You Don't Win an Award for Being Wrong Longer

One of the weird things about startups is that you don't win an award for doing the same wrong thing for longer.

创业公司有件怪事:同一件错的事做得更久,并不会有人给你发奖。

Down the line when you fail, none of them will care.

等你日后真失败了,他们谁也不会在意。

So excited to welcome one of the hottest, most talked about startups from the valley, Windsurf and their CEO, Varun.

太高兴了,今天请到硅谷最火、被谈论最多的创业公司之一 Windsurf,还有他们的 CEO,Varun。

Companies that are usually first to a new paradigm are companies that are willing to disrupt themselves.

通常最先踏进一个新范式的公司,都是那些愿意颠覆自己的公司。

When you have a new great idea, even the crappy version of that idea is already amazing.

当你有了一个了不起的新想法,哪怕是它最糙的版本,也已经很惊艳了。

Startups don't fail because they look like messes inside.

创业公司失败,不是因为内部看着一团乱。

Startups fail because they don't do the right thing well enough.

创业公司失败,是因为没把对的事做得足够好。

It is much easier to run a company that has only one thing that matters than it is to run a company with the same amount of revenue with five things that matters.

经营一家只有一件事要紧的公司,比经营一家收入相同、却有五件事都要紧的公司,容易得多。

What idea do you think that you are too in love with today?

你觉得今天的你,太迷恋哪个想法?

I think if I was to go empirically

我觉得,要是按经验事实看

ready to go

准备好了

Varun, I'm so excited for this dude.

Varun,这期我太期待了,老兄。

I've heard so many good things by the way from Lee Marie, from Neil Mehta, from many others.

顺便说一句,从 Lee Marie、从 Neil Mehta、从其他很多人那儿,我听过太多关于你的好话。

So, thank you so much for joining me, man.

所以真的非常感谢你能来,兄弟。

Yeah, thanks a lot Harry for having me.

嗯,非常感谢 Harry 邀请我。

Not at all dude, but I wanted to kind of pick apart.

客气什么,兄弟。不过我想把一些事拆开来聊。

I've literally just been running and I've been listening to some of your shows before and there are a couple of segments where I was like, damn, I wish we double clicked on that.

我刚跑完步回来,路上听了几期你之前上过的节目,有几段我当时就在想:该死,真希望那儿能再往下挖一层。

And there was one that I thought I wanted to start with.

其中有一段我想拿来开场。

You said it's very rare that the first thing you believe will be the right thing.

你说过,你最先相信的那件事,很少会是对的那件事。

How have you changed your mind on what a truly great idea is versus what's not?

关于什么算真正伟大的想法、什么不算,你的看法后来是怎么变的?

I think there's a lot that's right about the Peter Thiel-ism of you want to pick something that seems a little bit not obvious to start with, right?

我觉得 Peter Thiel 那套说法有很多是对的——你要挑一个乍看之下不那么显而易见的东西,对吧?

Um for for whatever company you're starting.

不管你创办的是什么公司。

And the problem is most ideas that are not obvious are just bad ideas.

问题在于,大多数不显而易见的想法,其实就只是烂点子。

Like most ideas that are non-conventional, just because you are non-conventional doesn't mean you're going to pick a good idea.

大多数非常规的想法也一样——你非常规,不代表你就挑到了好想法。

Um, and I think people sometimes forget that.

我觉得人们有时候会忘掉这一点。

But I think at a high level though, like if you pick an idea that is like obvious to everyone, there's probably no alpha, right?

但往大了说,如果你挑的想法人人都看得明白,那大概率就没有 alpha 了,对吧?

A big company is going to go out and already beat you to the market, right?

大公司会抢在你前面把市场做掉,对吧?

Because they have more distribution resources and capital than you.

因为他们的分发、资源和资本都比你多。

So startups usually win by picking non-conventional ideas to the public.

所以创业公司通常靠挑那些在大众看来非常规的想法取胜。

But you kind of need to go in and be humble enough to understand that probably there's a good chance that your weird idea is not actually a good idea.

但你进场时也得足够谦虚,明白你那个古怪的想法很有可能压根就不是个好想法。

What do you believe was your non-conventional insight when you made the pivot to Windsurf?

你觉得当初转向 Windsurf 的时候,你那个非常规的洞察是什么?

Yeah, so I mean I could even start before this like the first idea that we had even like our company name has changed twice now at this point.

对,其实我可以从更早讲起,讲我们最早的那个想法——到今天为止,我们公司名字都已经换过两次了。

It was called Exafunction when we first built GPU virtualization technology was that GPU workloads were going to power the world, right?

那时候叫 Exafunction,我们最早做的是 GPU 虚拟化技术,当时的判断是:GPU 负载将会驱动整个世界,对吧?

That was like sort of the the mission there.

那算是我们当时的使命。

A lot of us at the company had previously worked in autonomous vehicles and ARVR um at the at the company.

公司里很多人以前是做自动驾驶和 AR/VR 的。

We had a fairly small team that was actually accurate that GPUs were going to become very popular.

我们团队当时相当小,而这个判断其实是准的——GPU 会变得非常热门。

Like we were we basically said Nvidia is going to sell a lot of GPUs.

我们基本上就是说,Nvidia 会卖出海量 GPU。

The thing we were wrong about was how diverse the GPU workloads were going to be, right?

我们判断错的地方在于,GPU 负载会有多么五花八门,对吧?

We thought a bunch of different model architectures would be run.

我们以为会有一堆不同的模型架构在跑。

There were going to be hundreds of different model architectures and we were wrong.

以为会冒出几百种不同的模型架构——结果我们错了。

All of them ended up being transformers.

最后它们全都是 transformer。

And in a world in which all the architectures look the same, um there's very little reason for us to be a differentiated infrastructure provider in that category.

而在一个所有架构长得都一样的世界里,我们要在这个品类里做一家有差异化的基础设施提供商,理由就很薄弱了。

So we needed to pivot at that point.

所以那个节点上我们必须转向。

But that's some one of those things where you don't know going into the market that that's how the world is going to turn out.

但这就是那种事:你进入市场之前,并不知道世界最后会长成这样。

Uh right.

对吧。

And we were just wrong about our hypothesis.

我们的假设就是错的。

And uh so that's why like sometimes you have a take that is not very obvious at the time and then you could be wrong.

所以说,有时候你在当时持有一个不那么显而易见的判断,然后你可能就是错的。

Totally.

完全同意。

The trouble is many people are in love with their thesis.

麻烦在于,很多人太迷恋自己的投资主题。

That's why I hate thesis driven investors bluntly.

所以说白了,我讨厌主题驱动型的投资人。

But many people are in love with their ideas and they're told that actually the true grit and persistence of an entrepreneur is what separates those that win versus those that don't.

但很多人就是迷恋自己的想法,而且总有人告诉他们:真正把赢家和输家分开的,是创业者的坚韧和执着。

How do you think about that versus a brilliant statement which I agree with.

这一点你怎么看?对照你说过的一句我特别认同的漂亮话。

You said never be too in love with your ideas.

你说,永远别太迷恋自己的想法。

I think I think this is the hardest part about starting a company or or building a company in in the first place, right?

我觉得这是创办一家公司、或者说从零打造一家公司,最难的地方,对吧?

It requires these two almost counter beliefs in your head, right?

它要求你脑子里同时装着两种几乎相反的信念,对吧?

Irrational optimism because if you don't have any optimism at all, you won't do anything and you won't get out of bed, right?

一是非理性的乐观——因为你要是一点乐观都没有,你什么都不会去做,连床都爬不起来,对吧?

And uncompromising realism.

二是毫不妥协的现实主义。

Every single day, you need to be asking your question, do we have a reason to exist, right?

每一天你都得问自己:我们有存在的理由吗?对吧?

And if the answer is not really, you need to change your mind really fast and reorient the entire company really fast.

如果答案是“不太有”,你就得非常快地改主意,非常快地把整家公司掉头。

One of the weird things about startups is that you don't win an award for doing the same wrong thing for longer.

创业公司有件怪事:同一件错的事做得更久,并不会有人给你发奖。

Like it might feel better to tell your co-workers, tell your investors, tell your friends that hey I'm still doing the thing I did.

跟同事说、跟投资人说、跟朋友说“嘿,我还在做原来那件事”,感觉可能是好受一些。

But down the line when you fail, none of them will care.

但等你日后真失败了,他们谁也不会在意。

What idea do you think that you are too in love with today?

你觉得今天的你,太迷恋哪个想法?

So I'll give you an example.

我举个例子。

For me as an investor, discipline around portfolio construction.

对我这个投资人来说,是投资组合构建上的那套纪律。

I love the belief of a disciplined investor, the craft of portfolio building.

我信奉“有纪律的投资人”这个理念,也喜欢搭组合这门手艺。

The challenge is in today's world, you have to break most rules.

问题是,在今天这个世界,你必须打破大部分规则。

And it doesn't mean that you're a cowboy and you're gambling, but it means you're much more elastic than the more concrete investor mind.

这不是说你变成了个莽撞的牛仔、在赌博,而是说你要比那种更死板的投资人心态灵活得多。

That's what I'm too in love with.

这就是我太迷恋的东西。

What do you think you're too in love with?

那你觉得你太迷恋的是什么?

I think I think if I was to go empirically, we were too in love with our ideas every time.

我觉得,要是按经验事实看,我们每一次都太迷恋自己的想法了。

Like other people outside looking in might have been like this company has pivoted multiple times and that's a positive outcome, but I wished we had done it faster.

外面的人看进来,可能觉得这家公司转向过好几次、结果还挺好,但我真希望我们当时能转得更快。

I still beat myself over the head on why did we not pivot from the GPU virtualization to the code AI company three months earlier.

我到现在还在捶自己:为什么没能早三个月从 GPU 虚拟化转去做代码 AI 公司。

I think we had the signals but we waited until the companies we were working with were starting to go belly up before we actually pivoted the company.

信号我觉得我们其实早就拿到了,可我们一直等到合作的那些公司开始一家家垮掉,才真的把公司转向。

Right?

对吧?

A lot of our customers at the time were autonomous vehicle companies and it was peak ZIRP era, right?

我们当时很多客户是自动驾驶公司,那会儿正是 ZIRP 的巅峰期,对吧?

And then in the middle of 2022 when a lot of them were unable to raise their next funding round, that's when we started to question, hey, like the market is changing, the technology is also changing, but also our our future customers are not looking like they're going to be able to kind of satisfy our sort of our market, right?

然后到 2022 年年中,他们中很多家融不到下一轮,我们才开始怀疑:嘿,市场在变,技术也在变,而且我们未来的客户看起来也撑不起我们这个市场,对吧?

And uh I wish we had done that faster.

我真希望我们当时动作能更快。

And the same is also true for for ideas like Windsurf, right?

像 Windsurf 这样的想法也一样,对吧?

Um, I think we knew agents were going to be very powerful and we could have built Windsurf a couple months earlier and I think we we should have done that if we could have.

我觉得我们当时就知道 agent 会非常强,Windsurf 本可以早几个月做出来;能做到的话,我们就该那么做。

Can I ask you a question?

我能问你个问题吗?

I just had the CEO of Fiverr on the show and he said that time to clone is one of the most important things that have changed over the last few years.

我刚请了 Fiverr 的 CEO 上节目,他说过去这几年里变化最要紧的事情之一,就是“克隆时间”。

Time to clone being the amount of time it takes for someone to copy exactly what you have.

所谓克隆时间,就是别人把你手上的东西原样抄过去要花多久。

Um, my question to you, how important is it to be first if time to clone is shorter than ever before?

我想问你的是:如果克隆时间比以往任何时候都短,那么做第一个还有多重要?

I think time being first being first does two things, right?

我觉得做第一个能带来两件事,对吧?

Um I think it's a signal of how you're running the company.

它反映了你在怎么经营这家公司。

That's one.

这是其一。

Um and the second thing is it allows you to learn faster.

第二件事是,它让你学得更快。

And I'll maybe double click on the first one.

我先把第一点展开讲讲。

Companies that are usually first to a new paradigm are companies that are willing to disrupt themselves.

通常最先踏进一个新范式的公司,都是那些愿意颠覆自己的公司。

They're organizations that are open to the idea of changing at a dime, right?

它们是那种说变就变的组织,对吧?

uh very very quickly and they're al always investing in technologies that are not incremental and and willing to sort of sort of break themselves right

非常非常快,而且它们一直在投那些不是增量式的技术,愿意把自己拆掉重来,对吧

um and I think that's a good organization in a category like software or code AI where the space is changing so so quickly right

我觉得在软件、代码 AI 这种变化极快的品类里,这才是好的组织,对吧

that's one

这是其一

and then two I think it kind of boils down to what does it mean if you're first?

第二点,我觉得归根结底就是:做第一个到底意味着什么?

that means you get to learn from the market faster and that means you're first to the next idea too right

意味着你能更快从市场里学到东西,也意味着下一个想法你还是第一个够到,对吧

uh because you you see where all the dead bodies are in your category

因为你看得见你这个品类里的尸体都埋在哪

Right.

对。

Um, one of the nice things that I like to think is um, if we have a product in market earlier, what are all the things we can learn in the first month that enable us to build the next product in the market as well.

我很喜欢这么想:如果我们的产品更早进到市场里,那头一个月我们能学到的所有东西,能让我们把下一个产品也推进市场。

And I think that that could give us a compounding advantage with time.

我觉得随着时间推移,这会给我们一种复利式的优势。

Is it not more helpful seeing someone else go first, see what they do as a mistake and then learn from their mistakes and leverage the cost that they have and then do it without that cost for you?

看着别人先走一步,看他们哪儿做错了,再从他们的错里学,把他们付出的代价拿来用,自己却不用付——这难道不是更有利吗?

