本集简介
双语字幕
仅展示文本字幕，不包含中文音频；想边听边看，请使用 Bayt 播客 App。
迈克，欢迎来到节目。
Mike, welcome to the show.
很高兴能来，丹。
Great to be here, Dan.
见到你真好。
Good to see you.
对于还不了解你的人，你是Anthropic实验室的负责人，也是Instagram的联合创始人。
So for people who don't know you, you're the head of Anthropic Labs, and you're the co founder of Instagram.
今天我想和你聊聊Fable Five。
And today, I wanna talk to you about is Fable five.
Fable Five明天就要发布了。
So Fable five is dropping tomorrow.
我们今天录这期节目，是在发布前一天。
We're recording this the day before.
节目上线之后才会对外放出。
It's gonna come out after it drops.
但我真正想请你来，是想听听你用这个模型超过第一天之后的真实体验。
But what I really wanted to do is bring you on the show to tell me about what it's like to use this model beyond the first day.
我觉得，当这么强大的模型发布时，特别需要有人天天在用，告诉你它到底强在哪里。
I think when a model this powerful drops, it's so useful to have someone who's using it day in and day out to tell you, this is where it's powerful.
它真正带来了哪些改变。
This is how what it actually changes.
又有哪些地方其实没变。
This is what it doesn't change.
这样你才不会陷入那种AI狂热的误区。
So that you you're you kinda like don't you kinda don't get the same AI psychosis type thing.
你能真正想清楚，它到底该怎么融入你的日常生活。
You can actually think about, okay, like, this is how it fits into my life.
是的，完全没错。
Yeah, absolutely.
而且这过程本身也很有意思。
And and it's also just been interesting.
你知道吧，自从Fable发布前几个月，我们就一直在用这一系列的模型。
You know, we've had some, you know, models in this, you know, mythos class leading up to the Fable release, you know, for a couple of months now.
看到外界的人们怎么用这个模型，真的让人特别兴奋。
And it's I think it's very exciting to see how people will build with this externally.
不过你说得对，第一天的体验其实没那么重要，真正能看出价值的是用上几周之后。
But I think you're also right that day one impressions, I think it really comes from getting to use this over a couple of weeks.
我们之前也见过类似的情况，比如去年12月到1月用OPUS 4.5或4.6的时候。
I think we've seen that even with previous models, like the December into January usage, was an OPUS 4.5 or OPUS 4.6.
那时候特别关键，因为大家用久了才发现，原来自己之前根本没把模型用到极致。
So it was really important because people spend extended time on the model and then figure it out, oh, actually, I wasn't pushing it hard enough.
得再往前走一步。
I gotta go further.
得重新想想，这一代模型到底能做什么。
I gotta rethink what's even possible with this generation.
完全同意。
Totally.
我的意思是，我不确定。
I mean, I don't know.
我觉得Every内部有些人一直在用，他们都说：天啊，我好像得重新学一套技能才能用好这个模型。
I feel like there are people internally at Every who have been using it, who have been like, oh my god, I think I kind of need a new set of skills to use this model.
尤其是那些非技术背景、主要从事知识型工作的同事，他们根本不知道这东西能用来做什么。
And I think you can especially see this with people who are maybe more nontechnical internally and who are more on the knowledge work side of things where they're like, I don't even know what I would use this for.
而那些在编排智能体的人则说：天哪，我感觉有太多新东西要学了。
And the people who are orchestrating agents are like, holy shit, I feel like there's so many new things I need to learn.
所以我想问问你，你第一次用的时候和现在的感觉有什么不同？
So I'm curious for you, tell us about the difference between your impression when you first tried it and now.
是的，你说得对，适应工作流程确实是个关键点。
Yeah, I think that your point on adapting workflows is a really good one.
我说的不仅是工作流程，稍后我会细说，更重要的是，你该怎么重新思考这个模型的使用方式。
Quite literally workflows, I'll talk about that in a second, but also just in terms of, like, how do I, like, think about usage of the model?
因为说实话，当时时间点挺巧的，正好是我从首席产品官转到实验室，重新回到开发者角色的时候。
Because, you know, at first, the the timing was interesting because it kinda coincided with me transitioning from CPO into labs and going really back into into builder mode.
