Running an AI-Native Engineering Org

"Hey folks, do y'all hear me okay?"

嘿,大家好,都能听清我说话吗?

"Okay, I swear this is not a Claude Code thing, but do you guys mind if I take a photo?"

好,我发誓这跟 Claude Code 没关系,不过介意我先拍张照吗?

"Cuz Boris and Jared had their session at 2:00 and I really thought this was going to be empty."

因为 Boris 和 Jared 的场次是 2:00,我本来真以为这里会空荡荡的。

"I'm like, there is just no way people would still be coming in from that session."

我当时想,绝不可能还有人从那场赶过来吧。

"Oh my gosh. Thank you, prompt."

天哪,谢谢你们配合。

"I promise me and Boris don't just do selfie wars all the time."

我保证,我和 Boris 并不是整天只顾着比拼自拍。

"But good afternoon and thanks for attending."

总之,下午好,谢谢各位来捧场。

"So, yeah, my name is Fiona Fung and I lead Claude Code and Cowork engineering and product."

好,我叫 Fiona Fung,负责 Claude Code 和 Cowork 的工程与产品。

"So, I work really closely with Boris and Cat."

所以我和 Boris、Cat 合作得非常紧密。

"And before Anthropic, I had led and grown teams at Meta and then also Microsoft."

来 Anthropic 之前,我在 Meta、之后又在 Microsoft 带过、也带大过团队。

"And so, for today's talk, this is kind of like the overall agenda, but the whole idea is what are some of the lessons I learned in helping Claude Code and Cowork grow as we're building out this team."

所以今天这场分享,这大致是整体提纲,核心其实是:在我帮 Claude Code 和 Cowork 成长、一边搭建这支团队的过程中,我学到了哪些东西。

"And it's interesting as lessons I learned even if I think about my time at Meta or even Microsoft, but even Anthropic."

有意思的是,无论回想我在 Meta、Microsoft 还是 Anthropic 的日子,这些教训都成立。

"Like it's funny, I did this slide deck maybe like a month ago and already I've had to change some of the content."

说来好笑,这套 slide 大概是一个月前做的,可我已经不得不改掉一些内容了。

"Cuz for example, when I started this deck, there were no routines and even that way of working was different for me."

比方说,我开始做这套 slide 时还没有 routines,连那种工作方式对我来说都是全新的。

"And so, yeah, I want to kind of cover five themes that I've noticed."

所以,我想讲讲我观察到的五个主题。

"One, which is the bottlenecks have moved, they've shifted."

第一个:瓶颈已经移动了,发生了迁移。

"And so, when bottlenecks shift, what are then some of the team norms that we had to rewrite within the Claude Code team?"

那么,当瓶颈迁移时,我们在 Claude Code 团队内部又不得不重写哪些团队规范?

"I also wanted to share a little bit about all these team norms we had to rewrite, how we rolled them out, and also what are some of the signals that I get of, yeah, we're trending in the right direction."

我还想聊聊这些被我们重写的团队规范、我们是怎么推行它们的,以及有哪些信号让我判断:对,我们正朝着正确的方向走。

"And it's always going to be important for me to look at to see, is it still serving us, trending in the right direction?"

对我来说,持续去看「这套做法是否仍然托得住我们、是否还在正确的方向上」永远都很重要。

"And then I'll end it with a few questions that I still have for myself, and then also some suggestions for you to maybe take an action and embark to your teams to have conversations together."

最后,我会以几个我至今仍在问自己的问题收尾,再给你们一些建议,也许你们可以付诸行动,带回自己的团队,一起展开对话。

"So with that, the first section, the bottlenecks have moved. I call it the shift."

那么进入第一部分:瓶颈已经迁移,我把它叫作「范式转移」。

"But you'll probably hear me repeat this subtitle text a lot, which is what served you prior may not serve you any longer."

你大概会听我反复念这句副标题:过去成就你的,未必能继续托住你。

"And even when I think about all my experience, whether it was at Anthropic or Meta or Microsoft, the constant growth mindset is just a muscle that has served me really well."

回顾我所有的经历,不管是在 Anthropic、Meta 还是 Microsoft,持续保持 growth mindset 都是一块帮了我大忙的「肌肉」。

"And especially right now, I don't know if y'all are feeling it, the rate of change is just a little bit crazy, right?"

尤其是当下——不知道你们有没有同感——这变化的速度简直有点疯狂,对吧?

"Like I remember the first time I started doing some live coding was last year, and it was still making some bugs that I'm like, 'Ah, why are you using constants everywhere? That's not good engineering practice.'"

我记得我第一次做 live coding 还是去年,那时它还会冒出一些 bug,我会想:「啊,你怎么到处都在用常量?这不是好的工程实践啊。」

"And now it's just become so much more capable."

而现在,它已经强大太多了。

"But that's what I'm seeing as this interesting shift of, when the bottlenecks moved, how do you think about adapting in terms of everything else around that bottleneck?"

但我看到的正是这个有意思的转移:当瓶颈移动之后,你该如何重新思考围绕它的其他一切、做出调整?

"So maybe y'all are feeling this too, but for years engineering bandwidth was the expensive thing."

也许你们也有同感:过去很多年里,engineering bandwidth 才是那个昂贵的东西。

"Like coding throughput was really expensive."

也就是说,写代码的吞吐量非常昂贵。

"And when you think about all the processes we have of shipping software, a lot of it was around that."

想想我们交付软件的那一整套流程,很多都是围着这件事转的。

"Even when you think about how we used to do planning, like remember we used to do waterfall and then agile, everything used to be because engineering bandwidth was really expensive."

就连我们过去做规划的方式——还记得吗,我们先是 waterfall,后来是 agile——这一切之所以这么做,都是因为 engineering bandwidth 太贵了。

"Actually, I'll take a little segue here. When you think about our industry, we've always had to adapt."

其实我这里插一段题外话:回看我们这个行业,我们一直都在被迫适应。

"Like I'm going to put you all in a time machine. Come back with me all the way to the year 2000s."

我要把你们都塞进时光机,跟我一起回到两千年代。

"That was when I started my career. I worked on Visual Studio, and we were shipping Visual Studio 2005."

那是我职业生涯的起点。我做的是 Visual Studio,当时我们要交付 Visual Studio 2005。

"And I kid you not, in those days, if folks remember, we used to ship software on like CD-ROMs."

