say they build something awesome and then it takes a huge amount of tokens, and then the cost becomes pretty big, that's the point at which you want to optimize it.

但不要过早地这么做。

But don't do that too early.

你有没有见过一些公司，它们的令牌成本比员工工资还高？

Have you seen companies where their token cost is higher than their salary?

你认为这是我们未来会看到的一种趋势吗？

Is that a trend you think we're going to find and see?

在Anthropic，我们已经开始看到一些工程师每月在令牌上花费数十万美元。

Know, at Anthropic, we're starting to see some engineers that are spending hundreds of thousands a month in tokens.

所以我们已经开始看到这种情况了。

So we're starting to see this a little bit.

有一些公司，我们也开始看到类似的现象。

There's some companies that are we're starting to see similar things.

是的。

Yeah.

回到编程，你怀念写代码的感觉吗？

Going back to coding, do you miss writing code?

你有没有感到遗憾，作为软件工程师，你不再做这件事了？

Is this something you're kind of sad about that this is no longer a thing you will do as a software engineer?

这挺有意思的。

It's funny.

对我来说，当你学习工程时，它是非常实用的。

For me, you know, like, when when I learned engineering, for me, it was very practical.

我学工程是为了能动手做出东西。

I learned engineering so I could build stuff.

我是自学的。

And for me, I was I was self taught.

你在学校学的是经济学，而不是计算机科学。

You know, like, I studied economics in school, but I didn't study CS.

但我很早就自学了工程。

But I I taught myself engineering kinda early on.

我初中时就开始编程了。

I was programming in, like, middle school.

从一开始，编程对我来说就非常实用。

And from the very beginning, it was very practical.

所以我学编程的初衷其实是想在数学考试中作弊。

So I actually I've learned to code so that I can cheat on a math test.

那是我学编程的第一个目的。

That was the first thing.

那时候有图形计算器，我就把答案编进去。

Had these graphing calculators, and I just programmed the answer into

TI-83？

TI-eighty three?

TI-83 Plus。

TI-eighty three plus.

是的，没错。

Yeah, yeah, exactly.

加号。

Plus.

加号。

Plus.

是的。

Yeah.

所以我把答案都编进了计算器，然后下一次数学考试，或者第二年，题目变得太难了。

So I programmed the answers in, and then the next math test, whatever, the next year, it was just too hard.

我没法把所有答案都编进去，因为我根本不知道题目是什么。

I couldn't program all the answers in because I didn't know what the questions were.

所以我得写一个小小的求解器，就是一个能解答这些代数题之类的程序。

And so I had to write a little solver so that was a program that would just solve these these algebra questions or whatever.

然后我发现可以用一根小数据线把程序传给班上其他同学，这样全班都能得A了。

And then I figured out you can get a little cable, can give the program to the rest of the class, and then the whole class gets A's.

但后来我们都被抓到了，老师让我们别再这么干了。

But then we all got caught and the teacher told us to knock it off.

从一开始，对我来说编程就一直非常实用，编程是一种构建东西的方式。

From the very beginning, it's always been very practical for me, where programming is a way to build a thing.

它本身并不是目的。

It's not the end in itself.

在某个时刻，我个人陷入了编程之美的深渊。

At some point, I personally fell into the rabbit hole of the beauty of programming.

所以我写了一本关于TypeScript的书。

So, like, I I wrote a book about TypeScript.

实际上，当时那场活动是世界上规模最大的TypeScript聚会，因为我爱上了这门语言，深入钻研了函数式编程等各种内容。

I sort of the actually, at the time, it was the world's biggest TypeScript meetup just because I fell in love with the language itself, and I kinda got in deep into, like, functional programming and and all this stuff.

我认为很多程序员都会被这些东西分散注意力。

I think a lot of coders, they get distracted by this.

对我来说，编程，尤其是函数式编程，始终有一种美感。

For me, it was always sort of there is a beauty to programming and especially to functional programming.

类型系统也有其美感。

There's a beauty to type systems.

当你解决一个非常复杂的数学问题时，会有一种特别的兴奋感。

There's a certain kind of this buzz that you get when you solve a really complicated math problem.

当你平衡好类型，或者程序本身非常优美时，这种感觉有点类似。

It's kind of similar when you kind of balance the types or the program is just really beautiful.

但这并不是最终目的。

But it's really not the end of it.

对我来说，编程更像是一种工具，是实现目标的方式。

I think for me, coding is very much a tool, and it's a way to do things.

话虽如此，并不是每个人都这么想。

That said, not everyone feels this way.

比如，团队里有一位叫莉娜的工程师，她周末还会手写C++，因为她真的很享受手写C++的过程。

So for example, you know, like, there's one engineer on the team, Lina, who was still writing C plus plus on the weekends by hand because for her, she just really enjoys writing C plus plus by hand.

每个人都不一样。

And so everyone is different.

我认为，即使这个领域在变化，一切都在变化，始终都有空间去做这样的事。

And I think even as this field changes, even as everything changes, there's always space to do this.

只要你愿意，总有机会享受这种艺术，亲手去做一些事情。

There's always space to enjoy the art and to kind of do things by hand if you want.

你担心自己的工程师技能会退化吗？

Do you worry about your skills atrophying as an engineer?

这是你担心的问题，还是说，你知道的，这本来就是会发生的？

Is that something you worry about, or is it just like, you know, this is just how it's gonna go?

我觉得这就是事情发展的自然方式。

I think it's just the way that it happens.

我个人不太担心这个。

I I don't worry about it too much personally.

对我来说，编程是一个连续的过程。

I think for me, like, programming is on is on a continuum.

你知道，很久以前，软件其实还算是比较新的东西。

And, you know, like, way back in the day, you know, like, software actually is, like, relatively new.

对吧？

Right?

比如，如果你看看今天用运行在虚拟机上的软件编写程序的方式，这种编程方式从上世纪60年代左右就已经是这样了。

Like, if you look at the way programs are written today using software that's running on a virtual machine or something, this has been the way that we've been writing programs since probably the 1960s.