大概过了一个半月到两个月，我们内部才第一次用上这种模型。
I think it was about a month and a half or two months into that that we first had, you know, one of these models available internally.
我当时坐在那儿，感觉自己像个新手一样，因为现在我用来提示模型的方式，甚至对任务拆解的思路，都显得过时了。
And I I sat there, and I and I was like, I feel like a total newbie again because I feel like the way that I am prompting or even thinking about decomposing a task is really out of date now with this model.
不只是提示方式，连思考问题的时间尺度和交互模式，我觉得也都得跟着升级。
Like, it's no longer and it's even thinking about the time horizon or the sort of, like, interactivity model, I think, has to evolve as well.
以前可能是：我有个功能点子，咱们先做吧？
Like, going from, I think, early on would be like, I have an idea for this feature.
现在完全不是这样了，对吧？
Can we start by doing like, absolutely not, right?
现在是：太好了，让我把意图表达得更清楚一点。
To, great, like, let me express more of the intent.
我记得三月四月的时候，还特别惊讶。
And then just being I I remember, like, you know, know, March, April be like, wow.
只给一次提示，效果就已经非常惊人了。
On the one shot, it's already incredibly impressive.
但它还能理解我们接下来要怎么推进，以及整个大背景。
But then it also understands the intent around how we're gonna evolve this and understands, like, the global context as well.
所以我觉得到目前为止，这种变化特别有意思。今天早上我跟人聊天时还提到，我现在想工作的方式完全不一样了。
So I think that's been a really interesting evolution till now where, you know, I was I was talking to somebody this morning where, you know, I think about doing work.
我坐飞机的时候就想，嗯，这部分工作其实完全可以远程完成。
I had a flight, and I was like, okay, can do most of this work remotely.
我甚至都不担心网络会断，因为我知道只要把上下文指令设好，比如用循环反馈机制，它就能自己把事情搞定。
And I don't even worry that, like, the Wi Fi is gonna drop out because I know that if I set up the right, you know, context instructions, like, flash loop, you know, I'll see it it'll see it through.
过去两个月，我经常在睡前跟Claude说晚安，给它布置一个挺复杂的任务，然后早上醒来发现——通常凌晨两点它就干完了，接下来四个小时它就在那儿默默收尾。
And and I think my last two months have been full of a lot of times where I will, you know, wish Claude a good night, set it up on, like, a pretty complex task of something of this, like, monoclass, and wake up to you know, actually, it's usually done by, like, two in the morning, and I guess it just totals its thumbs for the next four hours.
但它真正厉害的地方在于，能自己把事情闭环，从困境里脱身，比如：
But, like, really impressive ability to, like, complete the swing, get itself out of the situation where it's like, okay.
行了。
Alright.
迈克让我连夜处理一个复杂的任务。
Well, Mike asked me to do this complex task overnight.
我卡住了，因为那个远程服务挂了。
I got stuck because this remote service went down.
我先给它搭个临时的后端框架应付一下。
I'm gonna write a, like, scaffolded, like, back end for it for now.
我会把这一步记下来。
So I'll, you know, I'll document that.
我会把整个流程走一遍。
I'll, you know, go all the way through.
我心里大概有数，这么做能推进到什么程度。
I have a, like, good mental model of, like, how far that's gonna get me.
等服务恢复了，我再把它修好。
And then when it comes back online, I'll fix it.
我会把这个情况记清楚。
I'll keep track of that fact.
最让我震撼的是，你真的能把这种级别的任务交出去，完全放心它最后会妥善处理。
It's just, like, it it is I think the most impressive thing for me is, like, you're just being able to, like, delegate that kind of level of task and just trust that the right thing will happen by the end.
当然了，你最终还是要检查结果，这中间还有一整套验证流程，我们其实应该好好聊聊，因为我觉得这是完成整个流程中非常关键的一环。
And, of course, like, you'll review the result, and there's still, like, a whole verification thing that we we can and should talk about because I think that's an important part of still completing the the the swing there.