我可没开玩笑,那年头——还记得的人就知道——我们是靠 CD-ROM 来发行软件的。

"Before CD-ROMs, it was actually floppy disk."

在 CD-ROM 之前,其实是软盘。

"And so, I still remember VS 2005, there were really hard deadlines."

所以我还记得 VS 2005,那时有非常死的 deadline。

"We had to hit those deadlines cuz we had to get the software to the manufacturing lab to print on the CDs, to put in the boxes, to ship in the stores."

我们必须卡着这些 deadline,因为要把软件送到生产车间刻进光盘、装进盒子、再铺到店里去卖。

"And so, when you even think about that, when we were able to distribute software online, that also changed how we ship software."

所以你想想看,等到我们能在线分发软件时,这同样改变了我们交付软件的方式。

"And so, that's what I'm finding really interesting of this new shift that I'm seeing: engineering bandwidth, it's no longer the expensive thing."

而这正是我觉得这次新转移最有意思的地方:engineering bandwidth 不再是那个昂贵的东西了。

"So, for example, on the Claude Code team, for sure, coding is rarely the slow part anymore."

举个例子,在 Claude Code 团队里,写代码几乎已经不再是拖慢进度的那一环了。

"And I would say it's not even that it's not the slow part, it's just also the throughput has really, really increased."

我甚至想说,不只是它不再拖后腿——而是它的吞吐量真的、真的大幅提升了。

"So, it's not only like, 'Yay, we're all getting to build more.' It's just the amount that we're generating has also changed a lot."

所以这不只是「耶,我们都能造更多东西了」这么简单——而是我们产出的总量本身也变化巨大。

"And so, what we saw was, when your bottleneck shifts from the coding and the actual act of typing — if you remember, it used to be writing code was expensive or writing tests or refactoring."

于是我们看到:当你的瓶颈从写代码、从敲键盘这件事本身移开时——还记得吗,过去是写代码贵、写测试贵、重构贵。

"I remember all of these conversations of, 'We have to schedule some time to do refactoring. Oh, but we have to do product work, and this is expensive. When are you going to find that time to do it?'"

我还记得那些对话:「我们得排出点时间来重构。哦,可我们还得做产品的活,而这很费成本。你到底什么时候才能挤出时间来重构?」

"All of that has shifted now. That is no longer the bottleneck."

如今这一切都变了,那些都不再是瓶颈了。

"And so, when that happened, sometimes I notice the bottlenecks end up shifting towards other areas."

而当这种情况发生时,我注意到瓶颈往往会转移到别的地方去。

"And so, what happened? Like, for example, verification, review, cross-functional partners, security."

那么发生了什么?比如:verification、review、cross-functional 的协作方、security。

"Because coding is no longer the bottleneck, and also we're doing so much more of it, these are some of the new bottlenecks that we're seeing."

正因为写代码不再是瓶颈,而且我们写得又多了那么多,这些就成了我们看到的新瓶颈。

"So, it's really always us asking, 'Is this code correct? Who reviews this code?'"

所以我们其实一直在问:「这段代码对吗?谁来 review 它?」

"That's like probably one of the top questions I get from all fellow eng leaders."

这大概是所有同行的 eng leader 问我最多的问题之一。

"Like, 'How are humans keeping up with how you guys are doing code reviews?'"

他们会问:「按你们这种 code review 的节奏,人到底是怎么跟得上的?」

"And interestingly, also how is it maintained? Because now it is also a lot easier for us all to generate a lot of code."

有意思的是,他们还会问:这些代码又是怎么维护的?因为如今我们所有人生成大量代码都容易多了。

"So also thinking about that maintenance cost, too."

所以那笔维护成本,也得一并考虑进去。

"So with that, these were some of the processes that I noticed quietly stop working."

说到这儿,这些就是我注意到的、那些悄悄失效的流程。

"And I love that phrase 'quietly stops working,' because a lot of times we all put in processes hopefully for a reason, right?"

我特别喜欢「悄悄失效」这个说法,因为很多时候,我们立下一套流程,总归是有原因的,对吧?

"Like we're thinking, 'Hey, there was a gap here, or we want to improve.'"

我们当时想的是:「这里有个缺口」,或者「我们想改进一下」。

"But what I've found over the years is, rarely do processes kill themselves."

但这些年我发现,流程几乎从不会自己消亡。

"We tend to just layer more and more and more processes on."

我们总是一层又一层地往上加流程。

"Like I remember that on one team, we had so many SLAs."

比如我记得有个团队,我们定了一大堆 SLA。

"There was a P0 bug SLA, a high-pri sub-review — anyways, there were so many SLAs."

有 P0 bug 的 SLA,有高优先级的复查 SLA——反正就是 SLA 多到数不过来。

"After a while I'm like, 'Oh no, we need to stack rank priorities so that all the engineers knew which SLA was going to be even more important.'"

过了一阵子我就想:「糟了,我们得给优先级排个座次,让所有工程师都知道哪个 SLA 更重要。」

"And I remember even at that time I thought, 'Hey, we should start thinking about what things we should defrag a little bit.'"

我记得当时我就在想:「我们是不是该开始琢磨,有哪些东西可以做点『碎片整理』了。」

"So yeah, these are the processes that might have served you."

所以说,这些都是曾经帮过你的流程。

"Do you remember again, that line — what may have served you may not serve you any longer."

还记得那句话吗——过去成就你的,未必能继续托住你。

"Like the planning norms: we used to spend a lot more time pre-planning, because coding time was expensive."

比如规划的习惯:过去我们会花更多时间做前期规划,因为写代码的时间很金贵。

"Code ownership — there used to also be a lot of questions of who, who, who wrote this code? Who owns it?"

还有代码归属——以前总有一堆问题:这段代码到底是谁、谁、谁写的?谁负责?

"That's a little bit of another question now."

现在这已经是另一回事了。

"Code reviews — we'll get into that a little bit."

代码评审——这个我们待会儿会细聊。

"Team makeup is interesting, too."

团队构成也很有意思。

"It's not only that roles are blurring, right?"

不只是角色之间的界线越来越模糊,对吧?

"Now we have AI to help augment non-engineering roles."

现在我们有了 AI,可以去增强那些非工程岗位。

"My non-engineering partners are also all shipping code."