I think the problem in software in general is they could have made a lot of mistakes internally during research and development that we we could never kind of take a look at

我觉得软件这行普遍的问题是,他们在研发过程里可能内部踩过很多坑,而这些我们根本看不到

and organizationally now I believe our company has so much wisdom in our category uh that there are ideas that we would not even touch because we are just like hey we've tried this before right

而在组织层面,我现在相信我们公司在这个品类里攒下了非常多的智慧,有些点子我们碰都不会碰,因为我们会说,嘿,这个我们早试过了

and it didn't actually work

结果确实没成

and that's something that's like built into the DNA of the company uh as we continue to innovate

随着我们不断创新,这些东西已经刻进了公司的 DNA

um and that's just not easy to just look at outside looking in.

而这些,外人从外往里看是很难看出来的。

What have you tried before that didn't work that the world doesn't know about?

有哪些是你们以前试过、没成,而外界并不知道的?

Yeah, so we we shipped a beta version of this product, but we were dog fooding it for such a long time.

好,这个产品我们是发过 beta 版的,但在那之前我们自己 dogfood 了非常久。

I'll give you maybe a couple examples here.

我给你举几个例子吧。

The first thing is we actually shipped a beta product for code review last year and we actually tried out different ways of shipping a code review product.

第一件事是,我们去年确实发过一个代码审查的 beta 产品,而且试过好几种做代码审查产品的路子。

We had a Chrome extension.

我们做过一个 Chrome 扩展。

We did something internally where we built a parallel website and we were able to try a bunch of things internally and none of them felt quite right.

内部还搞过一次,搭了个平行的网站,在里面试了一堆东西,但哪个感觉都不太对。

And now a couple weeks ago, we actually shipped a code review product and it feels much better and is providing a lot more value.

而就在几周前,我们真正发布了代码审查产品,感觉好太多了,带来的价值也大得多。

And we shipped it in the way we shipped it because of all the failed attempts that we made last year.

我们之所以用现在这个方式发,正是因为去年那些失败的尝试。

I'll give you a second example here.

我再给你举第二个例子。

For the code agent product that we launched with Windsurf, we actually had our first cut of that product in the beginning of last year and it was not good.

我们随 Windsurf 推出的 code agent 产品,其实去年年初就做过第一版,那一版不行。

It was not good for a variety of reasons and that enabled us to go back to the drawing board and fix up the different elements that would allow us to ship Windsurf, right?

不行的原因有很多,而这让我们能推倒重来,把那些能撑起 Windsurf 的环节一个个修好,对吧?

which was that we needed to get really better at codebase understanding as well as the models needed to get better.

也就是我们得在 codebase 理解上真正做得更好,同时模型也得变得更好。

But then once the once the models had already gotten better, our internal R&D on codebase understanding had gone good enough that we were ready to go and ship the product.

但等模型真的变好了,我们在 codebase 理解上的内部研发也已经足够好,我们就可以直接把产品发出去了。

So this is one of those things where if we had just waited for the rest of the world to have it, we would need to have played catch-up on multiple axes, right, just to ship the product.

所以这就是那种情况——如果我们只是等着世界上其他人先做出来,那光是把产品发出来,我们就得在好几个维度上补课,对吧?

Can I ask you do you advise founders to raise as much as possible early because it is about the number of at-bats that you have the experiments that you can take and if you raise more you have more experiments

我能问一句吗?你会建议创始人早期尽可能多融钱吗?因为关键在于你有多少 at-bats、能做多少次实验,融得越多,实验就越多

fundamentally do you advise that given the experience that you have

说到底,以你的经验,你会这么建议吗?

so we're we're obviously like just to shed some light here we actually raised our Series A on the GPU virtualization idea

所以我们——先交代一下背景,我们的 Series A 其实是靠 GPU 虚拟化这个想法融来的

Greenoaks actually led that led that round

那一轮是 Greenoaks 领投的

um whether or not they led

至于他们是不是领投——

they led your seed as well

你们的 seed 也是他们领投的

no they didn't

不,他们没有

yeah

嗯

they led our seed too, which uh whether or not they should have invested in the Series A on that one, I'll I'll I'll leave that to to the audience.

seed 也是他们领的,至于在那个方向上该不该继续投 Series A,我留给听众自己判断。

Uh but I think what that did give us is the confidence that hey, like if we pivot, we will still have the cash to go out and and pursue an idea.

但这确实给了我们底气——嘿,就算我们 pivot,手里还有现金去追新的想法。

And we actually launched Codeium, the extension product before this entirely free.

而我们推出 Codeium,也就是这之前的那个扩展产品,是完全免费的。

And we were able to do that because of the infrastructure expertise inside the company, but also we were confidently able to do that because we knew we had cash in the bank, right?

我们能这么干,一方面靠公司内部的基础设施功底,另一方面也是因为知道账上有钱,心里有底,对吧?

uh to go out and like actually ship this product and not have to worry about any details there, right?

让我们能放手把产品发出去,不用操心那些细枝末节,对吧?

So, I actually agree with you.

所以我其实同意你的说法。

It does give you more shots at that, but I think you need to be a company that is able to pivot very quickly, right?

钱多确实给你更多次出手的机会,但我觉得你得是一家能极快 pivot 的公司,对吧?

Um like most companies, I would say when they have 10, 15, 20 people, they're unwilling to change the idea, their idea entirely.

大多数公司,我觉得到了 10 个、15 个、20 个人的时候,就不愿意把自己的想法整个推翻了。

Um, we on the other hand decided we wanted to work on Codeium over a weekend, me and my co-founder, and we told the company on Monday and everyone started working on it that Monday despite the fact that we had we were making a couple million in revenue off the previous business.

而我们呢,是我和我的联合创始人在一个周末里决定要做 Codeium 的,周一告诉全公司,所有人从那个周一就开始做了——尽管我们当时靠上一个业务还有几百万美元的收入。

Is there anything that you can structurally do to increase the response time of an organization to move faster post pivot?

在结构上有没有什么办法,能让一个组织反应更快、在 pivot 之后跑得更快?

Yeah, I think I think the answer is for a startup, one of my true beliefs is that companies don't succeed because they do many things well.

有,我觉得对创业公司来说,我真正笃信的一点是:公司成功不是因为它很多事都做得好。

they succeed because they do maybe one thing really well.

它们成功,是因为可能只有一件事做得特别好。

Like it's kind of like what I said.

就像我刚才说的那样。

I think having one good idea already is like a miracle for most startups, right?

对大多数创业公司来说,能有一个好想法本身就已经是奇迹了,对吧?

And when you think about it from that perspective, when you think about a company, um, think about two companies, you know, any product has some sort of exponential curve at which it's able to grow, right?

从这个角度想,你去看一家公司,或者拿两家公司来比,任何产品都有一条属于自己的指数增长曲线,对吧?

There's some R value attached to a product.

每个产品都对应着某个 R value。

Uh, the right thing for a startup to do is always focus on the product with a higher RV value.

对创业公司来说,正确的做法永远是把力气压在 R value 更高的那个产品上。

It never makes sense to be diverting your resources to work on two different products that have different exponential growth curves, right?

把资源分散到两条增长曲线完全不同的产品上,永远说不通,对吧?

If they're completely disjoint.

前提是这两件事完全不搭界。

So when you do a pivot, I think the right answer is you should just give up what you were doing in the past because the thing you were doing in the past probably will have no sort of relevance to your business going forward.

所以做 pivot 的时候,我觉得正确答案是直接放弃过去在做的事,因为过去那件事跟你往后的业务大概率毫无关系。

Right?

对吧?

The revenue multiple on that business is not going to look at all like the revenue multiple of the other business.

那门生意的收入倍数,跟另一门生意的收入倍数根本不是一回事。

And I think if you can go cold turkey and actually go and make your organization completely work on the new thing, uh you're more likely to succeed.

我觉得如果你能一刀切,真让整个组织都扑到新方向上,成的概率会更大。

But that's scary, right?

但这挺吓人的,对吧?

You need to tell people, tell your investors, tell your customers, you're no longer going to support that.

你得跟大家说、跟投资人说、跟客户说:那个东西我们不再支持了。

Uh but I think once you make the tough decision, it's really easy.

但我觉得,一旦这个艰难的决定做下去,后面反倒真的很轻松。

Dude, revenue cannibalization when you're doing millions in revenue is hard.

老兄,收入都做到几百万了还去蚕食自己,这可太难了。

So yes, I completely agree with you.

所以是的,我完全同意你。

You said that success is about doing one thing really well and it's hard enough to have one thing.

你说成功在于把一件事做到极致,而光是有这一件事就已经够难了。

you have that, but what thing would you most like to do that you're not doing because you retain that element of focused mindset?

你已经有了,那有什么事是你最想做、却为了守住这份专注而没去做的?

Yeah, I think in our category, maybe if I was to just say this, in our category, a lot of our users of Windsurf are actually people that are non-developers.

嗯,我觉得在我们这个品类里——如果非要我说的话——Windsurf 有很多用户其实不是开发者。

They're people that are vioding apps and they're actually using it for productive use cases.

他们是在 vibe code 应用的人,而且真的用在了正经的生产场景里。

Like internally at our company, we have someone who's a non-developer who leads partnerships that has used our product and built apps to replace a lot of sales tool spend like over $500,000 of sales tool spend internally.

比如我们公司内部,有个负责合作伙伴关系的非开发者,用我们的产品做了几个应用,把大量销售工具的开支替掉了——内部省下 50 多万美元的销售工具开支。

And I think there's a lot of things we can do to make

我觉得我们还有很多事可以做,来让

so they they they've coded 500,000.

所以他们等于是写代码写出了 50 万。

Yeah.

对。

They built apps inside the company for like a partner portal and quoting tool and so on and so forth that these are sales tools that in the past cost hundreds of thousands of dollars a year.

他们在公司内部做了几个应用,比如合作伙伴门户、报价工具等等,而这类销售工具过去一年要花掉几十万美元。

Now the reason they deserve a payriseise phone, right?

这就是他们该涨薪的理由,对吧?

Yeah.

对。

I mean, we should we should get we should consider that.

我是说,这个我们真该考虑一下。

Um, but the reason why that was the case before um not to go on a tangent was because no one really these are very bespoke tools, right?

但以前之所以是那样——不是要扯远——是因为没人真的……这些都是非常定制化的工具,对吧?

These are very bespoke tools that do a singular function and no single company wants to build this out themselves, right?

这些工具非常定制化,只干一件事,没有哪家公司愿意自己把它做出来,对吧?

in the past, but now with software being so cheap to build, like ephemeral software or simpler simpler software being so cheap to build, um you can now have nontechnical people kind of building these apps.

过去是这样,但现在做软件这么便宜,像那种用完即弃的软件、或者更简单的软件,做起来这么便宜,现在非技术的人也能自己搭这些应用了。

And I think Harry, like one interesting avenue for us is how do we make that user much more productive than they are right now, but there's always this tension, right?

所以 Harry,我觉得对我们来说一个挺有意思的方向是:怎么让这类用户比现在高效得多。但这里始终有个张力,对吧?

If we make that user much more productive, we might be giving up the ability to make developers that operate on large code bases much more productive, too.

如果我们把这类用户的效率大幅拉起来,可能就得放弃让那些在大型 codebase 上干活的开发者也大幅提效。

It's just a focus game.

说到底就是个聚焦的取舍。

Is Windsurf about making engineers 10x better or is it about making the normal person able to be an engineer?

Windsurf 到底是要让工程师强 10 倍,还是要让普通人也能当工程师?

Yeah.

嗯。

So I think we are focused on the former.

我觉得我们聚焦的是前者。

Uh we're focused on the former and a consequence of that is the latter right because the technology underneath is capable of doing that but the focus is definitely the former.

我们聚焦前者,而后者是随之而来的结果,对吧,因为底层技术本来就有这个能力,但焦点绝对在前者。

Will you converge over time?

你们会随时间趋同吗?

Will Cursor and you converge to eat the Lovables and Bolts and will the Lovables and Bolts converge to try and eat you?

Cursor 和你们会不会趋同过去、把 Lovable、Bolt 这类产品吃掉,而 Lovable、Bolt 这类又会不会趋同过来,试着把你们吃掉?

You know, I think once once and this is uh this is I think a really common question we get which is that I think those tools that are very focused on like the Lovables, the Bolts that are very focused on the non-developer.

我觉得，这是我们经常被问到的一个问题：像 Lovable、Bolt 这类工具，非常聚焦在非开发者身上。

I think with time they're going to need to build out more and more tunables, more and more ways to configure the product wi with time.

我认为随着时间推移，它们会不得不加进越来越多可调项、越来越多配置产品的方式。

Um but I think I think for us if we can deeply understand code like large code bases it'll naturally fall out that with very little effort non-technical people can build apps.

但我觉得对我们来说，如果我们能深刻理解代码、理解大型 codebase，那自然而然就会带来一个结果：非技术人员花很少力气就能做出应用。

But maybe my point is if we build for that use case right now it's it's we're we're going to be kind of forgoing the ability to help developers.

但我想说的可能是，如果我们现在就冲着那个场景去做，就等于放弃了帮开发者的能力。

But I think it will converge it will converge with time uh between both of these categories.

但我认为这两个品类会随着时间收敛、合流。

Like a company that deeply understands your codebase will be able to with very little natural language build an entire app that is consistent with existing code bases inside the company.

比如，一家能深刻理解你 codebase 的公司，只用很少的自然语言就能造出一整个应用，而且跟公司内部已有的 codebase 保持一致。

If you had infinite resources I'm so enjoying this by the way.

假如你有无限的资源——顺便说一句，这场对话我太享受了。

I'm sorry I had this beautiful schedule but it.

抱歉，我本来排了个漂亮的日程，结果……

Uh if you had infinite resources is there anything that you would do differently to how you're doing it today?