但它确实逼着我重新思考，用这类模型时，什么才算真正高效？
But it's really forced me to rethink, like, what is being productive with one of these models look like?
我们之前聊过很久，当这些模型更像伙伴或同事时，会是什么样子？
It it is much more like, we've talked for a while about, you know, like, what is it like when these models are more of, a companion or a coworker?
现在我真觉得，它就像一个可以托付大量工作的团队成员。
And it really feels like now it's, like, a teammate that I can delegate, like, a lot of work to.
那你现在的日常操作流程是怎样的？
And what is your what is your day to day flow like right now?
我注意到一点，如果你直接丢给它一个大任务，自言自语地交代清楚，然后让它自己跑上几小时甚至过夜，这确实是我在用过的模型里最令人惊叹的。
Because one of the things I notice is if you if you just give it a big task and you monologue into it and you just like let it go for a few hours or overnight, it's like the most impressive model that I've ever tried.
但问题是，它太慢、太贵了，我总觉得不适合拿来处理日常琐事。
But you know, it's so slow and it's so expensive that you you I I feel like I don't wanna use it for day to day tasks.
所以你平时到底是怎么用它的？具体怎么安排？它在你工具栈里占什么位置，和其他模型相比呢？
So what is your actual flow like in terms of how you use it day to day, and where does it slot in versus other models?
是的
Yeah.
我也越来越多地在前期和它讨论架构设计的问题了。
I've ended up having a lot more architectural planning conversations upfront with it as well.
这成了另一个有趣的转变，我觉得所有模型在这方面都还有很大提升空间。我特别感激Instagram的经历——从最初那个用胶带绑在洛杉矶服务器上的简陋版本，一路发展到能扩展并最终整合进整个Facebook基础设施的过程。
So that's been, like, another interesting change where and I think this is an area that I think all models need to continue to improve, and I'm really grateful for the Instagram experience of having to, like, start, from our initial version that was, like, duct taped on a server in LA to, like, being able to scale it and eventually integrate it with, like, all of, like, the Facebook infrastructure.
因为你慢慢会培养出一种判断力，知道每个阶段该用什么样的基础设施抽象和复杂度才合适。
Because you kind of develop a sense of what what infra abstractions and complexity are appropriate for each stage of it.
但我并不是每次都会和Fable来回讨论，比如‘这个实现方式不错’。
And I I still don't always go back and forth with Fable where it'll be like, oh, this is a good, you know, implementation.
我是打算很快就把这个上线的。
Like, well, I I do plan on shipping this, like, fairly soon.
我觉得我们得考虑用不止一台服务器，这种来回的讨论挺重要的。
Like, I think we should probably think about more than one server and kind of like that back and forth Mhmm.
确实很重要。
Is important.
但很多这类规划工作，我经常会直接问它。
But, like, a lot of that sort of planning and I'll often actually ask it.
我意识到，Fable 在思考问题时能如此全面，尤其是在你和它一起做规划的时候。
It's kind of a the the thing I've realized is Fable can, like, be so sort of sort of complete in its thinking in terms of how much you are sort of planning with it.
比如，只要说‘你能帮我做个 HTML 页面，把我们刚才聊的内容展示出来，方便分享给团队吗？’——这其实很有用，哪怕只是个 Markdown 文档也好，不过我还是更喜欢图示。
Like, often just saying, can you just, like, make an HTML page, like, that represents what we just talked about so I can share it with the team is actually valuable, or even just a markdown document, but I like having diagrams.
所以这成了一个很有趣的用法：先和它一起规划、理清思路，再产出一份能让团队对齐的文档。
So that's been an interesting, like, use of, like, let's plan with it, let's think it through, and then let's have some sort of document that we can align the team on.
而且这种现象我在实验室和其他团队也见过，不只是 Anthropic：你确实能快速做出东西，但提前达成共识特别重要，哪怕先做个原型，再反推回架构设计，也完全可行。
Because and this is a dynamic I've seen in labs and just teams beyond Anthropic, which is you can build a lot very quickly, and forcing more of that early alignment, even if you do an initial prototype and then back it out into more of a sort of plan architecture, that works too.