我那些非工程的搭档,现在也都在交付代码了。

"So what happens when roles start blurring and you don't have those silos anymore?"

那么,当角色开始模糊、原先那些壁垒不复存在时,会发生什么?

"And then that also goes to knowledge share."

接着这又牵扯到知识共享。

"Knowledge sharing and onboarding — that's another signal we're noticing at Claude Code, how we used to do things changes a little bit, too."

知识共享和新人上手——这是我们在 Claude Code 团队注意到的又一个信号,我们过去的做事方式也在悄悄改变。

"And so, in the first section we talked about the shift."

所以,在第一部分里我们聊了这场转变。

"So within the Claude Code team, what are some of the norms that we have to rewrite?"

那么在 Claude Code 团队内部,有哪些习惯是我们不得不重写的?

"I want to share some of those with you, and hopefully some of them will resonate with you, or you might find helpful."

我想跟大家分享其中几条,但愿有些能让你产生共鸣,或者对你有点用。

"Number one is code review — like human judgment of who actually needs it, and we'll kind of go through all of these."

第一条是代码评审——比如靠人来判断,到底谁才真正需要它,这些我们会逐个过一遍。

"Onboarding has also changed, and how we do planning — you heard me talk about planning a lot."

新人上手也变了,还有我们做规划的方式——你已经听我念叨过很多次规划了。

"Hiring, especially with roles blurring, and team makeup — how that has changed for us, too."

招聘也是,尤其是在角色变模糊之后;还有团队构成——这对我们来说同样变了。

"And also org shape — that's kind of one of my favorite spicy topics."

还有组织形态——这算是我最爱的「辛辣」话题之一了。

"I'll share that story with y'all when I started proposing that at Anthropic."

等讲到我在 Anthropic 第一次提出这个想法时,我会把那段故事讲给你们听。

"I love my recruiting partners, they're awesome, but I remember one recruiting partner really did think I was crazy."

我很爱我那些招聘搭档,他们都很棒,但我记得有一位真的觉得我疯了。

"So I want to share that with y'all, too."

所以这段我也想分享给大家。

"So, how have planning changed? But also technical debates."

那么,规划方式发生了什么变化?还有技术争论也是。

"Planning, we do a lot less of it."

规划这件事,我们做得少多了。

"Like I would also say, and also the timing, I call it like JIT planning, almost like JIT compiling."

我还想说一点,就是时机问题,我把它叫做 JIT planning,几乎就像 JIT compiling 一样。

"Because even when I first joined, I'm like, 'Don't we need a six-month roadmap?'"

因为我刚加入的时候还在想:"我们难道不需要一份半年路线图吗?"

"And we put some effort in, we wrote it, it was pretty good for three months, and then I came back over the new year and so many things had changed already."

于是我们花了点功夫,把它写出来,前三个月它都挺管用,可等我过完新年回来,已经有太多东西变了。

"So, I realized, well, six-month roadmap just seems like a little bit too long."

所以我意识到,半年路线图好像有点太长了。

"So, again, it's how do you make sure you kind of do just the right amount in the right time, because, again, prototyping and code generation is just not the bottleneck that it used to be."

所以问题还是:你怎么确保在对的时间只做恰到好处的那一份规划,因为说到底,原型开发和代码生成早已不是过去那个瓶颈了。

"The technical debate one is a fun one, too."

技术争论这一点也很有意思。

"So, in technical debates, code wins."

在技术争论里,代码说了算。

"I'll share with y'all, when I first joined the Claude Code team, I wanted to do a refactoring."

跟大家分享一下,我刚进 Claude Code 团队时,想做一次重构。

"I wanted to like get to learn about the code base."

我想借此熟悉一下代码库。

"And me and Boris had a healthy technical debate of, you know, which way to go, and I almost leaned into my old toolbox."

我和 Boris 就该走哪条路,展开了一场良性的技术争论,我差点就掏出了自己的老一套。

"I almost tapped him on the shoulder to go, 'Let's go to that room and have a whiteboard,' and I'm like, wait wait wait wait wait a minute."

我差点拍他肩膀说:"走,我们去那间会议室,找块白板画一画",可我突然想,等等,等一下。

"Nowadays, I can just generate all the different options we've been discussing."

现在,我可以直接把我们讨论的各种方案全都生成出来。

"I generated three PRs."

我生成了三个 PR。

"And the cool part about that for the technical debate is I really cared not only about the implementation of the API, but also the impact to all the callers into the API."

而这对技术争论最妙的地方在于,我真正在意的不只是这个 API 怎么实现,还有它对所有调用方的影响。

"And so when I can have Claude help me generate the three different versions, it allowed us to have a debate of not only implementation, but also impact to the colleagues."

所以当我能让 Claude 帮我把三个不同版本都生成出来,我们就不仅能争论实现方式,还能争论它对同事们的影响。

"So I think that's another really interesting kind of pivotal change."

所以我觉得,这是另一个非常有意思的关键转变。

"Um, so yeah, like you know, when building is cheap, arguing expensive, again, how does that shift your team norms a bit?"

所以你懂的,当构建变得廉价、争论反而昂贵时,这又会怎样稍稍改变你们团队的规矩呢?

"I do want to call out, this makes it even more important to make sure you set up team culture for how you think about alignment."

我特别想强调,这让一件事变得更加重要:你必须在团队文化里立好关于"如何达成共识"的规矩。

"For example, what totally won't fly is, because code is so much faster for us to generate, it shouldn't be like the last person who checks in wins."

举个例子,绝对行不通的是:因为代码生成快了这么多,就变成"最后一个提交的人赢"。

"You know, like I'm going to stay up at 3:00 p.m. to submit this PR, I set up a routine so that I get the last word in."

就比如,我熬到 3:00 p.m. 提交这个 PR,我还设了个 routine,好让我抢到最后发言权。

"So definitely a no-no."

这绝对是要不得的。

"That makes it even more important to make sure you have good team culture that can have open, honest technical debates, but also a good team alignment."

这让另一件事更加重要:你得有健康的团队文化,既能开诚布公地展开技术争论,又能保持良好的团队共识。

"Uh, so I talked a lot about kind of what we reduce in planning."

前面我讲了不少我们在规划上"减掉"了什么。

"On Claude Code, we definitely have reduced the design doc before every code ritual."