假如你有无限的资源，有什么事你会跟今天的做法不一样？

I think we would be making many more bets.

我觉得我们会押更多注。

We'd be making many more bets inside the company like trying many more things.

我们会在公司内部押更多注，尝试更多东西。

Like as I said probably 50% of the things that we try inside the company or more actually fail right we are actually investing a lot of our product technology on things that are in the future uh and most of it fails actually.

就像我说的，我们在公司内部试的东西大概有 50% 甚至更多其实都失败了；我们把很多产品和技术投入押在面向未来的事情上，其中大部分其实都失败了。

I would like to fail even more but the weird thing about startups is the one thing that works pays for the hundreds of things that fail uh internally at a at a startup.

我其实希望失败得更多，但创业公司奇怪的地方在于：成的那一件，会替内部失败的那上百件买单。

I I spoke to Neil Mehta about stuff that I had to ask you before and he said you do have to ask him about how he builds product and his product building cycle.

我之前问过 Neil Mehta 有哪些是我必须问你的，他说：你一定得问他怎么做产品、他的产品打造周期是怎样的。

you mentioned mentioned there 50% fail the internal cycle.

你刚才提到内部循环里有 50% 会失败。

Can you talk to me about some unconventional ways that you think about internal product building that you think are important?

你能讲讲，在内部做产品这件事上，你有哪些你认为很重要的非常规做法吗？

Yeah, I think there's conventional uh wisdom and you know from other people that a company with more engineers is just going to succeed.

嗯，我觉得有一种从别人那儿听来的传统观念：工程师更多的公司就一定会赢。

But that sort of kind of makes software engineering and just R&D feel like a factory building process.

但那种说法有点把软件工程、把整个研发弄得像工厂流水线。

I think the right way to actually look at it is when a product or an idea has no legs and you're proving it out, you should actually have very few people working on on an idea, right?

我觉得真正正确的看法是：当一个产品或一个想法还没立住、你还在验证它的时候，其实应该只让极少的人做这件事。

Because actually like this is a very common everyone will understand this.

因为这其实非常常见，谁都能理解。

Imagine having 10 people working on something that has not been proven up.

想象一下，让 10 个人一起做一件还没被验证过的事。

Everyone has opinions.

人人都有观点。

Everyone has ideas and nobody's ideas are wrong, right?

人人都有想法，而且谁的想法都不算错。

Because no one has proven anything out.

因为还没有人验证过任何东西。

So it's very hard to get alignment.

所以很难达成一致。

It's very hard for people to work in one direction without causing communication issues.

很难让所有人朝一个方向使劲，又不出沟通问题。

Um instead what is actually really good is can you have a handful of people with an opinionated stance that can go out and prove out an idea and then the question is uh with that if you can if you can go and get a nugget of value and this is like a hard question for any product how do you know when to stop working on it, right?

反过来，真正管用的做法是：找几个有明确主张的人去把一个想法验证出来；然后问题就变成——如果你真拿到了一点价值内核，这对任何产品来说都是个难题：你怎么知道什么时候该停手？

Um and I think there's this interesting like kind of piece of wisdom that that we've built and a lot of us at the company have worked in hard tech before this like companies like autonomous vehicle companies before this is that when you Like it already proves something amazing, right?

我觉得有一条挺有意思的经验是我们自己攒出来的，公司里很多人在这之前做的是硬科技，比如自动驾驶公司，那条经验就是：当你……它其实已经证明了某种了不起的东西。

Uh like for instance, when we built the first working version of like the agent in Windsurf, um the even the crappy version of that was able to do things that we were never able to do in the past like eight or nine months ago, right?

比如说，我们在 Windsurf 里做出 agent 的第一个能跑的版本时，哪怕那个版本很糙，也能做到我们过去、大概八九个月前根本做不到的事。

And we were like, okay, this thing has legs.

我们当时就想：好，这东西站得住。

And then at that point, once you've proven the crappy version has legs, you can then go and resource more uh more people on the project to actually go out and and uh and and pursue it more deeply.

到那一步，一旦你验证了糙版本站得住，就可以往这个项目上加人，去更深地推它。

That no wonder Neil loves you cuz he always talks to me about like the real customer delight moment and that shock and awe that they feel when they have that first agent experience that you mentioned.

难怪 Neil 这么喜欢你——他总跟我讲那种真正让客户惊喜的时刻，就是他们第一次用上 agent 时那种震撼，也就是你刚说的。

Um I totally get that.

我完全懂。

Can I just dig in on a couple of elements before we carry on?

在我们继续之前，我能就几个点再挖一挖吗？

You mentioned like a handful of people.

你提到「几个人的小团队」。

How many is that?

具体是几个人？

How do you structure these teams on new bets?

新押注上，这些团队你怎么搭？

It's it's maybe three or four people.

大概三四个人。

Okay.

好的。

And couple of engineers and a designer.

几个工程师加一个设计师。

Yeah.

是的。

Or or actually in the in the case of if it's just purely like a systems technology or it's like it could be just you know kind of pure engineers that are building this out.

或者说，如果纯粹是系统层的技术，那也可能就是一群纯工程师在做这件事。

One of the cool parts about our product is we are a developer product.

我们产品有个很酷的地方：我们本身就是面向开发者的产品。

So our developers themselves have some intuitions on on a crappy version and how useful it could be at least the v0, right?

所以我们自己的开发者对一个糙版本、对它至少在 v0 阶段能有多大用，是有直觉的。

Because we're building a product for developers.

因为我们就是在给开发者做产品。

Do budget-wise and timewise do you set that, do they set that, how does that work?

预算上、时间上，是你来定还是他们来定，这是怎么运作的？

I don't think we actually think about that and I think that's that's partially because we're in an unconstrained market right now, right?

我觉得我们其实不太想这个，一部分原因是我们现在处在一个不受约束的市场里。

The value of great technology that could accelerate software development is so valuable to our customers that um for us like I guess we largely speaking don't think about it from like a perspective of what is the budget of a particular project.

能加速软件开发的好技术，对我们的客户价值太大了，所以对我们来说，大体上不会从「某个项目预算多少」这个角度去想。

Now what we do look at inside the company is what are the what is the progress we're unlocking in a project over time and whether or not we should decide to move on and work on something else like table something and work on it later.

我们在公司内部真正会看的是：一个项目随时间推进解锁了多少进展，以及我们要不要决定放下它去做别的，比如先搁置、以后再捡起来。

Like that is something that we decide and that's not a democratic process inside the company, it is it is like a little bit of a top-down process inside the company.

这是我们来定的，而且这在公司内部不是个民主过程，多少有点自上而下。

Is there is there anything that you've tabled that with the benefit of hindsight you're like, oh Varun, I should have kept going with that one?

有没有哪件事是你搁置过、事后回头看会想「哦 Varun，这个当初该接着做下去」的？

Yeah, I think I think so.

有，我想是有的。

I think uh an example of something we probably did table was the autocomplete experience which is the first product that we actually built out.

我觉得一个我们大概确实搁置过的例子是 autocomplete 的体验，那是我们最早做出来的产品。

I think we actually delayed on building a much better product experience sooner rather than later.

我觉得我们确实拖了，没有早点把产品体验做得好得多。

Part of the reason why we the product experience wasn't as good actually was because we were stuck in VS Code and VS Code didn't give us the UI capability to make the product much better and once we had Windsurf we were able to invest much more in the capability.

产品体验之所以不够好，部分原因是我们被困在 VS Code 里，而 VS Code 没给我们把产品做得更好所需的 UI 能力；等有了 Windsurf，我们才能在这块投入更多。

You mentioned VS Code there.

你刚才提到了 VS Code。

It brings me to the topic of kind of moats and defensibility.

这就带到护城河和防御性这个话题了。

I'm obviously an ambassador for my sins today as well as being a terrible British podcaster.

我今天显然还挂着个大使的头衔——也算我倒了霉——外加一个糟糕的英国播客主持人。

Um, moats dead in a world of AI.

在 AI 的世界里,护城河已经死了。

You you you're not blind to this.

你对这一点不会视而不见。

Everyone says Cursor, Windsurf, I can switch between the two.

人人都说 Cursor、Windsurf,我随时能在两者之间切换。

Help me understand dead.

跟我讲讲这个「死了」。

And how do you think about that?

你怎么看这件事?

You know, I think startups by and large just this idea of a startup having a moat is usually kind of silly uh for the for the most part.

我觉得初创公司大体上——「初创公司有护城河」这个想法,多数时候其实挺可笑的。

Like the conventional seven powers, you know, like if if anyone in the audience has actually listened, you know, read the book.

比如传统的《七种力量》,如果听众里真有人听过、读过那本书的话。

Um I love it.

那本书我太喜欢了。

Yeah, it's a great I think it's a great book.

是啊,那是本很棒的——我觉得那是本很棒的书。

I think it's just a premature for a startup uh for the most part because okay, fine, you have a company with 50 engineers.

我只是觉得初创公司谈这个,多数时候还为时过早,因为——好吧,就算你有一家 50 人工程团队的公司。

Um you could have the best engineers in the world.

你可以有世界上最好的工程师。

I think we have amazing engineers like like a majority of our engineers are from MIT.

我觉得我们的工程师很了不起,大部分都来自 MIT。

I think we do get the top percentage top few percentage of MIT to come to the company.

MIT 里最顶尖的那几个百分点,我们确实招进了公司。

And um but still like it's still like if we have 50 to 100 engineers, it's still only hundreds of engineering years that have gone into our product.

但即便如此,就算我们有 50 到 100 个工程师,投进产品里的也只有几百个工程师年。

This is not something where ultimately someone else can't build something in this in this as well.

这不是那种别人最终做不出同样东西的领域。

And I think the only moat in our category is speed.

我认为在我们这个品类里,唯一的护城河就是速度。

It's kind of like what I just said.

就是我刚才说的那件事。

It's learning where the dead bodies are, learning how to compound an advantage.

是摸清坑都埋在哪儿,是学会怎么把优势复利起来。

And I think maybe an example of this that I I like to bring up is even a company like Nvidia.

我爱举的一个例子可能是,连 Nvidia 这样的公司也一样。

I think everyone outside looking in is like CUDA is the real moat or something like that.

我觉得外面的人往里看,都以为 CUDA 才是真正的护城河什么的。

Um I think that's just inaccurate, right?

我觉得这根本不准确,对吧?

When you look at large companies like OpenAI, Anthropic, Google, let's say Google if they were buying Nvidia, I know that they build their own TPUs.

你看 OpenAI、Anthropic、Google 这些大公司,比如说 Google 要买 Nvidia——我知道他们自己造 TPU。

They spend tens of billions of dollars on chips.

他们在芯片上花几百亿美元。

Are you telling me if CUDA didn't exist, they would not use Nvidia GPUs?

你是想告诉我,如果 CUDA 不存在,他们就不用 Nvidia 的 GPU 了?

No.

不会的。

They'd find a way to write assembly, the lowest level code, and make that run on the GPUs.

他们会想办法写汇编,写最底层的代码,让它在 GPU 上跑起来。

It's that expensive for them, right?

对他们来说这笔开销就是这么大,对吧?

The reason why people use it is it's just awesome.

大家用它,只是因为它实在太好用了。

They've made it really capable at doing low floating point math.

他们把低精度浮点运算做得非常强。

They've made the interconnect really good.

他们把互联做得非常好。

The computer's really fast and every year they have a ticking time bomb on their head.

算力非常快,而且每年他们头上都悬着一个倒计时。

If they don't make the hardware, the interconnect um kind of the memory bandwidth much much faster every year.

如果他们不让硬件、互联、还有显存带宽每年都快上一大截。

They're going to lose like their profit margin is going to shrink, right?

他们就会失去——他们的利润率就会被压缩,对吧?

And AMD will be on their case, right?

AMD 也会紧咬着不放,对吧?

And that, you know, that's a that's a little bit of a surprising sort of uh sort of feeling like it doesn't have the like Nvidia doesn't have the same properties of a company that a company like Google has, right?

这就有点出乎意料了,感觉 Nvidia 并不具备 Google 那类公司的那些属性,对吧?

Uh but it's still one of the most valuable companies out there.

但它仍然是世上最有价值的公司之一。

I agree totally in terms of the moats for startups like if you have two companies starting start line today.

初创公司的护城河这件事我完全同意,比如今天有两家公司站上同一条起跑线。

But if one of them becomes part of a very large company, say an OpenAI or an Anthropic and they suddenly have distribution and they have brand and they have resources which buys you know chips at scales that are unparalleled compared to what any other startup can do.

但如果其中一家成了某家巨头的一部分,比如 OpenAI 或者 Anthropic,它一下子就有了分发、有了品牌、有了资源,能以其他初创公司望尘莫及的规模买芯片。

Suddenly seven powers becomes increasingly prominent and moats exist.

《七种力量》一下子就越来越显眼,护城河也就成立了。

No, I think the reason why some of this wisdom is potentially not not correct is I think speed is still important in that case, right?

不,我觉得这套说法之所以可能站不住脚,是因为我认为那种情况下速度依然重要,对吧?

And and the reason why is like why does a startup ever win, right?

原因在于,初创公司到底为什么能赢?

Why does a startup ever win in an important category?

初创公司凭什么能在一个重要品类里赢?

Why did Google win in in the early 2000s when Microsoft was already a behemoth at that point?

2000 年代初 Microsoft 已经是个庞然大物了,Google 为什么还能赢?

I think it's just if if a company has good insights and is able to outexecute strategically another company, they will win.

我觉得就是,一家公司只要有好的洞察,又能在战略上把另一家执行掉,它就会赢。

They will win in a in a space.

它就能在一个领域里赢下来。

Um, I would I would argue that distribution is more important than speed.