所以已经过去六十年左右了。

So it's been sixty years or something like that.

再之前，用的是打孔卡片。

Before that, it was punch cards.

再之前，用的是开关。

Before that, it was switches.

再之前，用的是硬件。

Before that, it was hardware.

再之前，其实就是笔和纸。

And before that, it was you know, like literally pen and paper.

那是一整间屋子的人，用纸笔做数学计算。

It was like a room a room full of people that were doing math on on paper.

所以，编程一直都是这样不断变化的。

And so, you know, programming has always changed in this way.

在某些方面，你仍然希望理解下一层的机制，因为这能帮助你成为更好的工程师。

In some ways, you still want to understand the layer under the layer because it helps you be a better engineer.

我认为在未来一年左右，这种情况还会持续。

I think this will be the case maybe for the next year or so.

但我认为很快这就不那么重要了。

But I think pretty soon it just won't really matter.

它只会像运行在程序下面的汇编代码一样，或者类似的东西。

It's just gonna be kind of like the assembly code running under the program or something like this.

从情感上讲，我一直觉得必须不断学习新东西。

At an emotional level, I feel like I've always had to learn new things.

作为一名程序员，这其实并不觉得是全新的，因为总会有新的框架出现。

And as a programmer, it's actually not it doesn't feel that new because there's always new frameworks.

总会有新的编程语言。

There's always new languages.

这在我们这个行业里已经是习以为常的事了。

It's just something that we're quite comfortable with in the field.

但与此同时，这并不适用于所有人。

But at the same time, this isn't true for everyone.

我认为对一些人来说，他们会感受到更大的失落、怀旧或衰退之类的情绪。

And I think for some people, they're gonna feel a greater sense of, I don't know, maybe like loss or nostalgia or atrophy or something like this.

我不知道你有没有看到，但埃隆说过，为什么AI不直接写二进制代码呢？

I don't know if you saw this, but Elon was saying that why isn't the AI just writing binary straight to binary?

毕竟，所有这些编程抽象的意义何在？

Because what's the point of all this programming abstraction in the end?

是的。

Yeah.

这是个好问题。

That's a good question.

我的意思是，如果你愿意，它完全能做到这一点。

I mean, it totally can do that if you wanted to.

天哪。

Oh, man.

所以，我在这里听到的是，始终存在一个问题：我应该学习编程吗？

So what I'm hearing here is in term there's always this question, should I learn to code?

学生在学校应该学习编程吗？

Should people in school learn to code?

我从你那里听到的观点是，一两年后，你其实并不需要掌握它。

What I heard from you is your take is in a year or two, you don't really need to.

我的观点是，对于那些现在使用代码助手或智能代理来编程的人，你仍然需要理解底层的原理。

My take is I think for for people that are using that are using quad code, that are using agents to code today, you still have to understand the layer under.

但没错，一两年后，这就不重要了。

But yeah, in a year or two, not gonna matter.

我在想，这最接近的历史类比是什么？

I was thinking about what is the right historical analog for this?

因为我们必须将这件事置于历史背景中，弄清楚我们曾经经历过哪些类似的转变，什么样的心理模型最适合理解它。

Because somehow we have to situate this thing in history and figure out when have we gone through similar transitions, what's the right mental model for this.

对我来说，最接近的类比是印刷术。

I think the thing that's come closest for me is the printing press.

如果你看看15世纪中期的欧洲，当时的识字率其实非常低。

And so if you look at Europe in the mid 1400s, literacy was actually very low.

当时人口中不到1%是抄写员，他们负责所有书写工作。

There was sub 1% of the population, it was scribes, that they were the ones that did all the writing.

他们也是唯一阅读的人。

Were the ones that did all the reading.

他们受雇于领主和国王，而这些领主和国王自己往往并不识字。

They were employed by lords and kings that often were not literate themselves.

因此，这个极小比例的人口承担了所有这些工作。

So it was their job of this very tiny percent of the population to do this.

后来，古腾堡和印刷术出现了。

And at some point Gutenberg and the printing press came along.

有一个惊人的数据：在印刷术发明后的五十年里，印刷出来的材料总量超过了此前一千年的总和。

There was this crazy stat that in the fifty years after the printing press was built, there was more printed material created than in the thousand years before.

因此，印刷材料的数量急剧增加。

And so the volume of printed material just went way up.

成本大幅下降。

The cost went way down.

在接下来的五十年里，成本下降了大约一百倍。

It went down something like 100x over the next fifty years.

如果你看识字率，实际上花了一段时间，因为学习读写非常困难。

And if you look at literacy, it actually took a while because learning to read and write is quite hard.

这需要一个教育体系。

It takes an education system.

这需要空闲时间。

It takes free time.

这需要不用整天在农场干活，才能有时间接受教育之类的事情。

It takes not having to work on a farm all day so that you actually have time for education and things like this.

但在接下来的两百年里，全球识字率上升到了70%。

But over the next two hundred years, it went up to 70% globally.

所以我认为，这可能是我们将看到的类似转变。

So I think this is the kind of thing that we might see is a similar kind of transition.

实际上，有一份有趣的历史文献，记录了15世纪一位抄写员对印刷术的看法。

And there was actually this interesting historical document where there was an interview with some scribe in the 1400s about how do you feel about the printing press?

他们其实非常兴奋，因为他们说：我最不喜欢做的事情就是抄写书籍。

And they were actually very excited because they were like, actually, thing that I don't like doing is copying between books.

我真正喜欢的是在书中绘制插图，以及装订书籍。

The thing that I do like doing is drawing the art in books and then doing the bookbinding.

我很高兴现在我的时间被解放出来了。

I'm really glad that now my time is freed up.

这很有趣。

And it's interesting.

作为一名工程师，我对这一点感到担忧。

As an engineer, I felt a peril with this.

我的感受是，我不必再做那些繁琐的编码工作了，因为这一直以来都是编程中琐碎的部分。

This is how I feel, where I don't have to do the tedious work anymore of coding because this has always been sort of the detail of it.