在 Claude Code,我们确实削减了"每次写代码前先写设计文档"这个仪式。

"I would say certain teams and for definitely certain scenarios it's still really important to think about design docs, especially as we're doing like kind of async discussions."

我得说,对某些团队、某些场景,认真写设计文档仍然很重要,尤其是当我们在做 async 讨论的时候。

"But on Claude Code, most of our discussions is like, instead of a doc, a PR."

但在 Claude Code,我们大多数讨论的载体不是一份文档,而是一个 PR。

"That's kind of like one of the words we have of, 'Hey, we found an idea, go prototype.'"

这差不多就是我们的一句口头禅:"嘿,我们有个想法,去做原型吧。"

"That's the other thing. We don't really do a lot of product reviews because the landscape is changing fast."

还有一点,我们其实不怎么做产品评审,因为整个格局变化太快了。

"So let's prototype, let's actually get a lot of internal ants using it, and I'm really a big fan of shipping it out to all of you, and then hearing all the excellent feedback."

所以我们直接做原型,让大量内部同事(ants)先用起来,而我特别喜欢把它直接发布给在座各位,然后听到各种很棒的反馈。

"And again, that has changed our planning ritual to be a lot less design docs, and mostly discussions are in PRs or prototypes."

所以说,这又一次改变了我们的规划仪式:设计文档少了很多,讨论大多发生在 PR 或原型里。

So we reduced that, but what did we double down on?

所以我们砍掉了那部分,但我们加倍投入的是什么?

And I think this is an area where we actually need to do more and continue to be better — it's verification.

我觉得这块恰恰是我们需要做得更多、持续做得更好的地方,那就是验证。

Because again, the throughput is different, and there are new ways to break.

因为还是那句话,吞吐量变了,出错的方式也变多了。

So how can you scale out? And I call it kind of shift left.

那你要怎么扩展开来?我把它叫做某种 shift left。

Like in the old days, you would get code out, and I would love for me to find bugs before any of you find it.

就像过去那样,你把代码发出去,而我特别希望由我在你们任何人之前先找到 bug。

And what's better than me finding a bug is actually what I call shift left — more automation, so we catch it earlier to the source.

而比我自己抓到 bug 更好的,其实就是我所说的 shift left——更多自动化,让我们更早、更靠近源头就把它拦下来。

So that is something that I think we need to continue to double down on.

所以这是我认为我们需要持续加倍投入的事情。

And the other thing that's interesting is also because roles are blurring.

另一件有意思的事,是因为角色之间的界限正在变模糊。

For example, my designers — I would love to have more confidence that when I checked in this code, I don't break something.

比如我的设计师们——我特别希望能更有把握:当我提交这段代码时,不会弄坏什么东西。

I remember I fixed a bug in resume or something, and the next day I was catching up on Boris's threads and I saw someone tag him up, "I'm noticing a bug."

我记得我修过一个 resume 相关的 bug,第二天我在翻 Boris 的消息串时,看到有人 @ 他说:"我注意到一个 bug。"

I remembered having that sinking feeling. I don't know if y'all have ever — did I just catch a bug?

我那一刻心里咯噔一下。不知道你们有没有过这种感觉——是我刚刚写出了一个 bug 吗?

So because of the throughput, I just really want to make sure everybody, regardless of roles, has a lot higher confidence for the change that they're putting in.

所以正因为吞吐量变大了,我真的很想确保每个人,不管什么角色,对自己提交的改动都有高得多的把握。

Who made this change? And so my advice here is, because all our PRs are assisted by Claude, it's a little bit of an odd question.

这个改动是谁做的?我的建议是,因为我们所有的 PR 都有 Claude 协助,这个问题本身就有点怪。

And so for us, what is more helpful than just that question is what I call double clicking into it.

所以对我们来说,比起单问这一句,更有用的是我所说的"往下点一层"。

In the old days, when you used to ask, "Who made this change?" — what question are you really trying to answer?

过去你问"这是谁改的?"时,你真正想回答的到底是什么问题?

Are you looking for who caused this regression?

你是想找出谁造成了这次回退?

Definitely don't want to blame, but just want to know who was the last person that touched this code that might have caused this break.

绝不是要追责,只是想知道最后碰过这段代码、可能导致这次出错的是谁。

Or are you looking for an expert to answer a customer question? Or are you looking to gain context?

还是说你想找个专家来回答客户的问题?又或者你只是想补足上下文?

So whatever is that double click question, also think about: is there a way you can automate it?

所以无论"往下点一层"的那个问题是什么,也想想:有没有办法把它自动化?

It's funny, I mentioned that to change this slide deck.

说来好笑,正是为了这件事我改了这页幻灯片。

When I started this talk about a month ago, part of my every morning was I would bring up my desktop Claude Code and go into a customer feedback channel, and I'd want to ask it to summarize.

大约一个月前我开始准备这次演讲时,我每天早上都会打开桌面版 Claude Code,进到一个客户反馈频道,然后想让它帮我做个总结。

And that was just a morning ritual I would do with my morning cup of coffee.

这就成了我配着早晨那杯咖啡的固定仪式。

Now it's routines, right? Like I can set that up, and it's even better than what I had done before, which was how I kick it off.

而现在它成了 routine,对吧?我可以把它设好,而且比我之前手动启动的做法还要好。

With my morning cup of coffee, I always get to a pretty good overview.

就着早晨那杯咖啡,我总能拿到一份相当不错的概览。

So that's my point — code ownership is a little bit more fuzzy, but on the flip side, double click to what question you're really trying to answer and see how Claude can actually help you with those.

所以我的意思是——代码的归属变得更模糊了,但反过来说,往下点一层、看清你真正想回答的是什么问题,再看看 Claude 究竟能怎么帮你。

How do you keep up with code reviews?

你要怎么跟上代码审查的节奏?

So I'm really glad if y'all saw the keynote this morning — Cat talked about it. We definitely leverage Claude code review heavily.

如果你们看了今早的 keynote 我会很高兴——Cat 讲过这一点。我们确实大量用上了 Claude 的代码审查。

Now what's interesting is going to be: where do you trust Claude a lot, but then where do you still want a human?

现在有意思的问题在于:哪些地方你非常信任 Claude,而哪些地方你仍然想要一个人来把关?

And actually, as Cat showed y'all, Claude also does a great job babysitting PRs.