我倒觉得,分发比速度更重要。

Listen, you're the multi-billion dollar founder, so like please correct me, but I'm like, partnership agreements is really what propelled Google in the early days to be so successful.

听着,你才是那个身价几十亿美元的创始人,尽管纠正我,但我总觉得早期真正把 Google 推上去的,其实是那些合作协议。

But it was also that there were lots of companies in the category and they had the best product and for that reason a lot of people wanted to partner with them.

但也是因为这个品类里公司很多,而他们的产品最好,正因如此才有那么多人想跟他们合作。

Maybe let me put it this way, our space is like very very valuable, I would say.

或许我这么说吧,我们这个领域,我认为非常非常有价值。

But if you actually look at a lot of the hyperscalers, their products in the category are not amazing products, right?

但你要真去看那些超大规模云厂商,他们在这个品类里的产品并不出色,对吧?

Um, even though they have a lot of distribution already.

尽管他们手里已经握着大把分发。

Why is that?

为什么会这样?

And I think it's actually because it is hard to marshall together talent to actually go out and execute in this category very quickly, right?

我觉得其实是因为,要把人才聚拢起来、在这个品类里真正跑得飞快,太难了,对吧?

We're the kind of company where right now we're shipping a major release every 1 to two weeks.

我们是那种每一到两周就发一个大版本的公司。

Name a large company that is able to learn that quickly.

你倒是说出一家能学得这么快的大公司。

Why do large companies suck at velocity of development when the market is as hot as it is today?

市场都火成今天这样了,大公司在开发速度上怎么还这么差?

I think it's existential dread, right?

我觉得是生存焦虑,对吧?

A a startup has existential dread on on whether or not they can survive if they don't ship fast enough.

初创公司背着生存焦虑——发布得不够快,还能不能活下来。

if they don't learn fast enough, right?

学得不够快,还能不能活下来,对吧?

If I'm inside a behemoth, like I'm inside Amazon, right?

如果我待在一个庞然大物里面,比如我在 Amazon 里面,对吧?

And Amazon has a product in our category right now.

Amazon 现在在我们这个品类里也有一款产品。

Um, do you think the employees working on this are worried if they will have a job in 6 months if they don't ship a great product?

你觉得做这产品的员工会担心,交不出好产品的话,6 个月后自己还有没有工作吗?

If we had shipped the same quality product that some of these hyperscalers have shipped, we'd be out of the market a year ago.

我们要是交出这些超大规模云厂商那种质量的产品,一年前就被市场淘汰了。

But they probably still have their jobs right now.

但他们现在多半照样有工作。

It's the seven days a week that's becoming the norm again in Silicon Valley, huh?

所以硅谷现在又把一周干七天当成常态了,是吧?

I mean, we're we're an in-person company actually right now.

我们现在其实是一家线下办公的公司。

Yeah, we're like a 100% in-person company.

是的,我们是 100% 线下办公。

Can you build a company today remote as efficiently as the in-person 7 days a week Silicon Valley specialness that you're seeing return to the valley again?

今天用远程的方式建一家公司,能跟那种一周七天面对面、你正看到它重新回归硅谷的独特氛围一样高效吗?

I think it's just it might just be a little bit harder, right?

我觉得可能就是会难一点吧?

I think you need to be a much more I guess principled company to start with.

我觉得你一开始就得是一家原则性强得多的公司。

But I also think it's it's it's just an unfair advantage if a company is able to get get their engineers kind of in the same room and on the same page like every five minutes if necessary, right?

但我也觉得,一家公司要是能把工程师聚在同一个房间里、必要时每五分钟就把大家的认知拉齐一次,这本身就是不公平的优势。

It's an unfair advantage on speed uh compared to a company where people are on different time zones.

跟人分散在不同时区的公司比,这是速度上的不公平优势。

Like you don't know when you can reach out to someone.

你根本不知道什么时候才能找到某个人。

right now if if if if it's 2 or 3 PM everyone at the company can be called can be kind of like brought into a room at that very moment right so it just it just makes your flexibility a lot faster and in the early days Harry when I was talking about how how obvious it is that your first idea or your ideas are not going to be accurate um you should be able to pivot extremely fast and kind of like marshall everyone at the company as quickly as possible

而现在,下午两三点,公司里每个人都能被叫上、当场就拉进一个房间,所以你的机动性快得多;而且在早期——Harry,我前面讲过,你的第一个想法、你那些想法不会准,这是明摆着的——你得能极快地 pivot,尽可能快地把全公司的人集结起来

do you think in your category brand power is important people are associate with being a Windsurf user or a Cursor user.

你觉得在你们这个品类里,品牌力重要吗?人们会自认是 Windsurf 用户或者 Cursor 用户。

Do you think brand has that same power that it does in other categories?

你觉得品牌在这里跟在别的品类里一样有力量吗?

You know, I I would I would like to obviously I think having a good a large brand is just helpful.

我当然觉得,有一个好的、大的品牌就是有帮助的。

It enables you as you ship more and more new products for more and more people to be able to use it and not have to build that build the user base up from from scratch.

你推出越来越多新产品时,它能让越来越多人用上,不必从零重新攒用户盘子。

But if I just was to remind everyone, you know, kind of like a year or two ago, probably in Silicon Valley, GitHub Copilot was the product that everyone everyone was talking about, right?

但我提醒大家一下,一两年前在硅谷,所有人谈论的产品还是 GitHub Copilot。

Uh at that time and if I was to talk about it a year ago, right, actually a bit over a year ago was the Devin launch.

那时候——要说一年前,其实是一年多一点之前,是 Devin 发布。

And I think for 3 to 6 months, I was not hearing anything but Devin uh from from everyone, right?

我觉得有三到六个月,我从每个人嘴里听到的除了 Devin 还是 Devin。

And I think it's like in our category, we need to prove ourselves almost every day.

而我觉得在我们这个品类里,几乎每天都得证明自己。

If in three months we don't ship something amazing to our users, we will become irrelevant, right?

要是三个月里我们没给用户交付出点了不起的东西,我们就会变得无关紧要。

Uh to to a lot of them, right?

对其中很多人来说就是这样。

If we don't actually like keep up with with the pace of innovation, we will become irrelevant.

只要跟不上创新的节奏,我们就会变得无关紧要。

So I think it's it's a little bit of if you can maintain if you can continue to innovate at breakneck pace, the brand is definitely helpful, but I don't think the brand gives you the right to move slower as a company.

所以我觉得,只要你能一直以极快的速度创新,品牌肯定有帮助,但我不认为品牌给了公司放慢脚步的权利。

But should these businesses not be valued inherently less or at a much lower multiple if they're sustenance of value is able to deteriorate it more quickly than ever before.

但如果这些生意维系价值的能力比以往任何时候都衰减得更快,它们本来不就该估得更低、或者按低得多的倍数估值吗?

As you mentioned Devin, as you mentioned, GitHub Copilot, brilliant examples.

你提到的 Devin、你提到的 GitHub Copilot,都是绝好的例子。

Yeah.

是的。

And then suddenly have a reduction of value so fast.

然后价值突然就缩水得这么快。

Should they be valued in the same way?

还该用同样的方式给它们估值吗?

So, I think there are still qualities of our company that we haven't really talked about of like having a large enterprise business, being wall-to-wall at very large Fortune 500 companies, right?

我觉得我们公司还有些特质我们还没真聊到,比如有一块很大的企业客户业务,在超大型财富 500 强公司里做到全员铺开。

and being able to work properly for large teams which has inherent switching costs for the business right not the same class of switching costs as a as some of the products like I don't know Salesforce yet I don't think we built that out

以及能真正服务好大团队,这对企业本身就带有切换成本——跟 Salesforce 那类产品的切换成本还不是一个量级,我不认为我们已经做到那一步

there's a trade-off where if you focus so hard on making your product to switch off of if you also don't improve your product fast enough people people will just switch on to the onto the better product despite the fact that it is hard to switch off of your product right

这里有个取舍:如果你太用力地让产品变得难以被替换,却没有把产品改进得足够快,那就算切走很难,人们照样会切到更好的产品上去

I always remember seeing a tweet from Tom Blomfield who's a friend of mine.

我一直记得看过 Tom Blomfield 的一条推文,他是我朋友。

he was like I don't really get this space.

他说,我不太看得懂这个领域。

I was a Cursor user and I switched to Windsurf and it was like better or just as good or like he switched to Windsurf for this one project.

我原来用 Cursor,后来换到了 Windsurf,感觉更好、或者至少一样好——或者说,他是为某一个项目换到了 Windsurf。

Um I have no idea if he stayed.

我不知道他后来有没有留下来。

I'm not making a comment.

我不是在下判断。

Tom, don't kill me.

Tom,别宰我。

Um but um my question to you was like how do you think about switching cost and retention and is it that easy to switch between the two?

但我想问你的是:你怎么看切换成本和留存?在这两个产品之间切换真有那么容易吗?

We need to innovate on features and product fast enough that we're able to build differentiated experiences for for folks.

我们得在功能和产品上创新得足够快,才能给大家做出差异化的体验。

That actually takes a lot of work, right?

这其实要花很大功夫。

Like as I was saying, like most of our products that we end up building don't end up working.

就像我刚才说的,我们做出来的产品大部分最后都没成。

And even though we're shipping releases every couple weeks, some of the more complex pieces of technology that we've shipped, they take months to actually ship.

而且虽然我们每两周就发一版,我们发出去的一些更复杂的技术,真要做出来得好几个月。

So I'll give you an example.

我给你举个例子。

A couple weeks ago, we shipped a model that was comparable to some of the frontier models at the agentic workload.

几周前我们发布了一个模型,在 agentic 工作负载上跟一些前沿模型不相上下。

And our model SWE-1 is much faster, much cheaper for us to run and capable of operating over a large codebase as well.

我们这个模型 SWE-1 快得多,我们自己跑起来也便宜得多,而且一样能在大型 codebase 上作业。

And we were able to do that by learning from our users on how they use the product internally, right?

我们能做到,是靠从用户那里学他们内部怎么用这个产品。

And that was not a single month kind of experience for us.

这对我们来说不是一个月就能搞定的事。

It took us many months to actually build out that that model internally.

我们花了好几个月才在内部把这个模型做出来。

But these are the kinds of and now it processes hundreds of billions of tokens of code a day.

但这类事情就是——现在它每天处理几千亿 token 的代码。

That single model, right?

就这一个模型。

And it's running on our own GPUs.

而且它跑在我们自己的 GPU 上。

Um, but this is like one of those things that it takes time to kind of build out these capabilities.

但这就属于那种要花时间才能攒出来的能力。

uh just like how maybe you know for everyone in the audience and folks like Tom he probably never heard of our company 10 months ago and used our product 10 months ago right

就好比对在座各位、对 Tom 这样的人来说,他 10 个月前大概压根没听说过我们公司,10 个月前也没用过我们的产品

so I think these ch these things change extremely quickly but you're totally right the fact that it can change this quickly means that you need to be ultra vigilant right

所以我觉得这些东西变化极快;但你说得完全对,正因为它能变这么快,你才得极度警惕

the same reason why no one had heard of us in Silicon Valley like maybe 10 12 months ago and now like a lot of folks have heard of us um the same thing can happen in the other way too

同样的道理:大概 10、12 个月前硅谷没人听说过我们,现在很多人都听说了——反过来的事一样会发生

you mentioned the strength of enterprise in your business how much of your revenue/ usage today is enterprise versus um bottoms up.

你提到企业客户是你们业务的强项,今天你们的收入和使用量里,企业客户占多少,自下而上占多少?

Yeah.

嗯。

So I would say like probably of our total amount it's like over 50% is is is actually is actually enterprise.

我会说,在我们的总量里,大概超过 50% 其实来自企业客户。

What are the biggest questions lessons things that enterprises care about when they buy you?

企业客户买你们的时候,最大的问题、最大的心得、他们最在意的是什么?

I think it's there's a maybe a handful of things.

我觉得大概有那么几件事。

I I think the first thing is a lot of enterprises and large companies this is something that's not very obvious to folks outside looking in.

我觉得第一件是,很多企业客户和大公司——这点从外面看进来的人不太容易看清。

They have a lot of Java developers, right?

他们有很多 Java 开发者。

And for the Java developers, actually most most large companies have a lot of code in Java.

而说到 Java 开发者,其实大多数大公司都有大量 Java 代码。

They use an IDE called IntelliJ as part of the JetBrains family.

他们用的 IDE 叫 IntelliJ,属于 JetBrains 家族。

So this is like an interesting fact.

所以这是个挺有意思的事实。

Even though we have the Windsurf editor, we still support plugins into all the JetBrains ids and it has the same functionality as the Windsurf editor.

虽然我们有 Windsurf 编辑器,但我们照样为所有 JetBrains IDE 提供 plugin,功能跟 Windsurf 编辑器一样。

The reason why we forked the VS Code uh editor was because we didn't have enough flexibility to build out the UI that we wanted uh to give out the new agentic experience.

我们之所以 fork VS Code 编辑器,是因为灵活度不够,做不出我们想要的 UI 来承载新的 agentic 体验。

But we were able to give that out entirely in JetBrains in the plugins.

但在 JetBrains 里,我们靠 plugin 把那套东西完整地交付出来了。

And I'll give you an example.

我给你举个例子。

JP Morgan Chase, who is a customer of ours, over 50% of their developers are JetBrains users inside the company.

JP Morgan Chase 是我们的客户,公司内部超过 50% 的开发者用 JetBrains。

And for those people, we're able to support all of their developers.

而对这些人,我们能支持他们全部的开发者。

We're not a product that's only for their VS Code users, right?

我们不是一个只服务他们 VS Code 用户的产品。

Uh so they can move over to Windsurf.