这真的非常关键。
I think is really, really, really key.
而在这个过程中，人与人之间的沟通依然占据着核心位置。
And it ends being ends up being the place from, like, the human to human interaction still stays very, you know, very much part of the process.
之后，不管是 overnight 还是白天，让它分块执行这些任务，就变得特别重要。
And then from then on, I think, you know, either overnight or during the the the day, like, having it execute on those chunks of tasks is really important.
这意味着我现在要同时开更多的会话，因为我经常会觉得，嗯，有这两项工作要处理。
And it just means having a lot more concurrent sessions than I did before because I often will think, alright, there's there's these these two pieces of work.
我有时喜欢用一个长时间运行的云代码会话，让系统在后台分叉子任务，这样主流程还能保持响应。
I go back and forth between liking having one, like, very long running Cloud Code session and really asking it to do everything in background sort of forks sub agents so the main thread stays responsive.
但有时候我也干脆接受现实，今天就开五六个标签页，各自处理一些复杂又耗时的任务。
And then other times, just embracing, like, I'm just gonna it's one of those days where we're gonna have, like, five or six tabs that, like, tackle, like, long comprehensive work.
但我确实觉得，这种长期视角很重要，别着急，别焦虑。
But I do think that there's something to this, like, long horizon, like, don't, you know, don't worry.
我正在处理，你放心。
I'm I'm on it.
这需要点时间，而且得来回反复地沟通和调整。
It's gonna take me a while, and, like, more of, this back and forth.
我觉得这种工作模式，我们的产品也得好好琢磨一下怎么支持。
And that that modality, I think, is something that we'll have to figure out in our products as well.
我认为两种方式都得保留，它们之间其实会互相产生有意思的互动。
I think you wanna preserve both, and they they interact with each other in interesting ways.
而且，我的习惯通常是，我总喜欢至少保留一个上下文理解能力强、但响应速度又极快的 Claude。
And, like, my preference is usually I always like having at least one Claude that is high context, but also very, very fast response.
它的直觉反应就是，没问题。
And, like, its instinct is, great.
我会直接回答你，如果需要的话，我再启动一些后台任务。
I'm gonna answer you, and I'll kick something off if I need to.
如果不需要，那我就先等着，等下一个交互循环。
And if not, I'm just gonna, you know, hang tight and and and wait for the next kind of loop.
我觉得你说得对，我刚才就是在琢磨怎么解决这个交互方面的问题，或者说一些特别细节的地方。
I do think you're right that for the I'm just trying to fix this, you know, interaction question or something that's like very fine detailed.
比如 Fable 就会去深入思考这些细节。
Like, Fable will go off and think very hard about those things.
正因为这个原因，我觉得 Fable 是我第一个会去特意调整‘努力程度’参数的模型。我通常会想，好吧，我就需要微调一下 UI，那我就把它调到中等强度，看看效果怎么样。
And I think Fable is the first model where I've actually played more with the effort levels for that reason where I've been like, okay, this is I just needed to, like, tweak some UI.
我打算就把它设为中等，然后观察一下实际表现。
I'm gonna actually gonna fall like, you know, put it to medium or something and see how that plays out.
我在用 Opus 的时候没怎么这么做过，可能是因为它的调节范围感觉没那么宽，而 Fable 的范围确实挺大的。
I didn't find myself doing that as much with the Opus maybe because the range felt less, like, wide, where it really can feel quite wide with Fable.
那像那种快速提问呢？
What about, like, a quick question?
比如你正在外面跑，
Like, you're you're on the go.
你是会想到什么就问 Fable 什么吗？
Like, are you asking Fable, you know, random questions as they as they come to you?
感觉你像是在用火箭筒打蚊子，还是说你在来回切换？
Because it feels like you're using a rocket launcher to kill a mosquito or something, or are you flipping back and forth?
你问这个真有意思，因为我之前还真没这么想过。
It's so funny you asked that because I hadn't been.
你知道的，你心里想的是它在思考。
And, you know, you're like, it's thinking.
它正在很努力地思考这个问题。
It's thinking really hard about it.