而且正如 Cat 给你们演示的,Claude 在盯 PR 这件事上也做得很好。

So we definitely have Claude handle all the styling and lint and PR feedback requests, even catching some bugs and fixing them before it even does a full commit, and also adding tests.

所以我们让 Claude 全权处理所有的样式、lint 和 PR 反馈请求,甚至在正式提交之前就抓出一些 bug 并修掉,还会帮忙补测试。

That's really what we've leaned heavily into Claude for.

这些正是我们大力依靠 Claude 去做的事。

But where I still definitely want a human is that expertise. It's all about trust but verify.

但我仍然坚持要有人介入的地方,是那份专业判断。一切都是"信任,但要核实"。

So for example, legal review — I always definitely want to make sure I'm getting my legal partner still.

比如法务审查——我始终坚持还是要拉上我的法务伙伴。

It's about risk tolerance. So for example, trust boundaries and security-sensitive code, I still want to make sure I'm pulling in the experts.

这关乎风险承受度。比如信任边界和安全敏感的代码,我还是会确保把专家拉进来。

The other area, which is kind of fun, is also that product sense and taste.

另一个领域,也挺好玩的,是产品的判断力和品味。

One of the fun things I like to use Claude for is that I like to decorate Claude for the holidays or the seasons.

我喜欢用 Claude 做的一件趣事,就是给 Claude 按节日或季节做点装扮。

And I remember last holiday, I wanted to update Claude in the terminal to give him a little holiday theme.

我记得上个假期,我想在终端里给 Claude 换个样子,给他加点节日氛围。

And I asked Claude Code to turn Claude into a snowman. Claude wasn't that good at ASCII art in those days.

于是我让 Claude Code 把 Claude 变成一个雪人。Claude 那会儿还不太擅长 ASCII 字符画。

And I remember asking — that's where product sense really comes in — I asked my design partner, "Hey, can you review this for me?"

我记得我去问了——这正是产品判断力发挥作用的地方——我问我的设计伙伴:"嘿,能帮我看看这个吗?"

And she gave me such good feedback. She's like, "You turned Claude into the Mr. Peanut character."

她给了我特别到位的反馈。她说:"你把 Claude 弄成了 Mr. Peanut 那个形象。"

Because I was trying to make him snow-match. I was like, "Okay, I'm going to do something more simple."

因为我本想让他配上雪的感觉。我心想:"好吧,那我做个更简单点的。"

So Claude was like ice blue with snowflakes — but keep in mind that product sense as well.

最后 Claude 变成了冰蓝色配雪花——但别忘了,那份产品判断力同样重要。

"Which leads me to: what should my team makeup be?"

这就引出了一个问题:我的团队该怎么搭配?

"Because roles are blurring, Claude is augmenting."

因为角色的边界正在变模糊,Claude 在不断地增强每个人。

"I'll share with you, on Claude Code there's two profiles for engineers that I'd really heavily indexed on."

我跟你分享一下,在 Claude Code 团队,我特别看重两类工程师。

"One are like creative builders with product sense."

一类是有产品 sense 的创意型 builder。

"Usually you'll see these are, you know, the dreamers."

通常你会发现,他们是那种爱做梦的人。

"There will be a big sense of curiosity."

他们身上有一股很强的好奇心。

"They're really passionate about, 'Oh, here's a problem. Maybe I could ship a product that solves that problem.'"

他们会很兴奋地想:「噢,这里有个问题,也许我能做个产品来解决它。」

"But then there'll be a lot of iteration to make sure you're delivering a delightful experience."

但接下来要做大量迭代,才能确保你交付的是一种让人愉悦的体验。

"The other one is deep systems expertise."

另一类是有深厚系统功底的人。

"So when I first joined the Claude Code team, I noticed actually we were pretty good with product generalists and creative folks, but we were missing folks with like distributed systems expertise."

所以我刚加入 Claude Code 团队时注意到,我们的产品通才和创意型人才其实挺充足,但缺少懂分布式系统的人。

"And when you're building things like Claude Code remote to ensure we can run Claude's everywhere, you really still need those expertise."

而当你要做像 Claude Code remote 这样的东西、要让 Claude 能跑在任何地方时,你真的离不开这种专业能力。

"So whichever software engineering lead you're part of or supporting, think about those hard parts of where you might want to continue to still double down."

所以不管你带的或支持的是哪个软件工程团队,都要想清楚:哪些硬骨头是你想继续加码深耕的。

"But for sure, what I index less on is raw throughput, because thanks to the models, we just saw a lot more efficiency."

但有一点我越来越不那么看重了,那就是单纯的产出量,因为有了模型,效率本身已经高出一大截。

"The cross-functional gaps is another interesting one."

跨职能的空缺是另一个有意思的点。

"So for example, I remember that I wanted to do an update to how we do survey responses, but I didn't have a dedicated content designer to work with me."

举个例子,我记得有次想改一下我们收集问卷反馈的方式,但身边没有专门的 content designer 配合我。

"And I'm an engineer, my writing skills are quite terrible."

而我是个工程师,文笔相当糟糕。

"I struggle to write things in a short and succinct form."

要把东西写得短小精炼,对我来说很费劲。

"I don't want the surveys to overload you all in the terminal, because every line space is really important."

我不想让问卷在终端里把大家淹没,因为每一行的空间都很金贵。

"But that's where Claude — like in the old days I would be, 'Who is a group of content designers I can work with?' and I can kind of change things back and forth."

这正是 Claude 派上用场的地方——放在过去,我会到处找:「我能找哪几个 content designer 来一起改?」然后来回反复地磨。

"But now Claude has really helped me to augment that role and was a really good content design partner for me to make sure the verbiage is good and succinct."

但现在 Claude 实实在在地帮我补上了这个角色,成了一个很棒的 content design 搭档,帮我把措辞打磨得既准确又精炼。

"And on the flip side, I'd say our PM codes a lot, which is really fun to see."

反过来也一样,我们的 PM 写起代码来也很多,看着特别有意思。

"So again, with Claude you have non-traditional coders now being able to do more engineering, but you also have engineers that can now lean in to do things that was traditionally not on the technical side but more of where the content or design is."

所以再说一遍,有了 Claude,非传统出身的程序员现在能做更多工程的活,而工程师也能反过来去做那些传统上不属于技术、而更偏内容或设计的事。

"So it's very interesting — I found that Claude and AI has really augmented roles kind of like all around."