所以他们能整体迁到 Windsurf 上来。

So that's again an interesting fact that I think outside looking at people don't really understand that for the enterprise there are lots of these IntelliJ users that we need to satisfy

所以这又是个有意思的事实,我觉得外面的人不太明白:企业客户那边有大量 IntelliJ 用户是我们必须照顾到的

and when we go into an enterprise we don't want to tell them hey only 40% of your users can use our product we want to make sure all of their developers can use use our product

所以我们进一家企业客户时,不想跟他们说你们只有 40% 的人能用我们的产品——我们要确保他们所有开发者都能用上

do you think your customers will have more or less engineers in 5 years time

你觉得五年后,你的客户会有更多还是更少的工程师?

it depends on the type of customer right and and the word engineer is going to become a very flexible word as I was saying right as the technology gets better and better people that are technical adjacent are going to using Windsurf as well

这取决于客户是哪一类,而且像我刚才说的,"工程师"这个词会变得非常有弹性;随着技术越来越好,技术相邻岗位的人也会开始用 Windsurf

and we're already seeing this by the way at some of the companies.

顺便说一句,我们在一些公司里已经看到这种情况了。

So what does engineer in 5 years time mean?

那五年后的"工程师"意味着什么?

If I was to explain it to an engineer, if you were to go over time just for an engineer, right?

如果我要跟工程师解释,你就顺着时间线看看工程师这个职业。

If you go back 50 years, everyone is writing assembly.

如果回到 50 年前,所有人都在写汇编。

Then after that, they're writing C.

在那之后,大家写 C。

And by the way, people still write assembly though, they after that they write C.

顺便说一句,今天还是有人写汇编的,但那之后大家写 C。

Then it's like C++, then Java and Python.

然后是 C++,再是 Java 和 Python。

And then now a lot of people writing JavaScript, right?

而现在很多人在写 JavaScript,对吧?

And I'm just giving you an example on like ease of programming has gone up, but a lot of the people that are that write JavaScript have no idea how to write assembly at this point.

我只是举个例子:编程越来越容易,但今天很多写 JavaScript 的人根本不知道汇编怎么写。

So I think it's going to be something along those lines, right?

所以我觉得未来大概也是这个路数。

There will be people that are able to operate purely on a natural language based abstraction, right?

会有一批人纯粹在基于自然语言的抽象层上干活。

And the AI tools are able to explain to them how different components of the software stack work.

而 AI 工具能给他们讲清楚软件栈各个组件是怎么运转的。

So that just even though they're operating on top of natural language, they're able to still be valuable to the company.

这样,哪怕他们只在自然语言之上操作,照样能给公司创造价值。

But then ultimately down the line, there will always be production critical applications.

但往后走,终归永远会有生产关键的应用。

Let me give you an example.

我给你举个例子。

Let's say I'm JP Morgan Chase and I'm building my transaction processing system.

假设我是 JP Morgan Chase,在搭自己的交易处理系统。

You probably don't want to vibe code then, right?

这种时候你大概不会想 vibe code 吧?

like Harry, if you give me a $100, probably you don't want to like lose a hundred and for me to not get the $100, right?

比如 Harry,你给我 100 美元,你大概不希望这 100 块丢了、我也没收到,对吧?

I'm going to complain.

我会投诉的。

And that's some transaction processing system that's running in JP Morgan Chase that is probably doing millions of transactions or billions of transactions a day, right?

而那套跑在 JP Morgan Chase 里的交易处理系统,一天大概要处理几百万笔、甚至几十亿笔交易。

And for that, you probably want someone that can go all the way down to the weeds and validate that this actually makes sense because that's core to the way the business operates.

对这种系统,你多半希望有人能一路下到底层去,验证这东西真的说得通,因为那是这门生意运转的核心。

So, I think you're going to have a spectrum of people, right?

所以我觉得,人会分布成一个光谱。

people that can operate purely on top of natural language and they're building capabilities and technology and and apps inside companies, but then people also that can go to the weeds go down to the weeds.

有的人纯粹在自然语言之上操作,在公司内部做出各种能力、技术和应用;也有的人能钻进细节、一直下到底层。

Um, now how many people need to go down to the weeds?

那需要多少人能下到底层?

Probably fewer than there are today.

大概比今天要少。

Do we have PMs?

还会有 PM 吗?

I'm I'm always told that PMs are CEOs of the product and many have said that actually we're not going to have PMs anymore.

我总听人说 PM 是产品的 CEO,可也有很多人说,我们以后根本不会再有 PM。

ElevenLabs I had that the head of growth on the show the other day was like we don't have PMs.

ElevenLabs,前几天他们的增长负责人上我节目,他就说我们没有 PM。

Do PMs have a role in 5 years?

五年后 PM 还有位置吗?

I think the expectation for PMs is not going to be down the line.

我认为往后走,对 PM 的期待不会再是原来那样。

It's not going to be I tell someone to go and build something.

不再是我让某个人去把东西做出来。

It's going to be you should go out and build it too, right?

而是你自己也得去把它做出来。

Uh there will be need to be a lot more agency on their side.

他们那边需要有强得多的能动性。

But the benefit is these AI tools are going to give them that capability.

但好处是,这些 AI 工具会给他们这个能力。

Right?

对吧?

Let's say you do have a technical PM that is that understands code and that can write code.

假设你手上真有一个技术型 PM,懂代码、也能写代码。

They are even more deadly now than they were in the past.

他们现在的杀伤力比过去大得多。

You don't need to give them a team of 10 people to prove out their idea, right?

你不用再给他们配十个人的团队去验证想法。

And one of the things that I kind of don't like about PMs probably in in uh nowadays in a lot of cases is they spend so much time writing docs to try to convince the organization to do what they what needs to be done.

而我现在挺不喜欢 PM 的一点是,很多时候他们花大把时间写文档,去说服组织做那些本来就该做的事。

But instead they should just go and build a version of what they want to get done and then prove to the organization that that's what needs to get done.

但他们其实应该直接把想做的东西做出一个版本,再拿它向组织证明这就是该做的事。

Why do you need to spend so much time babying your organization playing a playing a game of thrones, right?

何必花那么多时间哄着组织、玩一场权力的游戏呢?

It doesn't really make a ton of sense in the future.

这在未来实在不太说得通。

Did we skip the design stage and go straight to prototyping in the future?

未来我们会跳过设计阶段,直接上手做 prototype 吗?

Probably the ability to very quickly mock up a design that is consistent with the design inside your company and prototype it is going to be much faster, right?

很可能,快速做出一版跟公司内部设计一致的稿子、再把它做成 prototype,这件事会快得多。

Like internally right now when we build apps, we're not very very quickly going to Figma at the very beginning if the app can be built very fast uh inside.

比如我们内部现在做应用,如果这个应用在内部能很快搭出来,我们不会一上来就急着跑去 Figma。

And then obviously if it's core to our product experience, we're going to go out and design the layout and make sure it's it's it's it's to our standards.

然后显然,如果它是我们产品体验的核心,我们还是会去设计布局,确保它达到我们的标准。

Uh but rapid prototyping can it can we can definitely skip like a laborious design stage right now.

但靠快速原型,现在确实可以跳过费力的设计阶段。

I'm just fascinated like where where all of it converges together.

我特别好奇,这一切最后会在哪里汇到一起。

As we said, you meet the Lovables and the Bolts in the middle and then the Figmas are coming in now and actually saying hey turn design into code and kind of coming in from outside.

就像我们说的,你们在中间跟 Lovable、Bolt 这类产品碰上,现在 Figma 这类也进来了,说嘿,把设计变成代码,算是从外面往里切。

I'm trying to figure out what it this kind of mess looks like in five years time respectfully, you know.

恕我直言,我在琢磨五年后这一团乱麻会是什么样子。

Yeah, I think I think it's a fair question.

是啊,我觉得这问题问得在理。

Maybe one thing is for people outside looking in the things that the Lovables and Figmas do is like maybe a very small fraction of what software is.

也许有一点:对外面看进来的人来说,Lovable、Figma 这类产品做的事,可能只占软件极小的一块。

Uh like ultimately like most software is not just like a website that that has like some backend attached to it, right?

毕竟大多数软件并不只是一个挂了后端的网站。

It's there's complex database systems uh software out there.

外面还有复杂的数据库系统、各种软件。

There's all of this infrastructure.

还有这一整套基础设施。

Most of Google is not the front end, right, of Google.

Google 的绝大部分并不是它的前端。

Like think about how much work goes into Google.

想想 Google 里面投进去了多少工作量。

Google's probably one of the most complex pieces of technology.

Google 大概是世上最复杂的技术产物之一。

There's a very complex re-ranker.

里面有一个非常复杂的重排序模块。

There's this database called Spanner.

有一个叫 Spanner 的数据库。

That's a geographically consistent database um that that exists that's probably tens of millions of lines of code, right?

那是一个地理一致的数据库,大概有几千万行代码。

And all of those things, there's no one that's designing something in Figma for modifying the Spanner codebase, right, inside the company.

而这些东西,公司里没有谁会为了改 Spanner 的 codebase 先去 Figma 里画个设计。

But we would like it so that they're using Windsurf to actually go out and do that.

但我们希望的是,他们用 Windsurf 去把这件事真正做掉。

So the way we sort of like to look at it is like I don't we don't sit and think about what these other companies are doing all the time.

所以我们看这事的角度是,我们不会整天坐着琢磨别的公司在干什么。

We're sort of thinking about what are all the ways in which software engineers spend their time and how do we make them much more efficient.

我们想的是,软件工程师的时间都花在哪儿,我们怎么让他们效率高得多。

So you're totally right that I think the design stage a lot of things are going to get disrupted inside.

所以你说得完全对,设计阶段里很多东西都会被颠覆。

But maybe one point I just wanted to add is that's maybe just the tip of the iceberg on what a software developer ultimately does.

但我想补一句:那可能只是软件开发者最终要做的事情的冰山一角。

I was going through some of your tweets as well and you said async remote agents will happen and I just wanted to double click on that first.

我也翻了你的一些推文,你说异步的远程 agent 一定会出现,我想先就这点深挖一下。

Can you expand on how you think about this and which tasks will be automated first?

你能展开讲讲你怎么看这件事,以及哪些任务会最先被自动化吗?

So the idea of async remote agents if you were to look at a product like Windsurf Windsurf is actually a local agent that runs runs locally and it can do very longunning tasks like you can migrate an entire codebase from one version to the next and it should be able to do that without you having to touch your keyboard right once you once you instruct it to which is very very powerful but this idea of being able to do it asynchronously is also very powerful.

说到异步远程 agent,你看 Windsurf 这样的产品,它其实是个本地 agent,跑在本地,能干很长的活儿,比如把整个 codebase 从一个版本迁移到下一个版本,你只要下了指令,它就该能做完,全程不用碰键盘,这非常非常强大;而能异步地做这件事,同样非常强大。

You can imagine like just firing something off from your phone and it just generating a like a pull request for you.

你可以想象,就从手机上把任务丢出去,它直接给你生成一个 pull request。

But I think one of the complexities of this is there's a building products in our category is I think there are three things that really are important right uh there's latency so how long it takes to get the response back there's quality right which is what is the what is the correctness right and then correctability how quickly can you make a change right and I think people kind of feel like if the technology is really good they can kind of ignore these facts but it's really really important if you build an asynchronous exper experience and it takes 10 20 30 maybe a couple hours to go out and build something, people's expectations on the output are going to be really high.

但我觉得复杂之处在于,在我们这个品类里做产品,真正重要的其实是三件事:一是延迟,拿到回应要多久;二是质量,也就是对不对;然后是可纠正性,你能多快把它改掉。人们好像觉得只要技术足够好,这些就可以不管,但它们真的非常重要——如果你做的是异步体验,要花 10 分钟、20 分钟、30 分钟甚至几个小时才交出点东西,人们对结果的预期就会非常高。

Your quality has to be really, really high.

质量必须非常非常高。

And if the quality is not high, it better be easy to correct.

而如果质量不高,那它最好容易纠正。

But it's not very easy to correct.

但它并不容易纠正。

Like when when the workflow of how people build software, if all I do at the very end is I create a pull request on GitHub, which is the final way that people review the code, and even 20% of it is wrong, let alone 10%.

看看大家做软件的工作流:如果我到最后只是在 GitHub 上开一个 pull request——那是大家审代码的最终环节——而里面哪怕有 20% 是错的,更别说 10% 了。

I I wrote it has to be at least 90% correct.

我写过,它必须至少 90% 正确。

People were like it needs to be 99% correct because and I think they I think they might be right because if even 10% is wrong people lose faith in the product because they waited hours to get the result right and they need to bring it back in locally.

有人说得 99% 正确才行,我觉得他们可能是对的,因为哪怕只有 10% 是错的,人们也会对产品失去信心——他们等了好几个小时才拿到结果,还得再把它拉回本地。

And modifying something that is 90% correct is not that easy because you need to understand the 10% that is wrong also which takes time too right.

而改一个 90% 正确的东西并不容易,因为你还得搞清楚错的那 10% 错在哪,这同样要花时间。

You're still human time limited at that point.

到那一步,你还是卡在人的时间上。

So my rough feeling on things is what are going to be done asynchronously remotely are things that where the task is so easy that it can actually get it done properly.

所以我大致的感觉是,能异步远程做完的,是那些简单到它真能做对的任务。

Like the complex tasks are still going to get done in Windsurf in the in the short term, right?

复杂的任务短期内还是会在 Windsurf 里完成。

Um just because things that are complex and require a lot of rapid iteration, it's unlikely that you're going to do something asynchronously, wait 30 minutes, go and look at it in GitHub and then decide, hey, it's not correct and bring it back.

只是因为复杂、又需要大量快速迭代的事情,你不太可能异步发出去、等 30 分钟、去 GitHub 看一眼,再判断说这不对,又把它拉回来。

This the feedback loop is too slow at that point.