但从上周开始，我就觉得算了。
Then since last week, I was like, no.
我问了它一些真正重要的事。
I was asking it something that, like, true.
其实我问Fable这些问题时，还有点不好意思。
I felt embarrassed actually asking Fable about.
大概是跟NBA总决赛有关的内容吧。
It was something like, probably something NBA finals related.
我就想，行吧。
And I was like, okay.
我把iOS应用换成了Sonnet。
I switched my iOS app to to Sonnet.
我当时就想，对啊。
I was like, oh, yeah.
你平时不都是用这个来快速提问的吗？
I you you use this all the time for fast questions.
这更像是数量级上的差别，那种感觉其实根本不是每秒处理多少令牌的问题。
It's like counter of magnitude, like, and feeling of, like and it's actually not even the the sort of, like, tokens per second.
真正关键的是回答背后花了多少思考，有时候答案根本不需要完全想透。
It's actually probably more around how much thinking goes into the answer, and sometimes, like, the answer does not need to be fully thought through.
所以我正在自己理清这个思路，我觉得这也是个很好的产品问题：一般来说，你不希望用户对这些选择想太多。
So, yeah, I I am I'm thinking myself through it, and I think this is a good product question for us too, which is, you know, in general, you don't want people to have to be thinking so much about these choices.
从长远来看，我们最好能归纳出一些更清晰、更容易理解的使用场景。
So ideally, what we can sort of coalesce around in the longer run is sort of, you know, maybe like some more bucketable use cases that are really grokable to people.
可能不同平台的情况还不一样，比如在iOS应用里，我大部分时候做的都是类似Fable这类任务，所以为每个平台固定一个模型选择或许是更好的做法。
Maybe it varies by surface where it's actually probably unlikely that most of the time with the iOS app, I'm doing Fable type tasks, you know, having a sticky model selection per surface might be the way to do that.
这我们需要从产品角度进一步探索一下。
And we'll have to sort of explore what that means from a product perspective.
但我确实有过这种感觉：这根本算不上一个值得纠结的问题。
But I've, for sure, have had the feeling of, like, this this is not a stable worthy question.
我该去问Sonnet才对。
I'm I should ask Sonnet.
能
Can
给我们看看你用它做过什么吗？
you show us something that you've built with it?
当然可以。
Yeah.
这次我们鼓励大家多用自己的账号，尤其是周末，特别有意思，毕竟我们有不少专门针对Anthropic的工具和资源。
So one of the things that we were we we did this this go around is we encouraged personal sort of like account usage for us, like, especially on the weekends, which was really fun because, you know, we have, you know, you can imagine, like, a lot of Anthropic specific, you know, tooling, etcetera.
但能暂时跳出来，纯粹用Claude Code来搞点东西，感觉真的很棒。
But it was really good to sort of step back and be like and just like, you know, pure Claude Code.
周末就干脆自己动手做个东西吧。
Let's like work on something over the weekend.
你是用终端应用还是桌面应用？
And you're in you're in the terminal app or you're in the desktop app?
这是个好问题。
That's a great question.
我大部分时间还是用终端应用。
I'm mostly still in the terminal app.
挺有意思的是，看我妻子——她不是专业工程师，更多是做UX设计和产品管理的——居然通过桌面版Claude Code爱上了它，我觉得这种方式帮她简化了不少抽象概念。
It's interesting watching my wife who's like not a professional engineer and more of a UX designer PM, like, really fall in love with Claude Code via the desktop app, and I think it's, like, sort of simplified some of the the sort of abstractions for her in that way.
但这次我还是用，是Ghosty还是Ghosttty？
But for this one, I was still is it ghosty or ghost t t y?
Ghosty，用的是终端应用。
Ghosty and and and the terminal app.
让我给你演示一下。
But let me show you.
这种事吧，每个人都有点自己独特的需要。
I this is one of those, like, everybody has some bespoke need around this.
我就想弄个好用的媒体追踪工具，当时正好在玩游戏。
Like, I wanted a good sort of media tracker experience, and I was like, you know, I'm playing games.
也在追剧。
Like, I'm watching TV shows.