所以很有意思——我发现 Claude 和 AI 几乎是全方位地增强了每一种角色。

"This was a spicy one."

下面这条比较辣。

"So I remember when I first joined Claude Code, everyone's like, 'Okay, you're going to grow the team by a certain amount.'"

我记得刚加入 Claude Code 时,大家都说:「好,你要把团队扩张到一定规模。」

"And I could tell recruiters were still using the typical, you know, 10 ICs to one manager, and then how do you start thinking about nesting?"

我看得出来,招聘官还在套那套老公式——10 个 IC 配 1 个 manager,然后再去考虑怎么往下分层。

"I really have leaned into keeping it really scrappy."

而我是真的倾向于把团队保持得精瘦、能打。

"And actually maybe I would step back — whether it's at Anthropic or Meta or Microsoft, whether it was Visual Studio or Facebook Marketplace or AR/VR devices or Claude, what I have found to really help me ship great product is heavy heavy heavy dogfooding."

其实我想往回退一步说——不管是在 Anthropic、Meta 还是 Microsoft,不管做的是 Visual Studio、Facebook Marketplace、AR/VR 设备还是 Claude,我发现真正帮我交付出好产品的,是极度极度极度地 dogfooding。

"Especially for leaders, where nowadays it's actually fun — I can still be in the code, but for a while it wasn't."

尤其对 leader 来说,如今这件事其实挺有乐趣——我现在还能动手写代码,但有那么一阵子是做不到的。

"And when you can't get your hands on the code, I would always make sure I would make time so that I'm actually using my product days in, days out."

而当你没法亲自碰代码时,我总会特意挤出时间,让自己日复一日地真正去用自家产品。

"And so that's why I wanted every manager in Claude Code to start out as an IC first."

这也是为什么我希望 Claude Code 里的每一位 manager 都先从 IC 干起。

"And also to earn some street cred with the team and really learn how to be an effective engineer."

同时也是为了在团队里挣得一些威信,真正学会怎么当一名顶用的工程师。

"And then I really structured the org to be as flat as possible, because I want us to be super agile."

然后我把组织尽可能做扁平,因为我希望我们足够敏捷。

"This was just my — my recruiters had some concerns, because I remember they said, 'You want to hire managers and they will start as an IC first. No manager would be interested in that.'"

只不过我的招聘官有些顾虑,我记得他们说:「你想招 manager,却要他们先从 IC 做起,没有哪个 manager 会愿意。」

"I'm like, 'Well, this is what dogfooding on the Claude Code team's about. And this is what I expect.'"

我就说:「这正是 Claude Code 团队 dogfooding 的意义所在,这也是我的要求。」

"And if someone's not interested, it's better for us to do an earlier separation."

如果有人不感兴趣,那不如趁早分道扬镳。

"But also, there's no way I would have been able to ramp or be able to do code, because my time is — there's just a lot of context switching without Claude."

但反过来说,要不是有 Claude,我根本不可能上手、也写不了代码,因为我的时间会被大量的 context switching 切得稀碎。

"And so for those of you in the room who are managers, I really encourage y'all to kind of lean in."

所以在座各位当 manager 的,我真心鼓励你们也都下场试试。

"I'll be honest, for a long time at Meta, I would still try every year to do one PR a year, just to — but the workflows would always change."

老实说,在 Meta 那段很长的日子里,我每年还是会硬着头皮做一个 PR,就是想保持手感——可工作流总在变。

"Like all the internal tools — by the time I learn one command, it would have changed."

就拿那些内部工具来说,我刚学会一个命令,它就又改了。

"Nowadays, I don't even remember git commands. I just always ask Claude to help me out with all of that."

现在我连 git 命令都记不住了,这些事我一律交给 Claude 帮我搞定。

"Now with sharing, what becomes your new source of truth?"

再说共享这件事——什么会成为你新的 source of truth?

"So for example, on our team on Claude Code, the code is the source of truth."

比如在我们 Claude Code 团队,代码就是 source of truth。

"That's why I would go back to when I'm answering customer requests — I just have my desktop Claude with desktop Claude Code, and then I have all my local repository, so I can actually answer a lot of customer questions."

所以回到我处理客户请求的场景——我手边就一台桌面版 Claude 配上 desktop Claude Code,再加上我本地的全部 repository,这样我就能解答大量客户问题。

"So for us, just having that code base be the source of truth also prevents some of the lag that you might have had before of how to keep the documentation correct with the code."

所以对我们来说,让代码库直接当 source of truth,也省去了过去那种滞后——你得费劲让文档跟代码保持一致。

"But I would say this is where it's like, do what makes sense for your team."

不过我得说,这一块还是那句话:怎么对你的团队合适就怎么来。

"So for example, if you still have a lot of really good specs, check those into the repositories, and then have Claude lean in to help — like, 'Hey, take a look at how to verify my code execution so it matches what I expect on the spec.'"

比如说,如果你手上还有很多很好的 spec,就把它们 check 进 repository,再让 Claude 下场帮忙——像是:「嘿,看看怎么验证我的代码执行结果,让它和我 spec 里预期的对得上。」

"And so how did we roll it out?"

那我们是怎么把它落地的呢？

"I had just gone through some of the norms that we had changed, and I think it's interesting, there's a blend of what do we mandate as team norms?"

我刚才过了一遍我们改掉的一些规矩，我觉得有意思的是，这里面是一种混合：哪些东西我们强制定为团队规范？

"Like make sure you're gaining alignment with the team."

比如确保你和团队达成一致。

"And then where do I really enable each pod — we have pods within Claude Code — to do what makes sense for them as well?"

然后又有哪些地方，我真正放手让每个 pod 去做对他们自己最合理的事？我们 Claude Code 内部是分 pod 的。

"So in terms of the balance, in terms of forcing function, it's align with the teams on the must-do."

所以从平衡、从 forcing function 的角度看，原则是：必须做的事，要和团队对齐。

"So these are a few of the core Claude Code team principles that we really live and breathe."

下面这几条，就是我们 Claude Code 团队真正身体力行的核心原则。

"And by the way, I'll go back to that growth mindset and always think about — is something still serving you?"

顺便说一句，我要再回到那个 growth mindset：永远要想一想——这东西现在还对你有用吗？

"We keep these up to date, so every few months we'll be like, 'Hey, is this still having the same effect, or serving the purpose that we wanted when we started it?'"