到那时,反馈闭环就太慢了。

To what extent do users care about latency have you found?

你发现用户到底有多在意延迟?

Very very much.

非常非常在意。

I'll give you an example.

我给你举个例子。

Like even for a tab product 10 milliseconds which is like the product that gives you not only the autocomplete but but the automatic refactors as the user is typing every 10 milliseconds affects a percentage points in acceptance rate right now for the product.

就算是 tab 这类产品,10 毫秒——就是那个不只给你 autocomplete、还在你打字时自动帮你重构的产品——每 10 毫秒都会影响这个产品现在采纳率上的几个百分点。

When we think about these async remote agents that have to be incredibly responsive that have to be correctable what form factor do they take?

说到这些既要响应极快、又要可纠正的异步远程 agent,它们会是什么形态?

We mentioned that earlier like just pinging off a message.

我们前面提到过,比如就丢一条消息过去。

How do you envision the right form factor and what that will take?

你设想中对的形态是什么样?那又需要什么?

Yeah, I think it's it's it's not super clear, right?

我觉得这事还不是特别清楚。

It's possible that the right form factor is you need to fire this off from your IDE because people are people are in their work state in their IDE.

有可能对的形态是你得从 IDE 里把它发出去,因为人的工作状态就在 IDE 里。

Now, the question is how how common is the use case that people want to fire things off from their mobile phone, right?

现在的问题是,想从手机上发起任务这个场景到底有多常见?

And then if they do, is it fine for it to just be on Slack?

真要这么做,那放在 Slack 上就够了吗?

Do people want to continuously talk to something on their mobile phone?

人们想在手机上不停地跟一个东西聊下去吗?

I think the answer is maybe not because if I get thousands of lines of ch code changes, I don't know how to review code very effectively on my phone and then in which case like I don't want to interact with something on my phone back and forth.

我觉得答案可能是不想,因为要是给我几千行代码改动,我不知道怎么在手机上很有效地审代码,那样的话我就不想在手机上来回折腾。

So maybe I'm I'm game to do one and done interactions on my phone and then it better be 100% correct and I'll go back to my laptop and be like no it's correct and then stamp stamp it and then approve it and merge the code.

所以我也许愿意在手机上做一次性搞定的互动,那它最好 100% 正确;然后我回到笔记本上一看,对,是对的,盖个章、批准、把代码合了。

So it's there's a lot of trade-offs here.

所以这里有很多取舍。

When we talk about the right form factor for these remote async agents, it kind of makes me think about something that Satya said, and I I did write it down because I wanted to make sure that I got it right because I didn't want to misquote Satya, but he essentially said that obviously apps will collapse into agents and become just databases with business logics.

说到这些远程异步 agent 该是什么形态,这让我想起 Satya 说过的一句话——我特意记了下来,就是想说准,不想误引 Satya——他的意思基本是,应用显然会坍缩成 agent,变成只是带着业务逻辑的数据库。

Do you agree with him in that respect?

这一点上你同意他吗?

Salesforce becomes a database that agents just crawl all over, a HubSpot the same, and you basically have their utility value reduced to just databases.

Salesforce 变成一个 agent 到处爬的数据库,HubSpot 也一样,它们的效用价值基本上就被压缩成了数据库。

Yeah, I mean I'm I'm trying to think about why I think that's that is not exactly true, but you're totally right that I mean, you know, for any company, their entire state could could be distilled down to a database, right?

我在想,为什么我觉得这话不完全对——不过你说得完全没错,任何一家公司的全部状态,确实都能提炼成一个数据库。

Uh for for any company.

任何公司都是。

I don't know if I completely buy that in the short term, these complex applications like Salesforce and Workday completely get replaced by another company that just builds a database and an agent operating on top of it.

我不确定自己完全认同短期内 Salesforce、Workday 这类复杂应用会被一家只做数据库、再加一个 agent 在上面跑的公司彻底取代。

I think a lot of complex workflow has been built around the idea of using Salesforce.

我觉得很多复杂工作流,是围着“用 Salesforce”这件事长出来的。

So, I'll give you an example.

我给你举个例子。

When we built our sales team, everyone already knows how to use Salesforce down the line and now we built custom workflows inside Salesforce.

我们组建销售团队的时候,后面进来的每个人早就都会用 Salesforce,现在我们又在 Salesforce 里搭了定制工作流。

And you're right, those are all database transitions, but they affect the way humans operate with software.

你说得对,那些本质上都是数据库上的变更,但它们影响的是人怎么跟软件打交道。

It's very unlikely to me that what's going to happen is Salesforce will get dethroned by a company that is just a database with an agent operating on top of it.

在我看来极不可能发生的是,Salesforce 被一家只是数据库加一个 agent 在上面跑的公司掀下王座。

It just there's too much inertia for the company uh building out these complex workflows.

公司在搭这些复杂工作流上的惯性实在太大了。

And maybe here's the the final piece and maybe why I believe that I don't think agents are good enough today that you can kind of just let them operate on top of on top of a database and write to the database automatically write to the database uh without human supervision for some arbitrary workflow.

也许这是最后一块,也是我这么想的原因:我不认为今天的 agent 已经好到可以放手让它们在数据库上跑、自动往数据库里写,在没有人工监督的情况下去执行某个任意的工作流。

Right?

对吧?

Right now we're still in the stage right now where these agents are mostly reading from data sources.

眼下我们还处在这些 agent 主要只是从数据源读取的阶段。

They're not like modifying things at scale.

它们并没有在大规模地改东西。

Um, and that's just because maybe we don't have the tr the the we don't we don't trust these systems uh deeply enough right now.

而这只是因为,也许我们现在还没有足够信任这些系统。

Do you think we are overly optimistic about the progression of the agent landscape?

你觉得我们对 agent 这块的进展是不是太乐观了?

Obviously, as a venture investor, the only thing we give a about today is agents.

显然,作为风险投资人,我们今天唯一在乎的就是 agent。

Um, do you think we are overly optimistic about where we are?

你觉得我们对眼下所处的位置是不是太乐观了?

This is going to be the the most obvious statement, but I think people think these systems are way more capable than they like a lot of investors might believe they're more capable than they are um today.

这会是句最显而易见的话,但我觉得人们把这些系统想得比实际强太多——很多投资人可能觉得它们比今天的实际能力更强。

I'll give you an example.

我给你举个例子。

My goal here is not to not to, you know, kind of downplay anything here, but I remember last last March when kind of the Devin launch came out.

我不是想贬低什么,但我记得去年三月 Devin 发布的时候。

Everyone was giddy with excitement of the idea of like software engineering is getting replaced.

所有人都为“软件工程要被取代了”这个念头兴奋得不行。

Uh like a bunch of investors are like, "Oh, there's no reason to ever hire junior developer. You will have this. "

一堆投资人说:“哦,再也没必要招初级开发者了,有这个就行。”

And now we're very far from that right now, right?

而我们现在离那还远得很。

I don't think like or at the time we were very far from actually replacing a junior developer with with an asynchronous AI agent, right?

我不觉得——或者说在当时,我们离用一个异步 AI agent 真正取代一个初级开发者,还差得很远。

But I think what people don't understand is even if the technology is not there yet, I don't think they grasp how quickly the exponentials are improving, right?

但我觉得人们没搞明白的是,就算技术还没到位,他们也把握不住这些指数曲线改进得有多快。

Which is that in six months what is capable of these models is going to be very different than what is capable today.

也就是说,六个月后这些模型能做的事,会跟今天能做的非常不一样。

And I think people are it's it's very hard for people to understand that.

我觉得这一点大家真的很难理解。

Like what I just said about agents and the fact that they're really good at reading systems at scale, but people don't trust them to make rights to systems to internal databases at scale.

就像我刚说的 agent 那件事:它们很擅长大规模读取系统,但人们不信任它们大规模地往系统、往内部数据库里写。

That will probably change in six months, right?

这在六个月内很可能就变了。

uh because eight or nine months ago people were not using agents a ton.

因为八九个月前,人们还没怎么用 agent。

So yeah, I think people underestimate the speed at which these things improve.

所以是的,我觉得人们低估了这些东西变强的速度。

What do you think in 12 months will be crazy that we don't have today?

你觉得 12 个月后会有什么我们今天还没有、听上去很疯狂的东西?

Like you said there about kind of the transition and models bluntly improving so fast in such short time frames.

就像你刚说的那个转变,模型在这么短的时间里,说白了进步得太快了。

If you think about what we do today that we would consider crazy that will be a reality in 12 months, what do you think it would be?

今天我们会觉得疯狂、12 个月后却会成为现实的事,你觉得会是什么?

I think a lot of things that maybe I understand the software development uh landscape a lot better um is the way how end these tools are going to be is going to be I think shocking to people.

我觉得很多事情吧——可能因为我对软件开发这个领域了解得更深一些——这些工具会做到多么端到端,我觉得会让人震惊。

I think right now if I was to say one of the biggest problems of these tools is mostly they help in the code writing experience.

现在这些工具最大的问题之一,我觉得是它们主要只在写代码这个环节上帮忙。

they're not helping in a lot of different aspects of the of the software development life cycle which is software developers design software right they deploy software review software right uh build navigate they do a lot of debug test they do a lot of different things and I think the reason why it's mostly been kind of siloed to one or two areas is actually when when you look at a software engineer and their job at a company they look at a lot of distinct data sources they look at like a logging system they might look at a database they might do all these other things and I think these agents are going to access to all of this data and all of this data, your browser data, all this stuff.

它们在软件开发生命周期的很多其他环节上帮不上忙——软件开发者要设计软件,要 deploy 软件、评审软件,要构建、要在代码里导航,还要做大量 debug、测试,做很多不同的事;我觉得目前它基本被圈在一两个环节里,原因其实是:你看一个软件工程师在公司里的活儿,他要盯很多不同的数据源,要看 logging 系统,可能要看数据库,可能还要做各种别的事;而我认为这些 agent 会拿到所有这些数据——你的浏览器数据,所有这些东西。

And I think what's going to be possible is debugging very complex tasks.

我觉得接下来能做到的,是 debug 非常复杂的任务。

Designing systems is going to be something that these agents are going to give much more leverage for than they are today.

设计系统这件事上,这些 agent 给出的杠杆会比今天大得多。

And I think this idea that software engineering, there's like a part of software engineering that AI is not going to touch because it's too hard, I think is is just going to be wrong.

而「软件工程里有一块是 AI 碰不了的,因为太难」这种想法,我觉得就是错的。

I think every aspect is going to become 10 times more effective.

我觉得每个环节都会变得高效 10 倍。

Can I ask you a really hard one?

我能问个特别难的问题吗?

You mentioned Devin there.

你刚提到了 Devin。

If you look at say a Cursor, what have they done well that you've learned from?

拿 Cursor 来说,他们有哪些做得好、你从中学到了东西?

And what do you think they've maybe made a mistake on that you've learned from?

又有哪些你觉得他们可能做错了、你也从中学到了东西?

I don't know.

我不知道。

I don't know about a mistake, but I think for them just actually I think they took a really good approach on building high quality UIUX.

说错误我不好讲,但我觉得他们在打磨高质量 UI/UX 上走的路子确实非常好。

Um, and that actually proved to be uh what people really loved, right?

而这一点后来证明恰恰是人们真正喜欢的,对吧?

And I think that's that's like very impressive that they were that they that they were able to do this right off the bat.

我觉得挺让人佩服的,他们一上来就做到了这一点。

I think that's that's that's not actually what our initial intuition was.

这其实不是我们最初的直觉。

Um and I think they they executed wrong.

我觉得他们执行得不对。

What do you mean by that?

你这话是什么意思?

It wasn't where you placed importance in terms of that initial UIUX being seamless.

是说最初的 UI/UX 做到丝滑这件事,不是你们当时的重点?

Yeah, I think for us to go out and build Windsurf, we actually felt that we needed a massive like technical breakthrough of like actually a new product experience.

是啊,我觉得对我们来说,要做出 Windsurf,我们当时真觉得需要一个巨大的技术突破,一种真正全新的产品体验。

And I think that was the only reason why we actually decided to go out and build like actually have our own ID.

我觉得那是我们最后决定去真做一个自己的 IDE 的唯一原因。

I like if we didn't do that, I think we're the type of company that would not just go out and build UIUX just for the like, you know, for a much better experience by default unless we had great technology.

如果不是那样,我觉得我们这类公司不会只为了默认体验好很多就去做 UI/UX,除非我们手里有很强的技术。

I like to think of ourselves as like we're a technology company and I think we ship product to maximize the amount of technology that our users can consume productively, right?

我更愿意把我们看成一家技术公司,我们发产品是为了让用户能高效消化掉尽可能多的技术,对吧?

Um, and we want UIUX that is not harming people's experience of our of our product.

而我们要的 UI/UX,是不损害用户使用我们产品体验的那种。

To what extent is the quality of your customer experience defined by the software that you create, the UI, the UX versus the model progression that lies beneath it being OpenAI Anthropic?

你们的客户体验质量,有多少是由你们自己做的软件、UI、UX 决定的,又有多少取决于底下的模型进步,也就是 OpenAI、Anthropic?

Yeah, I think I think you can't downplay the fact that that the latter is very helpful, right?

是啊,我觉得你没法淡化这一点:后者非常有帮助,对吧?

The frontier model is getting cap getting more and more capable.

前沿模型的能力越来越强。

Every new capability unlocks maybe a new type of product experience that wasn't capable in the past.

每多一项新能力,可能就解锁一种过去做不到的新产品体验。

But I think it's a little bit of both, right?

但我觉得两边都占一点,对吧?