我收到一堆推荐，但就想自己弄个东西，完全按我的需求来，贴合我平时的实际使用场景。
I get all these recommendations, and I just wanted to build something, like, that was personal to me and, like, sort of fit some of the use cases that I that I had.
我一开始定的两个最重要标准，一个是加东西得特别简单。
And, like, I would like the two biggest criteria that I started with was, like, one, like, really easy to add things.
你可以直接跟Claude说，让它帮你加。
And so, like, you could talk to Claude.
Claude会自动在所有内容里搜索，然后把最合适的信息推给你。
Claude does the gentic search over everything, and then puts the right thing to then.
它还能主动发现，比如新一季剧集上线了，或者某个游戏出了续作，自己去查相关信息。
And then also proactively, like, you know, there's a new season or a new, like, sequel to a game that it could go off and and and research those things.
大部分界面用的是Fable的一次性方案，那已经挺厉害了。
Most of the UI was like, you know, Fable one shot, which was which was already impressive.
但今年在实验室里，我一直在思考一个问题：怎么让像Claude这样的软件团队，更贴近实际的软件本身？
But then the the thread I've been pulling out a lot in labs this year is how do you sort of bring the software team, which is Claude these days, closer to the software itself?
所以这事儿，大概就是某个周六早上突然冒出来的想法。
And so this was, like, maybe, you know, Saturday morning.
这个周末全用来陪孩子了，所以很多工作都是断断续续做的——带孩子去徒步，回来接着干。
I had a full weekend with with kids stuff, so a lot of this was sort of kick off work, go do you know, go for a hike with the kids, come back, you know, continue to do the work.
有时候在徒步路上还会顺便看看工作进展。
Sometimes check-in on the work on the hike.
我本不该这样，但确实挺舒服的，切换到远程模式看看那边在忙什么。
I probably shouldn't, but, know, it was, like, nice to, like, pop into remote mode and and see what was going on there.
你知道的。
You know?
我一般不会这么频繁地这么做。
I try not to do that too much.
但我突然有个想法，就是——
But I had this idea around, hey.
我们能不能做个快速验证？我老说‘快速验证’这个词，用在这些模型上。
Like, could you could we, like, do a spike on I say spike a lot with with these models.
我想试试，如果能直接在软件内部修改软件本身，会怎么样？——其实这两个功能我都做出来了。
I'm like, can we do a spike on, what if you could actually modify the software from within itself, which is, you know and it was I built both.
一个是 React Native 版本，另一个就是网页版。
It was, a React Native version, and then this version, which is just the web version.
我之前已经弄了个聊天式的功能，你可以直接让 Claude 通过链接添加内容，说白了，我希望所有软件都这样——再也不用去翻菜单做任何事了。
So I already had like a chat type thing where you could sort of ask Claude to, you know, add things by URL, which is like, you know, I want every software to have this where I should never have to like navigate a menu to do anything ever again.
从很多方面来说，Dan，我当时是想把智能体原生架构的理念推到极致，就是——
And this is like, in many ways, Dan, like, the I was trying to distill the, like, agent native architectures to, like, its, like, fullest degree, which is,
对。
like Yeah.
还要让智能体能直接修改这个应用。
Also have the agent be able to modify the app.
所以，智能体原生架构的第一步，就是这个产品里的每一样功能，都能通过智能体访问，都有对应的工具调用之类的。
So, like, maybe, like, phase one of agent native architecture, like, every single thing in in in this product is, you know, accessible from the agent and and and, like, has tool calls, etcetera.
这将来大概会成为基本要求。
That's, like, you know, hopefully becoming table stakes.
可惜现在大部分软件还做不到，但能实现真是太好了。
It was sadly not in a lot of software, and it's great.
我当时就想，这到底是个啥体验？
Because I was like, what's that like?
因为有人推荐过，巴西有个剧讲的是戈亚尼亚的放射性事件。
Because somebody had recommended there's a in Brazilian, there's, a show about radioactive stuff in Goiania.
我一时想不起名字，但Claude一下就给找出来了。
I did not remember what it was called, and Claude was able to figure it out.
这比自己瞎琢磨可强太多了。
It was like so much better than being like trying to figure that intuitively.