这一直是其中最枯燥的部分，还要折腾Git，使用各种不同的工具。

It's always been the tedious part of it and kind of messing with a git and kind of using all these different tools.

那并不是有趣的部分。

That was not the fun part.

有趣的部分是决定要构建什么，并且想出这些点子。

The fun part is figuring out what to build and coming up with this.

与用户交流。

Talking to users.

思考这些大型系统。

It's thinking about these big systems.

思考未来。

It's thinking about the future.

与团队中的其他人协作，这正是我现在能更多做的事情。

It's collaborating with, you know, other people on the team, and that that's what I get to do more of now.

令人惊叹的是，你正在开发的工具能让任何人做到这一点。

And what's amazing is that the tool you're building allows anybody to do this.

即使没有任何技术背景的人，也能做到你所描述的这些。

People that have no technical experience can do exactly what you're describing.

我一直在做一堆随机的小项目，每当我卡住时，只要说‘帮我搞定这个’，就能立刻突破瓶颈。

Like I've been doing a bunch of random little projects and it's just like anytime you get stuck, just like help me figure this out and you get unblocked.

我以前确实是工程师，在职业生涯早期干了十年，还记得花大量时间在库和依赖项上，总是想：天啊，我该怎么办？

Like I used to, yeah, I was an engineer for earlier in my career for ten years, and I just remember spending so much time on like libraries and dependencies and things, and just like, oh my god, what do I do?

然后去Stack Overflow上找答案，而现在只需要说‘帮我搞定这个’，就会直接给你步骤一、二、三、四。

And then looking on Stack Overflow, and now it's just like, help me figure this out, and here's step one, two, three, four.

好的，我们有了

Okay, we got

是的

Yeah.

没错

Exactly.

没错

Exactly.

今天早些时候我刚和一位工程师聊过。

I was talking to an engineer earlier today.

他们在用 Go 写某个服务，你知道的，已经一个月了，他们把服务搭建起来了。

They're like they're writing some service in Go, and, you know, it's been, a month already, and they they built up the service.

它运行得相当不错。

Like, it's it's working quite well.

然后我就说，好吧。

And then I was like, okay.

那么，写代码的感觉怎么样？

So, like, how do you feel writing?

他说，其实我还不太懂 Go，但我认为我们会看到越来越多这样的情况。

And he was like, you know, like, I I still don't really know Go, but and I I think we're gonna start to see more and more of this.

如果你知道它能正确高效地运行，那其实没必要掌握所有细节。

It's like, if you know that it works correctly and efficiently, then you you don't actually have to know all the details.

显然，软件工程师的生活已经发生了巨大变化。

Clearly, the life of a software engineer has changed dramatically.

从过去一两年开始，这简直像一份全新的工作了。

It's like a whole new job now as of the past year or two.

你认为AI接下来会最影响哪个角色？是在科技领域内，比如产品经理、设计师，还是科技领域之外？你觉得AI下一步会走向哪里？

What do you think is the next role that will be most impacted by AI within, either within tech, like product managers, designers, or even outside tech, just like, what do you think, where do you think AI is going next?

我认为受影响最大的会是与工程相关的周边岗位。

I think it's gonna be a lot of the roles that are adjacent to engineering.

所以，是的，可能是产品经理。

So, yeah, it could be like product managers.

也可能是设计。

It could be design.

也可能是数据科学。

It could be data science.

它将扩展到几乎所有能在电脑上完成的工作，因为模型在这方面会变得越来越强大。

It is gonna expand to pretty much any kind of work that you can do on a computer because the model is just gonna get better and better at this.

你知道，像协作者这样的产品只是实现这一点的最初方式，但这只是第一步。

And, you know, like, is the coworker product is kind of the first way to get at this, but it's just the first one.

我认为它能把AI，特别是具身AI，带给那些之前从未使用过它的人。

And it's the thing that I think brings AI to agentic AI to people that haven't really used it before.

人们现在才开始第一次真正感受到它的存在。

And people are starting just to to get a sense of it for the first time.

当我回想起一年前的工程工作时，没人真正了解什么是智能体。

When I think back to engineering a year ago, no one really knew what an agent was.

没人真正使用过它。

No one really used it.

但如今，这已经成为我们工作的常态。

But nowadays, it's just the way that we do we do our work.

当我观察当今的非技术性工作，或者半技术性工作，比如产品工作、数据科学之类的。

And then when I look at nontechnical work today, or maybe semi technical, like product work and data science and things like this.

当你看看人们使用的AI类型时，总是这些对话式AI。

When you look at the kinds of AI that people are using, it's always these conversational AI.

就像一个聊天机器人之类的。

It's like a chatbot or whatever.

但之前没人真正使用过智能体。

But no one really has used an agent before.

这个词‘代理’被频繁滥用，完全失去了本意。

And this word agent just gets thrown around all the time, and it's just so misused.

它已经毫无意义了。

It's lost all meaning.

但‘代理’实际上有非常明确的技术含义，即它是一种能够使用工具的AI，一个大语言模型。

But agent actually has a very specific technical meaning, which is it's an AI, it's an LLM that's able to use tools.

所以它不只是会说话，还能真正采取行动并与你的系统交互。

So it doesn't just talk, it can actually act and it can interact with your system.

这意味着它可以使用你的Google文档，发送电子邮件。

And this means it can use your Google Docs and it can send email.

它可以在你的电脑上运行命令，做各种这类事情。

It can run commands on your computer and do all this kind of stuff.

所以我认为，任何以这种方式使用计算机工具的工作，都将是下一个被改变的对象。

So I think any kind of job where you use computer tools in this way, I think this is going to be next.

这是我们整个社会必须去解决的问题。

This is something we have to figure out as society.

这是我们整个行业必须弄清楚的事情。

This is something we have to figure out as an industry.

对我来说，这也是为什么在Anthropic做这项工作感觉非常重要且紧迫的原因之一，因为我们对此非常重视。

And I think for me also, this is one of the reasons it feels very important and urgent to do this work ad Anthropic, because I think we take this very, very seriously.

因此，我们现在有了经济学家、政策专家和社会影响方面的专家。

And so now we have economists, we have policy folks, we have social impact folks.