Um may maybe maybe to put it put it kind of uh kind of bluntly like we were the first agentic a IDE experience right uh for when we launched Windsurf uh we were we were the very first ones obviously now that idea is not very it's not non-consensus like it was the obvious thing to do but these models like Sonnet had existed for quite a while but it actually took us some work to actually put together uh like using these models but also using our own models internally to make this a usable experience with the UIUS to be able to review all the software very quickly before people actually this the product experience kind of took off.

说得直白点,我们是第一个 agentic 的 IDE 体验,对吧,Windsurf 上线的时候我们显然是最早的那一批;当然现在这个想法已经不太算非共识了,它成了明摆着该做的事,但像 Sonnet 这样的模型其实已经出来相当一阵子了,我们还是花了不少功夫,才把这些模型——同时也用上我们自己内部的模型——拼成一个可用的体验,再配上能让人飞快评审所有代码的 UI/UX,这个产品体验才真正跑起来。

So I think it's a it's maybe a little bit of both but I don't want to underplay how much the models actually affect you know the value people get from these experiences.

所以我觉得两边可能都占一点,但我不想低估模型到底在多大程度上影响了人们从这些体验里拿到的价值。

People often compare the model landscape to the cloud landscape.

人们经常把模型的格局类比成云的格局。

The thing that I I find challenging about that is cloud you generally stick to.

我觉得这个类比难办的地方在于,云你一般是黏住不换的。

You're on AWS you're not switching between AWS and Google and anyone else.

你上了 AWS 就是 AWS,不会在 AWS、Google 和别家之间来回切。

To what extent do you think model landscape will be one where you have a preferential relationship with an AWS of the world in cloud or Anthropic or OpenAI and models versus actually you just work between many different model providers and you switch between them for different functions?

你觉得模型的格局有多大可能像云里那样,你跟某个「AWS 级」的玩家、或者模型里的 Anthropic、OpenAI 建立一种优先关系,还是说你其实就是在很多家模型提供商之间穿梭,按不同用途来回切换?

I think you're right.

我觉得你说得对。

The the space is very early is the reason why it's kind of like the switching cost of these c of the category is not very high, right?

这个领域非常早期,所以这个品类的切换成本并不高,对吧?

Moving from one model provider to the next.

从一家模型提供商换到下一家。

It's very low switching costs ultimately.

归根到底切换成本非常低。

And by the way, that's because there's no state in these models for the most part.

顺便说一句,那是因为这些模型基本上没有状态。

So for the most part, it is almost like a Twilio like experience where it's like send text message and you kind of send a text message, but maybe I could send text messages with a lot of different applications.

所以大多数时候,它几乎就是 Twilio 那样的体验——发条短信,你就发条短信,但我也可以用很多别的应用来发短信。

But I don't take that as a big negative.

但我不觉得这是多大的负面。

I think what that means is we're still evolving in this category, right.

我觉得这说明这个品类还在演进,对吧。

Down the line you could imagine models can actually internally keep a lot of state, right, about a user, about about all the data they have inside their company, about all the data they they use locally.

往后你可以想象,模型内部其实能保留大量状态,对吧——关于某个用户,关于他们公司里的所有数据,关于他们在本地用到的所有数据。

If you were to look at a lot of large code bases, they are actually getting to billions of tokens of code, right.

你去看很多大型 codebase,它们的代码量其实已经到了数十亿 token,对吧。

What if that's actually something stateful that you can inject and inject and eject context about this from a particular model provider?

如果这变成某种有状态的东西,你可以从某一家模型提供商那里注入、注入,再弹出关于它的 context 呢?

Like, I'm sure there are ways in which they can increase their switching costs down the line as the technology gets better and better.

我相信随着技术越来越好,他们往后是有办法把自己的切换成本抬上去的。

But I think this is what it comes down to, right, Eric?

但我觉得最终归结到的就是这一点,对吧,Eric?

In a category that is evolving so quickly and when the models are improving so quickly, it is actually bad for them to overinvest in ways in which they can increase switching costs.

在一个演进这么快、模型也进步这么快的品类里,过度投入去抬高切换成本,对他们其实是坏事。

That's just doesn't make sense because let's suppose they did do that but they missed the next improvement in the model, like they for they they didn't they didn't do test time compute or something, they would be behind and then it wouldn't matter that they increase the solution cost.

这根本说不通,因为假设他们真这么干了,却错过了模型的下一次进步,比如没做 test time compute 之类的,他们就落后了,那时候抬高了切换成本也没意义了。

So I think ultimately your the analogy to cloud providers is not accurate right now.

所以我觉得归根到底,你说的云厂商那个类比,现在是不准确的。

They don't have the same properties.

它们不具备同样的属性。

It's not the same kind of business right now.

现在它不是同一种生意。

Um, but could it be down the line?

但往后有没有可能变成那样?

I think it could.

我觉得有可能。

Today models do have non-commoditized properties.

今天的模型确实有一些没被商品化的属性。

Anthropic is often hailed as being better for developers, for example, just as like a most base one.

比如 Anthropic 就常被夸对开发者更好,这算最基础的一个例子。

To what extent do we actually see the complete commoditization?

我们究竟会在多大程度上看到彻底的商品化?

I mean there are literally 12 13 different providers in China who are racing faster than ever and we'll reach some asymptotic utility point where all of them hit the same level and I'm sure the US will be the same.

我是说,光中国就有整整 12、13 家不同的提供商在以前所未有的速度狂奔,大家会到达某个效用的渐近点,所有人都撞到同一个水平,我相信美国也会一样。

To what extent do we hit that versus no no Anthropic is the one for developers, X is the one for whatever else.

我们会在多大程度上走到那一步,还是说不不,Anthropic 就是开发者的那一家,X 就是别的什么场景的那一家?

I think it is unlikely that in the short term you will have a like a an ability for one model provider to just run away with a particular category um if the category is like very valuable, right?

我认为短期内不太可能有哪一家模型厂商在某个品类上直接一骑绝尘——尤其这个品类还非常有价值,对吧?

I think it's it's unlikely that that's the case now.

我觉得现在不太可能是这样。

I do believe in things like the scaling hypothesis which is to say with more scale you you are able to build better and better models and you do have like your your hands tied up if you're operating with less capital in the category, right?

scaling 假说这类东西我是信的,也就是说规模越大,你能造出的模型就越好;而如果你在这个品类里手上资本更少,你就会束手束脚,对吧?

It's just it's just the the nature of the game um right now.

这就是眼下这个游戏的本质。

But I don't I don't think we're anywhere near the point in which you're going to have someone who's the defining leader of having a great model and no one else can touch it for for years, right?

但我完全不觉得我们已经接近那个点:某个人成了拥有最强模型的绝对领跑者,其他人好几年都碰不到他,对吧?

Um, and and maybe the the reason for that is actually because 6 months from now if a new technique exists and someone else finds the new technique um ultimately they're able to do it in a much more efficient way.

原因可能在于:6 个月后如果冒出一种新技术,而别人先找到了它,他们最终就能用高效得多的方式把事情做成。

And if they're able to do it in a much more efficient way, it means that if you just miss it and you're still applying your previous technique, you're going to be behind them, right?

而一旦他们能用高效得多的方式做,就意味着你只要错过这一步、还在用原来那套技术,就会落在他们后面,对吧?

So, it's almost like unless you believe that companies one company is going to get all the capital in the world uh and no one else is ever going to get the capital in the world and one company has a monopoly on all great ideas, it is unlikely that you're going to have someone that is able to like run away with with something in in the short term, if that makes sense.

所以几乎可以说,除非你相信有一家公司会拿走全世界所有的资本、别人永远拿不到,而且这一家还垄断了所有好点子,否则短期内不太可能有谁带着什么东西一骑绝尘——如果这话说得通的话。

Given the commoditization of that model layer as we said and no one being able to run away, to what extent the model providers have to move into the app layer to have any form of differentiation?

既然像你说的,模型层正在商品化,谁也没法一骑绝尘,那模型厂商在多大程度上必须走进应用层,才能有任何形式的差异化?

Maybe in the in the API side I think there are lots of ways you can build better APIs for a lot of different applications.

也许在 API 这一侧,我觉得有很多办法能为各种不同的应用做出更好的 API。

And it's kind of like maybe you can you can you can kind of focus on the applications you do support.

有点像是,你可以专注在自己确实支持的那些应用上。

And an example of this is like let's say you you were kind of working with a company like us um and we were using a model provider, you when we output code, maybe a lot of the code looks the same as existing code and you can actually optimize from a latency perspective to help our workload versus another workload that exists for images out there and you can be a much better kind of vendor to us than another company.

举个例子,假设你在跟我们这样的公司合作,而我们用的是某家模型厂商——我们输出代码时,很多代码跟已有代码长得一样,你其实可以从延迟角度做优化来服务我们这种负载,而不是外面那种跑图像的负载,这样你对我们来说就是个比别家好得多的供应商。

So there is maybe specialization you can have on different axes to provide a better experience but as you can see you you are seeing like these model companies actually going up to the app layer too in whatever apps they see are like valuable to them.

所以在不同维度上你也许能做出专业化,提供更好的体验;但你也看到了,这些模型公司其实也在往应用层走,只要那些应用在他们看来有价值。

Um, so I think you're seeing a little bit of both.

所以我觉得两种情况都能看到一点。

People have too many opinions about what the right way to do things are and I think you you have to be rational on on what the best way to win is and I think that's the only thing that matters.

大家对什么才是正确做法的观点太多了,我觉得你必须在“什么是取胜的最佳方式”上保持理性,这才是唯一重要的事。

Uh, final thing I I do just have to touch on because I talked to you for a long long time.

最后一件事我非要聊一下,因为我已经跟你聊了很久很久了。

Um, Dario the other day was talking at at some event or conference or whatever and he was like 2026 we'll have like solo billion dollar companies without doubt.

前几天 Dario 在某个活动还是会议上讲话,他说 2026 年我们毫无疑问会出现十亿美元的一人公司。

I just wanted to touch on this with you.

我就是想跟你聊聊这个。

Do you agree?

你同意吗?

No, I don't believe in that.

不,我不信这个。

Please explain.

请解释。

This feels like a university essay.

这感觉像一道大学论文题。

Discuss.

请展开论述。

I mean I mean I think the like the solo billion dollar company would imply that no one else cares.

我是说,十亿美元的一人公司这种说法,言下之意是没别人在乎这件事。

Like we live in a capitalist market so there's competition and if you're able to do something with one person unless you're just a thousand times smarter and more capable than everyone else.

我们生活在资本主义市场里,竞争是存在的;如果一个人就能做成某件事,除非你比其他所有人聪明、能干一千倍。

Like someone with two people that are as smart as you are probably going to be able to do your idea.

两个跟你一样聪明的人,大概率也能把你的点子做出来。

And if they do that's going to compress your margins.

他们真做了,就会压缩你的利润率。

That's going to compress your ability to be the pervasive brand out there.

会压缩你成为那个无处不在的品牌的空间。

That's going to compress your customer acquisition cost or increase your customer acquisition cost.

会压缩你的获客成本——或者说抬高你的获客成本。

Like I don't know, seems like very hard for it to be the case that a single person is going to be able to run away with a billion dollar product uh without continued investment.

我不知道,一个人在没有持续投入的情况下就带着一个十亿美元的产品一骑绝尘,这事看起来非常难成立。

Like it just the implication of that is like no one else cares about this but magically you're able to do well.

因为它的言下之意就是:没别人在乎这件事,但你却神奇地做成了。

So I don't know.

所以我不知道。

What what do people who believe this to be true get wrong?

那些真信这套的人,错在哪儿?

What are they fundamentally mistaking?

他们从根本上误判了什么?

I I I actually I' never understood this idea of like a single person is going to be able to build a billion-dollar company.

我其实从来没理解过“一个人就能建成十亿美元公司”这个想法。

I've never kind of understood it.

一直没太理解。

I mean, because what is a company?

因为公司是什么?

A company is the sum of the discounted cash flows over time.

公司就是随时间推移的折现现金流之和。

And uh and like down down the line like you're going to need people probably or if the AI is so good that you don't need people like at some point I would just wonder like if like what are you really contributing and is the AI building the whole thing in which case like I don't know like actually that would reduce the that would increase competitive pressure because everyone else will have access to the AI too.

再往后走,你大概率会需要人;或者如果 AI 好到你不需要人了,那到某个时候我就会想:你到底真正贡献了什么?是不是 AI 在造这整个东西?真是那样的话,我不知道,那其实会降低……其实会加剧竞争压力,因为别人也一样能用上这个 AI。

Can I ask you, you know, you've been across multiple different cycles of the business, ups and downs.

我能问你一个问题吗——你经历过这门生意的好几轮周期,起起落落。

When did you question your love for it most?

什么时候你最怀疑自己对它的热爱?

I think we often say like, "Oh, I'm always inspired.

我觉得我们常爱说,“哦,我永远都充满激情”。

I'm not inspired every day to do what I do.

我并不是每天都有激情去做我做的事。

Some days I am, some days. "

有些日子有,有些日子没有。

Was there a period where you were not?

有没有哪段时期你是没有的?

And how did you get through that period?

你又是怎么熬过来的?

I think every time basically every time we've needed to make a strategic pivot like actually it's before the pivot that that it's it's probably the the most anxietyinducing because you feel like something could be much better than it is but you are not acting on it and I think that is potentially the the part that is that is the most uh kind of gut-wrenching.

我想基本上每次我们需要做战略转向的时候——其实是转向之前那段最让人焦虑,因为你觉得事情本可以好得多,自己却没动手;我觉得那可能是最揪心的部分。

Um, you know, it's actually an interesting point when we pivoted from Exafunction to Codeium.

有个很有意思的点,就是我们从 Exafunction 转向 Codeium 那次。

I actually think post pivot was was the freest I've ever felt because I had written off in my head uh both me and my co-founder that we're likely going to fail, but at least we're likely going to fail in something that we believe in.