接下来我更感兴趣的是：如果软件能自己实时修改自己，那会是什么样子？
But then the next step I was interested in is like, what would it mean to actually be able to modify the software from itself on the go?
所以你长按这个聊天窗口，我做的、Claude帮着做的，就是用我们托管的智能体来接收编辑请求，然后让你预览改动。
And so if you long press this little chat thing, so what it actually what I built, what Claude built, was a way where it used uses our managed agents to basically take on, like, edit requests, and then you can preview them.
我这里用了Versal的实时预览功能。
And I used, like, the Versal Live preview thing here.
这个功能居然一次就能搞定，真的很棒。
This like, this whole, like, feature was also one shot, was really cool.
哦，是的。
Oh, yeah.
后来我慢慢又加了一些功能。
And then I just added to it over time.
不过，如果你愿意，它确实能显示一个简单的改动对比视图。
But, know, it was like actually does like a little diff view if you wanted to.
你可以进入管理代理的对话里，看看它具体做了什么。
You can go into the manage agent conversation and see, like, what it did.
但我几乎从不看，因为说实话，我对这段代码的质量或者长期可维护性根本没那么在意。
Although, I almost never do because, again, it's like, especially don't particularly care on, like, the code quality of, like, or the, like, long term attenability of this software.
你也能看到它在这里还开过一个会话。
You can see that it had a session in here too.
但整个过程真的特别有趣。
But it's been really fun.
所以我经常在路上用它，比如前几天有个功能需求，我说：‘iOS原生版的悬浮按钮位置太靠下了，但在网页版上就刚好。’
So I'll be using it on the go and say, like, you know, I had a feature request the other day, like, oh, like, the floating action button was too low on native iOS, but it was okay on on there.
你能自己去搞定吗？
Like, can you go off and do it?
它真的完成了。
It did it.
现在用 Expo 工具链做起来特别有意思。
It was really fun with some of the, like, expo tooling now.
它居然能直接在手机上实时刷新，那种感觉真的很棒。
It actually, like, live reloaded on my phone, which was also, like, a really cool kind of kind of feeling.
但说到底，这东西真需要做成能服务百万用户的生产级应用吗？
But it was just like, you know, does this thing need to be like a, you know, production level thing that's gonna go to a million users?
不需要。
No.
但能有个东西，不只停留在周末玩一玩，而是能边用边改，形成完整的闭环，这种感觉特别好。
But it felt really good to have something where I felt like it didn't have to stop at just the weekend, and I could keep working on it just by using it and having this, like, kind of end to end close thing.
我觉得这很好地体现了 Fable 的构建能力，也反映出你我都在思考的问题：Claude 怎么才能真正融入软件，而不仅仅是作为使用工具。
So I felt like this was a good manifestation of both, like, Fable's building ability, but also, like, I think a lot of both of I have been thinking both you and I have been thinking about, like, how does Claude embed itself and, like, into software beyond just even the usage side of things.
这真的太棒了。
This is really cool.
我想让大家都明白，这东西已经做出来了。
I want people to understand, this has been built.
你完全可以做出类似的东西，也许不用搞自修改那部分，但做个这样的项目，做个十年、二十年都没问题。
You could build something like this, maybe not the self modifying part, you could build something like this for ten years or twenty years or something like that.
但现在开发的成本已经大幅降低了。
But cost to build has gotten dramatically lower.
想想看，当年做Instagram的时候，要花多少钱，跟现在比呢？
So think about how much it would have cost to do this in the Instagram days versus now.
你能帮我们理清一下，这中间到底发生了什么变化吗？
Can you help us understand how that has changed?
是的。
Yeah.
我确实经常回想起那个年代，那时候我觉得自己已经是个挺高效的程序员了。
I think and I think about this a lot when I think back to that that time as well because, you know, I I thought of myself as a very productive programmer in the early Instagram days.
我当时特别投入移动开发，对要做什么心里很清楚。
You know, I was, like, really into mobile development, and and we had, like, a good clarity of of things.