我们只是想多多讨论这个问题，以便整个社会能共同决定该怎么做，因为这不应该是我们单方面决定的事。

This is something we just want to talk about a lot, so as society, we can figure out what to do, cause it shouldn't be up to us.

所以一个关键问题，你其实隐约提到了，就是工作和失业之类的问题。

So the big question, which is you're kind of alluding to, is jobs and job loss and things like that.

有一种叫做杰文斯悖论的概念，意思是当我们能做更多事时，反而会雇佣更多人，实际情况并没有看起来那么可怕。

There's this concept of Jevan's paradox of just as we can do more, we hire more, and it's not actually as scary as it looks.

在AI成为工程工作重要组成部分的过程中，你到目前为止有什么样的体验？

What have you experienced so far, I guess, with AI becoming a big part of the engineering job?

你们的招聘人数比没有AI时更多了吗？

Just are you hiring more than if you didn't have AI?

关于工作的其他想法。

And just thoughts on jobs.

是的。

Yeah.

我的团队正在招聘。

I mean, for our team, we're we're hiring.

所以Quadco团队正在招聘。

So Quadco team is hiring.

如果你感兴趣，可以去看看Anthropic的招聘页面。

If you're interested, just check out the jobs page on on Anthropic.

我个人觉得，所有这些变化都让我更享受我的工作了。

Personally, it's you know, all all this stuff has just made me enjoy my work more.

我从未像今天这样享受编程，因为我再也不用处理那些琐碎的细节了。

I have never enjoyed coding as much as I do today because I don't have to deal with all the minutiae.

所以对我个人而言，这非常令人兴奋。

So for me personally, it's been quite exciting.

我们经常听到很多客户说，他们非常喜欢这个工具，非常喜欢Quad Code，因为它让编程再次变得愉快。

This is something that we hear from a lot of customers where they love the tool, they love quad code because it just makes coding delightful again.

这对他们来说真的非常有趣。

And that's just that's just so fun for them.

但很难预测这件事最终会走向何方。

But it's hard to know where this thing is gonna go.

而且，我再次不得不去寻找一些历史上的类比。

And I, again, I just like I have to reach for these historical analogs.

我认为印刷术就是一个非常好的例子，因为这项原本只被少数懂读写的人掌握的技术，变得对每个人开放了。

And I I think the printing press is just such a good one because what happened is this technology that was locked away to a small set of people, knowing how to read and write, became accessible to everyone.

它本质上是一种民主化的技术。

It was just inherently democratizing.

每个人都能开始做这件事了。

Everyone started to be able to do this.

如果当时不是这样，那么像文艺复兴这样的事情根本不可能发生。

And if that wasn't the case, then something like the Renaissance just could never have happened.

因为文艺复兴的很大一部分是关于知识的传播。

Because a lot of the Renaissance, it was about knowledge spreading.

它是关于人们用来交流的书面记录，因为当时没有电话之类的东西。

It was about written records that people used to communicate, because there were no phones or anything like this.

那时候也没有互联网。

There was no Internet at the time.

所以，这接下来会带来什么可能性呢？

So it's about like, what does this enable next?

对我来说，这是非常乐观的版本，也是我真正感到兴奋的部分。

And I think that's the very optimistic version of it for me, and that's the part that I'm really excited about.

这简直难以想象。

It's just unimaginable.

你知道吗？

You know?

如果印刷术没有被发明，我们今天就不可能这样交谈。

Like, we couldn't be talking today if the printing press hadn't been invented.

比如，我们的麦克风根本不会存在。

Like, our microphones wouldn't exist.

我们周围的一切都不会存在。

None of the things around us would exist.

如果当初不是这样，根本不可能协调如此庞大群体的行动。

It just wouldn't be possible to coordinate such a large group of people if that wasn't the case.

所以我想象一个几年后的世界，每个人都能编程。

And so I imagine a world, you know, a few years in the future where everyone is able to program.

那会解锁什么？

And what does that unlock?

任何人都可以随时开发软件。

Anyone can just build software anytime.

我真的不知道。

And I have no idea.

这就像在十五世纪时，没人能预测到今天这样。

It's just the same way that in the fourteen hundreds, no one could have predicted this.

我觉得也是同样的道理。

I think it's the same way.

但我确实认为，与此同时，这将会造成极大的冲击，并让很多人感到痛苦。

But I do think in the meantime, it's gonna be very disruptive, and it's gonna be painful for a lot of people.

再次强调，作为社会，我们必须进行这场对话，必须一起找到解决办法。

And again, as a society, this is a conversation that we have to have, and this is a thing that we have to figure out together.

所以对于听到这些、希望在这场混乱中取得成功的人，有什么建议吗？

So for folks hearing this that want to succeed and, you know, make it in this crazy turmoil we're entering, Any advice?

是应该多尝试AI工具，熟练掌握最新的技术吗？

Is it, you know, play with AI tools, get really proficient at the latest stuff?

你还有什么其他建议，可以帮助人们保持领先？

Is there anything else that you recommend to help people, stay ahead?

是的。

Yeah.

我觉得基本上就是这样。

I think that's pretty much it.

尝试这些工具。

Experiment with the tools.

熟悉它们。

Get to know them.

不要害怕它们。

Don't be scared of them.

直接投入进去，试试看，站在技术前沿，成为开拓者。

Just, dive in, try them, be on the bleeding edge, be on the frontier.

也许第二条建议是，尽量让自己成为一个通才，而不是像过去那样只专精于一点。

Maybe the second piece of advice is try to be a generalist more than you have in the past.

比如，在学校里，很多学计算机科学的人只学编程，其他东西学得很少。

For example, in school, a lot of people that study CS, they learn to code, and they don't really learn much else.

也许他们只学了一点点系统架构之类的东西。

Maybe they learn a little bit of systems architecture or something like this.

但我每天合作的一些最有效的工程师，以及一些最有效的产品经理等，他们都跨越了不同领域。

But some of the most effective engineers that I work with every day and some of the most effective, you know, like product managers and so on, they cross over disciplines.