这些原则我们会一直更新，每隔几个月就会问：「嘿，这条还像当初那样有效吗？还在实现我们最初想要的目的吗？」

"So for example — and I should replace this slide — it's not every engineer uses Claude Code, that's obvious."

举个例子——这张幻灯片我其实该换掉了——写「每个工程师都用 Claude Code」太显然了。

"It's actually every Claude Code team member, including cross-functional partners."

真正的说法应该是：Claude Code 团队的每一个成员都在用，包括跨职能的合作伙伴。

"And actually, we all use Cowork quite a bit, too."

而且其实，我们大家也都挺常用 Cowork 的。

"Claudify everything you can."

能 Claudify 的，就全都 Claudify 掉。

"This is one thing where we're like, 'You know what's better than one of us doing it? Having Claude.'"

这一点我们的态度是：「你知道比我们自己来做更好的是什么吗？让 Claude 来做。」

"So always think about — is there some way for you to automate, whether it's verification, more shift left?"

所以要时刻在想——有没有办法把它自动化，不管是验证环节，还是更多地 shift left？

"Always be thinking, whatever you're doing right now — is there some way that Claude could actually help you do it?"

永远要想，你此刻在做的任何事——有没有什么办法能让 Claude 真的帮你来做？

"And the last one's my favorite — explicit permission to kill old processes."

最后一条是我的最爱——明确授权大家去砍掉旧流程。

"Because again, processes will pile on themselves."

因为还是那句话，流程会自己层层累积起来。

"As you're supporting your teams, it's really important to always get that fast feedback of what are the things we're spending a lot on."

在支撑团队的过程中，很重要的一点是：随时拿到快速反馈，搞清楚我们到底在哪些事上花了很多力气。

"I remember when I first joined Claude Code, we used to do stand-ups."

我记得刚加入 Claude Code 的时候，我们是开站会的。

"And then the team got a little bit big, so then we had a spreadsheet where we would all put up our weekly progress."

后来团队变大了一点，于是我们改用一张表格，大家都把每周的进展填上去。

"And then I was like, 'Oh wait, we should just do a skill, right? Like a stand-up script.'"

然后我心想：「等等，我们干脆做一个 skill 不就好了？比如一个站会脚本。」

"So then we can just run Claude and all of us can be much more kept aware of what everybody else is doing."

这样我们直接跑 Claude，每个人就都能更清楚地知道别人在做什么。

"So that's another example of — one day I remember spreadsheet, wait, 'Does this still make sense anymore?'"

这又是一个例子——有一天我看着那张表格，心想：「这玩意儿现在还有意义吗？」

"So always question and always look to defrag and kill old processes."

所以要不断质疑，不断去做整理、砍掉旧流程。

"What I want to make sure is that I leave a lot of room for pods to adapt."

我想确保的一点是：给各个 pod 留出足够的空间去自我调整。

"Each team really has a lot of high agency for how they do triage, or leverage Claude to do triage, any planning rituals or stand-ups, how they think about on-calls, and also which workflows to Claudify first."

每个团队都有很高的 agency，去决定他们怎么做问题分流、怎么用 Claude 来分流，决定各种 planning 仪式或站会怎么开，怎么看待 on-call，以及先把哪些 workflow 拿来 Claudify。

"So we don't usually mandate, 'Thou shalt automate this.'"

所以我们一般不会下命令说：「尔等必须自动化此事。」

"We have some suggestions and learnings, but always give room to your team, especially as they may be touching on different problem areas."

我们会给一些建议和经验，但永远要给团队留余地，尤其是当他们各自面对的问题领域不一样的时候。

"So if I zoom out, what were the three things I prioritized when I joined Claude Code that I felt has to make the biggest difference?"

那么拉远一点看，我刚加入 Claude Code 时优先抓的三件事——我觉得最能带来最大改变的三件事——是什么？

"Keeping the team as flat as possible."

第一，让团队尽可能扁平。

"Managers release pods of work, but really keep it agile."

管理者把一块块工作以 pod 的形式放出去，但要真正保持敏捷。

"So for example, on Claude Code and Cowork, we have one overall team mission."

举个例子，在 Claude Code 和 Cowork 上，我们只有一个统一的团队 mission。

"Because sometimes when you start creating pods, each pod then wants to set up their own mission, and then anytime you have to shift, it might take a lot of time to walk people through that."

因为有时候你一开始分 pod，每个 pod 就想立自己的 mission，结果一旦你要调整方向，可能要花很多时间一个个去跟大家解释。

"But it's really as flat as you can — I felt that served us really well."

但真正的原则就是能多扁平就多扁平——我觉得这一点对我们帮助非常大。

"The second, Claudify everything — better than you doing it; I see how Claude can help you."

第二，把一切都 Claudify——比你自己亲手做更好；我看到了 Claude 能怎样帮你。

"It really frees us up to do more of the harder work."

它真正把我们解放出来，去做更多更难的工作。

"And again, the processes, they do pile on, so I encourage you to work with your team to see what are the processes that you should be able to let go."

还是那句话，流程确实会越堆越多，所以我鼓励你和团队一起，看看哪些流程其实是可以放手砍掉的。

"So does it actually work?"

那么，这套东西真的有用吗？

"I can't go into the explicit numbers, but these are three general metrics you can look at as you're rolling out changes on your team that might start steering you to — yeah, this seems like it's starting to be successful."

具体数字我没法讲，但有三个大致的指标，你在团队里推行改变时可以盯着看，它们会慢慢让你判断——对，这事好像开始成了。

"The onboarding ramp-up time — that has dramatically reduced."

第一个，新人上手的 ramp-up 时间——这个大幅缩短了。

"That's an interesting thing, like how soon an engineer or a designer or a PM can start being effective on your team."

这点挺有意思，就是一个工程师、设计师，或者 PM 多快能在你团队里真正开始出活。

"The PR cycle time shortening — I think this one's interesting to double-click into a bit."

第二个，PR 的周期时间缩短——这一条我觉得值得稍微展开 double-click 一下。

"Because it might actually help you identify a gap that's not just in terms of lack of AI adoption, but where the rest of the pipeline might be struggling to scale."