我其实觉得转向之后是我这辈子最自由的时候,因为我心里已经认了——我和我的联合创始人都认了——我们大概率会失败,但至少是在一件自己相信的事上失败。

In the past, yeah, we were making a couple million in revenue, but I know what it means to be like a very large successful company.

在那之前,是的,我们确实有几百万美元收入,但我知道一家非常大、非常成功的公司意味着什么。

You need to be able to make billions in revenue, right?

你得能做到几十亿美元的收入,对吧?

Uh so we were so far away from that outcome that it was almost like yeah this this feels very hard to do.

所以我们离那个结果实在太远了,几乎就是:是啊,这看着太难做到了。

But after pivoting we had a new lease on life and I actually feel like for us and for me the part that is the most anxietyinducing is when we believe we could be doing much better but there's random inertia friction that is preventing us from doing the right thing.

但转向之后我们重获新生;我其实觉得,对我们、对我来说,最让人焦虑的时刻是:我们明知自己本可以做得好得多,却有种莫名的惯性和摩擦挡着,不让我们去做对的事。

And even if doing the doing the right thing has a low probability of success, actually starting to work on it is is the most free experience that I feel like I've had at the company.

而且哪怕做对的事成功概率很低,真正动手去做,也是我在这家公司感受过的最自由的体验。

I heard you say once an interesting statement.

我有次听你说过一句很有意思的话。

You said you wait until you're drowning in a role before you hire for it.

你说你会一直等到自己在某个岗位上快淹死了,才去为它招人。

And I just thought it was really interesting because I was always brought up on the value of board members being they highlight hires that you need to make six months ahead so you don't have a bottleneck in six months time.

我当时就觉得这太有意思了,因为我从小被灌输的是:董事会成员的价值在于提前六个月点出你该招的人,这样六个月后你就不会卡在瓶颈上。

How do you think about those two seemingly opposing ideas?

这两个看起来相反的想法,你怎么看?

We've had cases where I would say it's a little bit of a of an antiattern on that one.

我们遇到过一些情况,在这一点上我会说有点像反模式。

Like for instance, when we before we hired our VP of sales um who who was who was really awesome um we we actually closed some very large enterprises inside the company without any salespeople.

比如说,在招我们那位 VP of Sales 之前——他非常出色——我们其实在公司里一个销售都没有的情况下,就签下了几家非常大的企业客户。

Some very large enterprise sales would look like inside the company and also it proved to the person we were going to hire, Graham, that we were an awesome company to potentially work at.

在公司内部亲眼看到超大额企业销售长什么样;而且这也向我们打算招的那个人 Graham 证明了,我们是一家值得来的好公司。

Right?

对吧?

So I think it's it's an interesting trade-off where maybe there's a belief that I have that domain experts are awesome, but if inside the company you're not able to get get an outcome that is 1% 10% as good inside the company, it's probably not the right time to go out and make the hire.

所以我觉得这是个很有意思的取舍:我或许是相信领域专家很棒的,但如果在公司内部你连 1%、10% 那么好的结果都拿不出来,那大概就不是出去招人的时机。

Like maybe let me put it this way.

或者我换个说法。

This is maybe the the the concise way to actually answer this.

这也许是回答这个问题更简洁的方式。

Startups don't fail because they look like messes inside.

创业公司失败,不是因为内部看着一团糟。

Startups fail because they don't do the right thing uh well enough.

创业公司失败,是因为没把对的事做得足够好。

That is the reason why they fail.

那才是它们失败的原因。

And I think people love to say, hey, like I don't have this role and that's why I'm not working.

而我觉得大家很爱说:嘿,我缺这个岗位的人,所以事情才推不动。

And it's like actually when you look at the most successful companies, they look like train wrecks inside, right?

但实际上你去看那些最成功的公司,内部看着都像车祸现场,对吧?

Ultimately, and what I think is important there is is as long as you are doing the basic things that make your product very successful, I think you can actually make these hires like as necessary.

归根到底,我觉得关键在于:只要你把那些让产品跑起来的基本动作做到位,这些招聘其实可以按需再来。

I don't think it's like critical that you make them six months in advance.

我不觉得提前六个月招人有多关键。

The worst thing for me would be making a hire and for them to not be important on day one and just to be, hey, chill out over here and wait for the next 3, 6, 12 months because then they're going to manufacture something to work on and I won't really care about what they're doing.

对我来说最糟的事,是招了个人,他第一天却无关紧要,只能跟他说“嘿,你先在这儿歇着,等上 3 个月、6 个月、12 个月”,因为那样他就会自己造点事情来做,而我根本不会在意他在干什么。

Final one before a quickfire.

快问快答之前,最后一个问题。

What did you believe on management that you've changed your mind on?

管理上,你有哪个原来的看法后来变了?

So like for me as an example, like I brought in this incredible person, Julian, he led workplace in Europe for 10 years and I always believed dude arrogantly as a young person.

拿我自己举例,我招进来一个特别厉害的人,Julian,他在欧洲带 workplace 带了十年;我年轻时一直很自大地以为,

I could read Elad Gil's high growth handbook and I know how to manage people and it's

读一读 Elad Gil 的《High Growth Handbook》,我就知道怎么管人了,可实际上

The wisdom and experience this guy brought from 15 20 years leading teams is so nuanced and granular.

这个人带团队十五二十年沉淀下来的智慧和经验,细腻、颗粒度之细,难以想象。

I learned so much from him about management every day now.

现在我每天都在从他身上学管理。

You can't learn it in a book.

这些书里学不到。

I've changed my mind on that.

这一点上我改了看法。

What would your change of mind be on management?

你在管理上改变的看法是什么?

Maybe the biggest change of mind I I did have was was what you originally said which is how large your team is is not a function of like output output and actually ruthless focus is the only thing that matters right so you can have a lot of people but the answer is if those people are not ruthlessly focused on one thing um you are going to be very very not successful like it is it is much easier to run a company that is focused that has only one thing that matters than it is to run a company with the same amount of revenue with five things that matters that with five different priorities and I think in retrospect that always feels that that feels obvious but at a startup like ours where there's so many things we can do there's always a couple shiny things you can always open Twitter and it's like X is solved Y is solved Z is solved but actually I think being able to say no to things and to focus on on the hand like the one thing that matters with many people and ruthlessly optimize optimize that

我最大的一次转变,大概就是你刚才说的那个——团队多大并不决定产出,真正唯一重要的是极致聚焦。你可以人很多,但这些人如果没有极致聚焦在一件事上,结果就是极其失败;经营一家聚焦的、只有一件事重要的公司,远比经营一家收入相同、却有五件事都重要、五个不同优先级的公司要容易。事后回想,这好像是废话,但在我们这样的创业公司里,能做的事太多了,总有几样亮眼的新东西冒出来,随手打开 Twitter,X 被解决了、Y 被解决了、Z 被解决了。可我觉得,能对事情说不,把很多人聚焦到那一件重要的事上,再极致地把它优化到底,

Um, it's very hard.

非常难。

Uh, but when you can actually get that right, you can you can make magic happen.

但真做对了,就能创造奇迹。

I could talk to you all day.

我能跟你聊一整天。

I do want to do a quickfire and I have some great quickfire questions for you.

但我确实想做个快问快答,给你准备了几个很棒的问题。

So, I'm going to start with one of that I'm adding.

先从一个我临时加的开始。

These ones.

就这几个。

Um, Neil Mehta, I love him as an ambassador and I'm an ambassador.

Neil Mehta,我很喜欢他这个投资人,我自己也是投资人。

So, I don't see the same Neil Mehta that you see.

所以我眼里的 Neil Mehta 跟你眼里的不是同一个。

What makes him so special for you as a founder?

作为创始人,他对你来说特别在哪儿?

I think his ability to analyze companies and be very intellectually flexible is is uh is actually pretty remarkable, right?

我觉得他分析公司的能力、思维上的极度灵活,其实相当了不起,对吧?

Like the breath of knowledge that he has about all these different businesses, but also to understand what makes them work, what are some of the ways they could improve and how that could be applicable to us um is is actually like very awesome.

他对各种生意的知识广度,还能看懂是什么让它们跑得通、有哪些地方能改进、这些又怎么用到我们身上,真的非常厉害。

And he's like a very kind of person with his time, right?

而且他对自己的时间很慷慨。

You would assume given the fact that you know he is he's he works with so many companies that he would not have a lot of time but he is extremely generous with his time.

你会以为,他同时跟这么多公司打交道,应该没什么时间,可他在时间上极其大方。

Who do you not have on your board that you would most like to have on your board?

现在不在你董事会、但你最想请进来的是谁?

I actually I would love to have someone like Scott Cook on our board.

我其实很想请 Scott Cook 这样的人进董事会。

Someone someone like uh someone who's the CEO of like a a CEO and founder of like an extremely successful software company.

就是那种——一家极其成功的软件公司的 CEO 兼创始人。

Um, yeah, someone someone I' I've had a handful of conversations with him and every time I walk away just with a deep understanding of just uh competitive dynamics and just what it takes to run a very enduring long long-term successful business.

我跟他聊过几次,每次聊完,我对竞争格局、对经营一家真正长青的生意需要什么,理解都深一层。

Dude, that is a great shout.

兄弟,这个答案太棒了。

Um, I haven't had that before.

以前没人这么答过。

If you had to make Windsurf open source tomorrow, how would you feel and what would you do instead to win?

如果明天必须把 Windsurf open source,你会是什么感觉,你会改用什么方式去赢?

You know, I I I feel like that's a that's a question I've been I've been getting uh recently.

这个问题我最近老被问到。

I don't think it would affect affect much if we had to make it open source because most of the work that it's kind of like the thing I said before right you know Google is yeah they have a front end but the front end is just a single box and if they open source the front end of Google I don't think anything would really change about the world right that that box the UX for that was not really the big deal.

真要 open source,我不觉得有多大影响,因为大部分工作——就是我前面说的那个道理:Google 是有前端,但那前端就是一个输入框;他们要是把 Google 的前端 open source 了,我不觉得世界会有什么真变化,那个框、那个 UX 本身不是关键。

I am aware of the fact that Copilot has done that but most of our logic is actually in the back end ultimately so I don't think that it would change a ton.

我知道 Copilot 已经这么做了,但我们的逻辑最终大部分在后端,所以我不觉得会变多少。

I also don't think it would change our adoption um of the of the product uh like most of the differentiation has nothing to do with the fact that the front end is actually open source or not.

也不觉得会影响产品的采用,大部分差异化跟前端开不 open source 根本没关系。

You have a sibling, they're 18 years old and they ask you, should I go to university?

你有个弟弟妹妹,18 岁,来问你:我该不该上大学?

What should I study?

该学什么?

And how should I position myself best for this new world of work?

在这个新的工作世界里,怎么给自己定位最好?

What do you advise them?

你怎么建议?

Just focus on problem solving, right?

就专注解决问题,对吧?

How do you how do you like break down problems?

你怎么拆解一个问题?

Uh be a high agency individual.

做个高能动性的人。

And I think computer science largely speaking is just the study of problem solving uh for the most part.

而且我觉得计算机科学大体上就是在研究怎么解决问题。

So maybe what doesn't matter as much is did you learn X language or Y language but can you break apart a problem into their P distinct pieces and actually go out and execute on them is is probably the most important thing.

所以没那么重要的是你学了 X 语言还是 Y 语言,重要的大概是:你能不能把一个问题拆成 P 个独立的部分,再真的去把它们做掉。

What's one thing that you think OpenAI did wrong that you've learned from?

OpenAI 做错了什么、而你从中学到了东西?

Oh maybe naming.

也许是命名。

Let's go with naming on that one.

这题就说命名吧。

Penultimate one.

倒数第二个。

You can change one aspect of your CEOship.

你能改变自己当 CEO 的一个方面。

What do you need to change most?

最需要改的是什么?

I think there are probably some details I I should be I shouldn't be actually like looking at very deeply that fundamentally don't matter about the long-term success of the company, but I just I'm a person that's very neurotic and I like to understand everything that's happening internally like what we're spending money on at a level that probably doesn't matter.

有些细节我其实不该看得那么深,它们对公司的长期成功根本不重要;但我这个人非常神经质,我喜欢把内部发生的一切都搞清楚,比如钱花在哪儿了,细到可能毫无意义的程度。

Uh and uh and I just like waste cycles in my head um that that could be put to better use.

结果就在脑子里空转,浪费了本可以用在更值的地方的算力。

Final one for you.

最后一个。

You can be known for having one impact on humanity.

你可以因为对人类的某一项影响被记住。

What impact do you want your future generations to say?

你希望后代说的是哪一项?

Varun, my grandfather, he did X.

Varun,我的祖父,他做了 X。

I think the mission of our company is is very exciting and and I wouldn't say it's just me, but but the team um is uh we want to reduce the time it takes to build technology by 99%.

我觉得我们公司的使命非常令人兴奋,这也不只是我,而是整个团队——我们要把构建技术所需的时间减少 99%。

And and if we are able to play a meaningful role of how that could shake out in the future, like I would be extremely excited about that.

如果未来这件事怎么落地,我们能在其中扮演一个有分量的角色,我会无比兴奋。

Brilliant.

精彩。

Listen, as I said, I heard so many things from Lee Murray, from Neil.

听着,就像我开头说的,我从 Lee Murray、从 Neil 那儿听了太多关于你的事。

I I was so looking forward to this.

我特别期待这次对话。

You've been fantastic.

你太棒了。

Thank you so much for putting up with my base level of knowledge and I so appreciate it, man.

谢谢你忍受我这点浅薄的认知,真的很感激,兄弟。

No, this was this was fantastic, man.

不,这次聊得太棒了,兄弟。

Thanks a lot.

非常感谢。