所以在四边形代码团队，每个人都写代码。

So on the quad code team, everyone codes.

我们的产品经理写代码，我们的工程经理写代码，我们的设计师写代码，我们的财务人员写代码，我们的数据科学家也写代码。

Know, our product manager codes, our engineering manager codes, our designer codes, our finance guy codes, our data scientist codes.

也就是说，团队里的每个人都写代码。

Like, everyone on the team codes.

然后，如果我看看具体的工程师，人们常常跨越不同的领域。

And then and then if I look at particular engineers, people often cross different disciplines.

因此，一些最出色的工程师是兼具产品和基础设施能力的混合型工程师，或是具有出色设计感的产品工程师，他们也能做设计；或者是对业务有深刻理解、能据此判断下一步行动的工程师，又或者是喜欢与用户交流、能准确传达用户需求以确定未来方向的工程师。

So some of the strongest engineers are hybrid product and infrastructure engineers, or product engineers with really great design sense, and they're able to do design also, or an engineer that has a really good sense of the business and can use that to figure out what to do next, or an engineer that also loves talking to users and can just really channel what users want to figure out what's next.

我认为，在未来几年里，获得最多回报的人不会只是原生AI用户，也不仅仅是因为他们能熟练使用这些工具，而是因为他们充满好奇心、是通才，能够跨越多个领域，思考他们所解决的更广泛的问题，而不仅仅是工程部分。

So I think a lot of the people that will be rewarded the most over the next few years, they won't just be AI native, and they don't just know how to use these tools really well, but also they're curious and they're generalists, and they cross over multiple disciplines and can think about the broader problem they're solving rather than just the engineering part of it.

你认为这三个独立的领域——工程、设计、产品管理——作为团队的思考方式仍然有用吗？

Do you find these three separate disciplines still useful as a way to think about the team?

它们分别是工程、设计和产品管理。

They're engineering design, product management.

尽管他们现在都在编程并参与思考要构建什么，你认为这三个角色在长期来看仍会保留吗？至少目前是这样？

Do you find those even though they are now coding and contributing to thinking about what to build, do you feel like those are three roles that will persist long term, at least at this point?

我认为短期内它们会继续存在，但我们已经开始看到这些角色之间有大约50%的重叠，很多人实际上在做同样的事情，而有些人则有专长。

I think in the short time it'll persist, but one thing that we're starting to see is there's maybe a 50% overlap in these roles, where a lot of people are actually just doing the same thing and some people have specialties.

例如，我写的代码比CAT RRPM多一些，而他更多地负责协调、规划、预测之类的工作。

For example, I code a little bit more versus CAT RRPM does a little bit more coordination or planning or forecasting or things like

是的。

this.

利益相关者对齐。

Stakeholder alignment.

利益相关者对齐。

Stakeholder alignment.

没错。

Exactly.

我确实认为未来会出现一种情况——到今年年底，我们会看到这些角色变得更加模糊，我认为在某些地方，‘软件工程师’这个头衔可能会逐渐消失，取而代之的是‘构建者’，或者每个人都会成为产品经理，每个人都会编程，类似这样的情况。

I I do think that there is a future where I think by the end of the year, what we're gonna start to see is these start to get even murkier murkier, where and I think in some places, the title software engineer is gonna start to go away, and it's just gonna be replaced by builder, or maybe it's just everyone's gonna be a product manager and everyone codes or something like this.

谁说招聘必须公平？

Who says hiring has to be fair?

我最近与每一位创始人和招聘经理交谈时，他们都感受到了同样的压力。

Every founder and hiring manager I've been speaking with these days is feeling the same pressure.

尽快招聘到最优秀的人才。

Hire the best people as fast as possible.

但招聘耗时耗力，达成共识困难，而优秀人才的竞争愈发激烈。

But recruiting is time consuming, alignment is hard, and competition for great talent keeps getting tighter.

这就是为什么 Elevenlabs、Brex、Replit、DEAL 以及另外 5000 多家机构都在使用 MetaView——这家为团队提供招聘真正不公平优势的 AI 公司。

That's why teams like Elevenlabs, Brex, Replit, DEAL, and 5,000 other organizations use MetaView, the AI company giving teams a real unfair advantage in hiring.

它们为你提供一套像招聘同事一样运作的 AI 代理。

They give you a suite of AI agents that behave like recruiting coworkers.

它们会根据你的具体要求寻找候选人，自动记录面试笔记，汇总招聘流程中的洞察，并帮助你识别人才库中最优人选。

They find candidates for you based on your exact criteria, take interview notes automatically, gather insights across your hiring process, and help you identify the best candidates in your pipeline.

AI 承担了招聘的繁琐工作，并为你提供一个真正的信息基准。

AI handles the recruiting toil and gives you a real source of truth.

这意味着每 hires 节省了数小时，团队能专注于最重要的事：赢得合适的候选人。

That means hours saved per hire and a team focused on what matters most: winning the right candidates.

别让竞争对手抢在你前面招到人。

Don't let your competitors outhire you.

MetaView 的客户职位招聘速度加快了 30%。

MetaView customers close roles 30% faster.

立即免费试用 MetaView，访问 metaview.ai/leni 额外获得一个月的候选人搜寻服务。

Try MetaView today for free and get an extra month of sourcing at metaview.ai/leni.

那就是 metaview.ai/leni。

That's metaview.ai/leni.

你提到你现在更享受编程了。

You talked about how you're enjoying coding more.

我其实就在 Twitter 上做了一个小规模的非正式调查。

I actually did this little informal survey on Twitter.

我不知道你有没有看到，我做了三个不同的投票。

I don't know if you saw this where I just asked I did three different polls.

我问了工程师们：自从使用AI工具以来，你们是更享受工作了，还是更不享受了？

I asked engineers, are you enjoying your job more or less since adopting AI tools?

然后我为产品经理和设计师分别做了单独的调查。

Then I did a separate one for PMs and one for designers.