因为它可能帮你发现一个缺口——不只是 AI 用得不够的问题，而是流水线的其他环节可能跟不上规模。

"So for example, as we're now putting in so much more code, sometimes can the product infrastructure and CI still keep up with the amount that engineers are checking in?"

举个例子，现在我们提交的代码量大了这么多，有时候你得问：产品基础设施和 CI 还跟得上工程师提交的量吗？

"So both of those should go down, but then what should go up a little bit is Claude-assisted commits."

所以前面那两项应该往下走，而应该稍微往上走的，是 Claude 辅助的 commit。

"For us, by default every commit is Claude-assisted — I don't think I've seen a non-Claude-assisted commit in probably the last four months or so."

对我们来说，默认每个 commit 都是 Claude 辅助的——大概最近四个月里，我都没见过一个不是 Claude 辅助的 commit。

"But hopefully these three things are metrics that you can take a look at in your team and see how it resonates."

希望这三项指标，是你能拿回自己团队里看一看、看它们是否对得上的东西。

"I would also say, outside of just number of commits, also think about the end goal."

我还想说，除了 commit 数量之外，也要想想最终目标。

"If you zoom out, what is the product that you're trying to make more delightful for users? Or what is the problem you're trying to solve?"

拉远来看，你想做的，是哪个让用户更惊喜的产品？又或者，你想解决的是什么问题？

"Because sometimes you see it on all the headlines — this company said X percent of code is now generated by AI."

因为有时候你会在各种头条上看到——某家公司宣称自己百分之多少的代码现在都是 AI 生成的。

"I think throughput is great, but really think about how you measure what it is that you're actually really trying to solve."

我觉得吞吐量当然很好，但你真得想清楚，该怎么去衡量你真正想解决的那件事。

"Like for example, we really want to make sure we're keeping an eye on quality and reliability, so those are some of the things that we're paying more attention to as well."

比如说，我们非常希望盯紧质量和可靠性，所以这些也是我们更加在意的东西。

"Okay, so my last section is to audit your own effort."

好,我最后一部分要讲的是:审视你自己的流程。

"I'll share with you transparently, actually, I still have a couple of questions that I'm still asking myself."

我跟你们坦白说,其实我自己也还有几个问题,到现在都还在问自己。

"The iOS and Android org is a very interesting one."

iOS 和 Android 这两个团队就是个很有意思的例子。

"Like when engineers can now more efficiently flex across different mobile platforms, does a more traditional way of having an iOS team and an Android team still make sense?"

当工程师现在能更高效地在不同移动平台之间灵活切换,那种传统的、把 iOS 团队和 Android 团队分开的做法,还说得通吗?

"That's something that I'm still kind of thinking through."

这个我自己也还在琢磨。

"How much do you push that fully automated review?"

全自动 review 你要推到多深?

"Again, when do you strike that balance between fast enough and we lost something important?"

还是那个问题:在「够快」和「我们丢掉了重要的东西」之间,你怎么找到那个平衡点?

"It goes back to the trust but verify."

这又回到了那句「信任,但要验证」。

"And what's interesting is you even might have heard today, earlier during Danella's talk, the model capabilities do keep improving."

有意思的是,你们今天可能也听到了,早些时候在 Danella 的演讲里就提过——模型能力一直在变强。

"So also, even if you might need to do more verify than trust for a certain section of work, that might also change with the next model."

所以,哪怕某一块工作你现在得多验证、少信任,等到下一代模型出来,这个情况也可能就变了。

"So it's why it's always important to re-evaluate."

所以这也是为什么,时不时重新评估总是很重要。

"And roles are blurring, so how do you make sure that everybody feels equally productive?"

而且各种角色之间的边界正在模糊,那你怎么确保每个人都同样觉得自己有产出、有价值?

"So if I were to leave you with one thing to take back, it's: pick your noisiest workflow."

所以如果要让你带一件事回去,那就是:挑出你最闹心的那个 workflow。

"And by noisiest workflow, it could be most expensive — something that you yourself might be dreading, or even your team might not really look forward to."

所谓最闹心的 workflow,可能是最烧钱的那个,可能是你自己一想到就发怵的那个,甚至是你团队都不太愿意去做的那个。

"And ask, is it still really serving what's the purpose there?"

然后问一句:它现在真的还在服务于它本来的目的吗?

"Actually, I'll share another funny story."

其实,我再讲个好玩的事。

"There was a team I was on where we used to have this weekly review."

我以前待过一个团队,我们每周都有一个例行的 review。

"It's a very expensive review, like 50 people in this large room."

这个 review 成本特别高,差不多 50 个人,挤在一个大房间里。

"And then I noticed everybody's on their laptops."

然后我注意到,所有人都在低头看自己的笔记本电脑。

"Except for when it's their time to give a status report, and then they pop their head up, say the status, and then go back down."

只有轮到自己汇报进度时才会例外:抬起头,把状态说一说,然后又埋下去。

"And I'm like, this is a very expensive meeting."

我心想,这会开得也太贵了。

"And I just asked that simple question of why are we having it?"

于是我就问了那个最简单的问题:我们到底为什么要开这个会?

"And just that one question, everybody's like, yeah, it's true."

就这一个问题,大家都说,对啊,确实如此。

"And so we canceled it."

于是我们就把它取消了。

"So, that's why it's always important to think."

所以,这就是为什么停下来想一想,永远都很重要。

"And so I figured this might be something fun for you to take back and look at — see what's one piece of workflow that you might consider."

所以我觉得这件事你们带回去琢磨琢磨应该挺有意思——看看有没有哪一个 workflow 值得你重新审视一下。

"Can you either automate it, or maybe even, is it still serving its intended purpose?"

你能不能把它自动化掉?或者干脆问:它是不是还在发挥它本来该有的作用?

"So, with that, that's the end of our talk."

好,讲到这儿,我们的分享就到此结束了。

"Thank you. Thanks a lot for attending."

谢谢大家。非常感谢你们来听。

"I really, really for sure thought this room was going to be empty, so thanks for not having me by myself."

我之前真的、真的非常确定这个房间会是空的,所以谢谢你们没让我一个人在这儿干讲。

"And I'm around here today and tomorrow, so if you have any questions or want to chat more, feel free to introduce yourself."

我今天和明天都在现场,所以如果你有任何问题、或者想多聊聊,尽管来跟我打招呼。

"I'm happy to chat. Thank you."

我很乐意聊。谢谢大家。