工程师和产品经理中，70%的人表示他们更享受工作了，大约10%的人表示他们更不享受了。

Both engineers and PMs, 70% of people said they are enjoying their job more and about 10% said they're enjoying their job less.

有趣的是，设计师中只有55%的人表示他们更享受工作了，20%的人表示他们更不享受了。

Designers, interestingly, only 55% said they're enjoying their job more, and 20% said they're enjoying their job less.

我觉得这非常有意思。

I thought that was really interesting.

这太有趣了。

That's super interesting.

我真想跟这些人都聊聊，不管是属于‘更享受’组还是‘更不享受’组的，就想知道原因。

I'd I'd love to talk to these people, you know, both in the more bucket and the less bucket, just to understand.

你有跟进过他们中的任何人吗？

Did did you get to follow-up with any of them?

有几个人回复了，我们实际上正在做一个后续调查，会在节目笔记中提供链接，以更深入地探讨其中一些内容。

They a few people replied, and we're actually doing a follow-up poll that we'll link to in the show notes of going deeper into some of this stuff.

但有很多因素会影响工作的趣味性，比如哪些让工作更有趣，哪些让工作更无趣。

But a lot of there's, like, you know, the factors that make it more fun and less fun.

设计师们实际上并没有分享太多关于为什么他们工作乐趣减少的原因，那些被问到的人很少提及。

The designers, they didn't share a lot actually of just the people that are actually asked, like, Why are you enjoying your job less?

我听到的不多，所以我很想知道那里到底发生了什么。

I didn't hear a lot, so I'm curious what's going on there.

是的。

Yeah.

我在Anthropic也看到一点类似的情况，我认为每个人的技术水平都相当高。

I'm seeing this a little bit with At Anthropic, I think everyone is fairly technical.

我们在招聘时就会对此进行筛选。

This is something that we screen for when people join.

即使是非技术岗位的候选人，也要经历大量的技术面试，而我们的设计师大多都会写代码。

There's a lot of technical interviews that people go through even for nontechnical functions, and our designers have largely code.

所以我认为对他们来说，这是一件他们一直享受的事情，因为现在他们不再去打扰工程师，而是可以直接去写代码。

So I think for them, this is something that they have enjoyed from what I've seen because now instead of bugging engineers, they can just, like, go in and code.

甚至一些以前不会编程的设计师也开始学习了。

And even some designers that didn't code before have just started to do it.

对他们来说，这太棒了，因为他们可以自己解决障碍。

And for them, it's great because they can unblock themselves.

但我非常想听听更多人的经历，因为我敢打赌，情况并不是都像这样一致的。

But I'd be really interested just to hear more people's experiences because I I bet it's not uniform like that.

是的。

Yeah.

所以如果你正在听这个节目，如果你觉得工作没那么有趣了，或者享受感降低了，请在评论区留言。

So maybe if you're listening to this, leave a comment if you're finding your jobs less fun and you're enjoying your job less.

因为根据你所说的，以及我从大多数人那里听到的信息，70%的产品经理和工程师反而更享受他们的工作了。

Because what you're saying and what I'm hearing from most people, 70% of PMs and engineers are loving their job more.

如果你不属于这个群体，那可能就有些问题了。

That's like, if you're not in that bucket, you could something's going on.

是的。

Yeah.

对。

Yeah.

我们确实看到人们也在使用不同的工具。

We we do see that people use also different tools.

例如，我们的设计师更多地使用Quad桌面应用来进行编码。

So for example, our designers, they use the Quad desktop app a lot more to to do their coding.

你只需下载桌面应用即可。

So you just download the desktop app.

里面有一个代码标签。

There's a code tab.

它就在协作功能旁边。

It's right next to co work.

而且这实际上是完全相同的Quad代码，所以是同一个代理和所有内容。

And it's actually the same exact Quad code, so it's like the same agent and everything.

我们已经使用这个功能很多很多个月了。

We've had this for, you know, for many, many months.

因此，你可以用它来编码，而无需打开大量终端，同时仍然能获得Quad代码的强大功能，最重要的是，你可以同时运行任意数量的Quad会话。

And so you can use this to code in a way that you don't have to open a bunch of terminals, but you still get the power of quad code, and the biggest thing is you can just run as many quad sessions in parallel as you want.

我们称之为多Quad操作。

We call this multi quadding.

所以，对于非工程师来说，这种方式更加原生一些。实际上，这又回到了将产品带到用户所在位置的理念。

So this is a it's little more native, I think, for folks that are not engineers, And really, this is back to bringing the product to where the people are.

你不想让别人使用不同的工作流程。

You don't wanna make people use a different workflow.

你不想让他们费劲去学习新东西。

You don't wanna make them go out of their way to earn a new thing.

无论人们在做什么，如果你能让它变得更简单一点，那就会成为一个更好的产品，人们也会更喜欢。

It's whatever people are doing, if you can make that a little bit easier, then that's just gonna be a much better product that people enjoy more.

这正是潜在需求的原则，我认为这是产品领域最重要的单一原则。

And this is just this principle of latent demand, which I think is just the single most important principle in product.

你能谈谈这个吗？

Can you talk about that actually?

因为我本来就想说到这个。

Because I was going to go there.

解释一下这个原则是什么，以及当你释放这种潜在需求时会发生什么。

Explain what this principle is and just what happens when you unlock this latent demand.

潜在需求指的是这样一种理念：如果你以一种允许用户以非设计初衷的方式使用或滥用产品的方式来构建产品，而用户用它来做他们真正想做的事情，那么这能帮助你作为产品构建者了解产品下一步该往哪里发展。

Latent demand is this idea that if you build a product in a way that can be hacked or can be kind of misused by people in a way it wasn't really designed for to do kind of something that they want to do, then this helps you as the product builder learn where to take the product next.

一个例子就是Facebook Marketplace。

So an example of this is Facebook Marketplace.

这个团队的负责人Fiona，她实际上是Marketplace团队的创始负责人，她经常谈到这一点。

So the the manager for the team, Fiona, she she was actually the founding manager for the Marketplace team, and she talks about this a lot.

Facebook Marketplace的灵感来源于一个观察，那大概是在2036年左右，当时Facebook群组中40%的帖子都是关于买卖东西的。

Facebook Marketplace is sort of based on the observation back in this must been like 2036 or or something like this, that 40% of posts in Facebook groups are buying and selling stuff.

这太惊人了。

So this is crazy.

人们在利用Facebook群组进行买卖，但这并不是那种安全意义上的滥用。

It's like people are abusing the Facebook groups product to buy and sell, and it's not it's not abuse in kind of like a security sense.

这种滥用在于，没有人设计这个产品是为了这个用途，但人们却自己摸索出了它的用处，因为它在这方面实在太有用。

It's abuse in that no one designed the product for this, but they're kind of figuring it out because it's it's just so useful for this.

所以这一点非常明显。

And so it was pretty obvious.

如果你打造一个更好的产品来让人们买卖，他们一定会喜欢。

If you build a better product to let people buy and sell, they're gonna like it.

从这一点来看，市场平台一定会大受欢迎，这一点非常清楚。

And it was just very obvious that marketplace would be a hit from this.

所以第一步是推出买卖群组，也就是专门用于让人们进行买卖的群组，第二步才是推出市场平台。

And so the first thing was buy and sell groups, kind of special purpose groups to let people do that, and the second product was marketplace.

我认为，Facebook dating的起源也差不多是这样。

Facebook dating, I think, started in a pretty similar place.

我认为当时的观察是，如果你查看个人资料浏览数据，会发现60%的资料浏览来自非好友、异性之间。

And I think the observation was if you look at people looking at if you look at profile views, so people looking at each other's profiles on Facebook, 60% of profile views were people that are not friends with each other that are opposite gender.

所以这是一种传统的约会模式。

And so this is this kind of like traditional kind of dating setup.

人们只是在偷偷关注彼此。

People are just creeping on each other.

所以如果你为这种需求打造一个产品，可能会成功。

So maybe if you can build a product for this, might work.

因此，这种潜在需求的概念我觉得非常强大。

So this idea of latent demand, I think, is just so powerful.

例如，Cowork 也是源于这一点。

For example, this is also where Cowork came from.

我们发现，在过去六个月左右，很多使用 Quad Code 的人并不是用它来写代码。

We saw that for the last six months or so, a lot of people using quad code were not using it to code.

有个人在 Twitter 上用它来种植番茄。

There was someone on Twitter that was using it to grow tomato plants.

还有另一个人用它来分析自己的基因组。

There was someone else using it to analyze their genome.

有人用它来从损坏的硬盘中恢复照片。

Someone was using it to recover photos from a corrupted hard drive.

像是婚礼照片之类的。

Was like wedding photos.

还有人以为自己在用它分析核磁共振成像。

There was someone that was using it for think they were using it to analyze an MRI.

所以有这么多完全非技术性的使用场景，这一点非常明显。

So there's just all these different use cases that are not technical at all, and it was just really obvious.

人们为了用终端做这些事，真是费尽周折。

People are jumping through hoops to use a terminal to do this thing.

也许我们该为他们直接开发一个产品。

Maybe we should just build a product for them.

其实我们很早就注意到了这一点。

And we saw this actually pretty early.

大概在五月的时候，我记得走进办公室，看到我们的数据科学家布伦丹正在电脑上使用Quad Code。

Back in maybe May, I remember walking into the office and our data scientist, Brendan, a quad code on his computer.

他只是打开了一个终端。

He just had a terminal up.

我感到很震惊。

And I was shocked.

我当时说：布兰登，你在干什么？

I was like, Brandon, what are you doing?

你居然学会了怎么打开终端，这可是个非常技术导向的产品。

You figured out how to open the terminal, which is it's a very engineering product.

就连很多工程师都不愿意用终端。

Even a lot of engineers don't want to use a terminal.

这是最底层的工作方式，完全深入到计算机的细节里。

It's just like the lowest level way to do your work, just really, really in the weeds of the computer.

所以他学会了怎么使用终端。

And so he figured out how to use the terminal.

他下载了Node。

He downloaded Node.

JavaScript。

Js.

他下载了Quad Code，并在终端中进行SQL分析。

He downloaded quad code, and he was doing SQL analysis in the terminal.

这太疯狂了。

It was crazy.

然后到了下一周，所有的数据科学家都在做同样的事情。

And then the next week, all of the data scientists were doing the same thing.

所以当你看到人们以这种方式滥用产品——用它来做原本设计之外、但对他们很有用的事情时，这强烈表明你应该直接开发产品，人们自然会喜欢。

So when you see people abusing the product in this way, using it in a way that it wasn't designed in order to do something that is useful for them, it's just such a strong indicator that you should just build a product and people are going to like that.

这是为特定用途量身定制的。

It's something that's special purpose for that.

我认为现在，潜在需求还存在另一个有趣的维度。

I think now there there's also this kinda interesting second dimension to latent demand.

传统的看法是：观察人们在做什么，让这些事情变得更简单，赋予他们更多能力。

This is sort of the traditional framing is look at what people are doing, make that a little bit easier, empower them.

在过去六个月里，我看到的现代视角略有不同，那就是关注模型试图做什么，并让它变得更简单。

The modern framing that I've been seeing in the last six months is a little bit different, and it's look at what the model is trying to do and make that a little bit easier.

因此，当我们刚开始开发Quad Code时，我认为人们在用LLM设计产品时，往往把模型框定在一个固定范围内。

And so when we first started building quad code, I think a lot of the way that people approached designing things with LLMs is they kind of put the model in a box.

他们会想：我要构建一个这样的应用。

And they were like, here's this application that I wanna build.

我要实现这样一个功能，让模型来完成。

Here's the thing that I wanted to do a model.

你只需要负责其中的一个组件。

You're gonna do this one component of it.

这是你与这些工具、API以及其他东西交互的方式。

Here's the way that you're gonna interact with these tools and APIs and whatever.

而对于Cloud Code，我们反过来做了。

And for Cloud Code, we inverted that.

我们说，产品就是模型。

We said the product is the model.

我们希望将其暴露出来。

We wanna expose it.

我们希望为其提供最少的框架，赋予它最基础的一组工具，让它能够完成任务。

We wanna put the minimal scaffolding around it, give it the minimal set of tools so it can do the things.

它可以自行决定使用哪些工具。

It can decide which tools to run.

可以决定运行它们的顺序，等等。

Can decide in what order to run them in and so on.

我认为这一切很大程度上源于模型自身潜在的需求。

And I I think a lot of this was just based on kind of latent demand of what the model wanted to do.

因此在研究中，我们称之为‘在分布内’。

And so in research, we call this being on distribution.

你希望看到模型真正想做什么。

You wanna see, like, what the model is trying to do.

在产品层面，潜在需求就是同一个概念，只是应用在了模型上。

In product terms, latent demand is just the same exact concept, but applied to a model.

你提到过协同工作。

You talked about co work.

你最初发布时，我看到你说你们团队在十天内就完成了它。

Something that I saw you talk about when you launched that initially is you your team built that in ten days.

这太不可思议了。

That's insane.

是的。

Yeah.

我觉得它上线了。

I think it came out.

我觉得它很快就有了数百万用户在使用，而这样的产品竟然只用了十天就做出来了。

I think it was, like, you know, used by millions of people pretty quickly, something like that being built in ten days.

那里还有什么吗？

Anything there?

除了说我们用 Cloud Code 来开发它之外，还有其他故事吗？

Any stories there other than just it was just, you know, we used Cloud Code to build it, that's it.

是的。

Yeah.

这很有趣。

It it it's funny.

Cloud Code，正如我所说，我们发布时并没有立即受到欢迎。

ClotCode, like I said, when we released it, it was not immediately a hit.

它随着时间的推移才变得受欢迎，并且有几个关键转折点。

It became a hit over time, and there was a few inflection points.

其中一个就是，你知道的，Opus 4。

So one was, you know, like, Opus four.

它真的、真的出现了爆发。

It just really, really inflected.

然后在十一月，它再次爆发，之后就一直在持续增长。

And then in November, it inflected, and it just keeps inflecting.

每天的增长速度都变得越来越陡峭。

The growth just keeps getting steeper and steeper and steeper every day.

但你知道，头几个月它并不是爆款。

But, you know, for the first few months, it wasn't a hit.

人们在使用它，但很多人不知道怎么用。

People used it, but a lot of people couldn't figure out how to use it.

他们不清楚它的用途是什么。

They didn't know what it was for.

模型本身那时候也还不够好。

The model still, like, wasn't very good.

Cowork在发布时就立刻成了爆款，比早期的Cloud Code要成功得多。

Cowork, when we released it, it was just immediately a hit, much more so than Cloud Code was early on.

我觉得很大程度上要归功于Felix、Sam、Jenny以及开发这个产品的团队。

I think a lot of the credit honestly just goes to, like, Felix and and Sam and the and Jenny and the the team that built this.

这真是一个极其强大的团队。

It's just an incredibly strong team.

而且，Cowork的诞生源于一种潜在的需求。

And, again, the the place Cowork came from is just this latent demand.

我们看到人们用Cloud Code来做这些非技术性的事情，于是我们在思考该怎么办。

Like, we saw people using Cloud Code for these nontechnical things, and we're trying to figure out what do we do.

所以有几个月，团队一直在探索。

So for a few months, the team was exploring.

他们尝试了各种不同的方案。

They were trying all sorts of different options.

最后，有个人突然说：好吧。

And in the end, someone was just like, okay.

如果我们直接把Quad Code放进桌面应用里会怎么样？

What if we just take quad code and put it in the desktop app?

这基本上就是成功的关键。

And that's essentially the thing that worked.

于是十天内，他们完全用Quad Code把它开发出来了。

And so over ten days, they just completely used quad code to build it.

你知道，CoWorker其实内置了一个非常复杂的安全系统，这些防护机制能确保模型做正确的事。

You know, co worker actually there's this very sophisticated security system that's that's built in and essentially these guardrails to make sure that the model kind of does the right thing.

它不会失控。

It doesn't go off the rails.

例如，我们随它一起打包了一个完整的虚拟机。

So for example, we ship an entire virtual machine with it.

而Quad Code编写了所有这些代码。

And quad code just wrote all of this code.

所以我们只需要思考一下，好吧。

So we just have to think about, alright.

如何让这个系统对非工程师用户来说更安全、更易于自主使用？

How do we make this a little bit safer, a little more self guided for people that are not engineers?

它完全是由Quad Code实现的。

It was fully implemented with quad code.

花了大约十天时间。

Took about ten days.

我们提前发布了它。

We launched it early.

你知道，它仍然很粗糙，边缘部分也还有很多不完善的地方。

You know, it was still pretty rough, and it's still pretty rough around the edges.

但这就是我们学习的方式——无论是产品方面还是安全方面，我们都必须比原计划更早地发布产品，以便获得反馈，与用户交流，了解他们真正想要什么，从而塑造产品未来的发展方向。

But this is kind of the way that we learn, both on the product side and on the safety side is we have to release things a little bit earlier than we think so that we can get the feedback, so that we can talk to users, we can understand what people want, and that'll shape where the product goes in the future.

是的。

Yeah.

我觉得这个观点非常有趣，而且非常独特。

I think that point is so interesting, and it's so unique.

一直以来都有这样的理念：尽早发布，从用户那里学习，获取反馈，不断迭代。

There's always been this idea, release early, learn from users, get feedback, iterate.

很难准确知道AI的能力边界，以及人们会如何使用它，这正是一个独特的原因，促使我们更早地发布产品，正如你所描述的那样，帮助我们发现这种我们原本完全不了解的潜在需求。

The fact that it's hard to even know what the AI is capable of and how people will try to use it is like, is a unique reason to start releasing things early, that'll help you as you exactly describe this idea of what is the latent demand in this thing that we didn't really know?

让我们把它放出去，看看人们会怎么使用它。

Let's put it out there and see what people do with it.

对。

Yeah.