从想法到做出一个完整的产品，那时候还得熬上好几天通宵，我本来就是那种睡到中午、熬到凌晨四点的人，根本没法兼顾家庭，所以后来我不得不改掉这习惯。
And I think the the gap from idea to fully realized version of, like, some complete product, like, you were still looking at, you know, four ish days of kind of my all nighters, which was, like, my natural state is up till four, you know, sleep until noon, which is not conducive to family life, so I've had to shift.
但那确实是我当时做东西的方式。
But that was, like, my my building thing.
我把那叫Instagram第一版，可能功能比这个还多一点，但也没多出多少，就是我连续熬了五天通宵，前后端都自己干，凯文负责做最初的滤镜，才把东西搞出来。
Yeah, I call it, you know, Instagram v one, which, you know, probably had more features than than this thing did, but not by an order of magnitude, was, like, five days of all nighters, me working on, like, the sort of front end and and back end, and Kevin working on the initial filters to get that that out.
而且这背后也是我之前好几年做iOS开发积累下来的经验。
And and this is also, like, you know, like, built on already, you know, many years that I've been working on on on iOS pieces as well.
后来产品上线后火了，我们有很多想法想继续迭代，但当时主要精力都用在维持服务器不崩，或者只加一个新功能，比如标签功能就得花一周，可还有那么多其他事等着做。
And then the iteration, you know, I think a lot about what we were gated on after that launch when things went well was we had all these ideas for where to take it, but we were just trying to keep the site up or we were just trying to, like, add the one incremental feature and, you know, hashtags take a week to build, but then there's, like, all the things that you wanna continue doing on it as well.
所以我觉得一方面是开发周期变短了，想法、概念和迭代本身还是需要时间，另一方面是你现在能快速在已有基础上不断调整，这真的很棒，也特别符合现在这种灵活开发的节奏。
And so I think it's both that shortening of time, like there's still the time required for the idea and the the concept and the iteration, and then the other piece, which is the you can then iterate on what you have, and I think a really I think really fun, but also like very, you know, sort of in the float kind of way.
现在呢，我作为一个专业的软件工程师、初创公司的创始人。
And then, you know, if now this is me as a professional software engineer, sort of startup founder.
除此之外，如果你有这个想法，我见过好多人经历这种情况：他们会想找家咨询公司来接手，但现在这过程太丢信息了，我到底想要什么，你知道吗？
Beyond that, if you had that idea, you know, and I saw multiple people go through this, like, was like, well, I'll try to find maybe consultancy that will take this on, but like now there's like, it's a really lossy process like, what I want it, you know?
对。
Yeah.
难道他们要为这个项目融资吗？
Is it they're gonna raise money for it.
我觉得最让人兴奋的是，这些模型不仅越来越自主，还在不断缩短意图和实现之间的距离，这改变了非技术人员的创造能力。
And I think that the thing that I think is, like, the most exciting part about these models getting not just more autonomous, but again, closing that gap between intent and execution is what I've seen it do to people's ability to build who are not, like, builders.
这些模型的发展轨迹是这样的：像Fable这类属于同一类别的模型已经出现了，未来还会有更便宜、更易获取的模型普及开来。
And the trajectory of these models has been, you know, if something Fable, you know, of this general mythos class is, like, in that class of models, and eventually, you know, models of you know, that are cheaper and more accessible to to other folks become available too.
随着这个过程推进，我觉得它正在打开无数可能性——前几天我就收到一条消息。
And, like, as that process happens, like, I just think it is just opening up so many like, I got a ping the other day.
如果你没看出来，我确实对这些东西特别兴奋。
I get very excited about the stuff, if you can't tell.
是公司内部的人发来的，我们之前给他们搭了个内部工具，把Fable和一些内部MCP系统整合在了一起。
From somebody internally, and we had built them an internal tool that kinda combined Fable and, like, access to some internal MCPs.
她说，这还是她这辈子第一次，她做招聘工作的。
And she said, like, it is the first time in my life and she works in recruiting.
她觉得，第一次感觉到，脑子里想的东西和现实中能实现的东西，就这么贴在一起了。
She's like, the first time in life where, like, I feel like the thing that's in my head and the thing that exists in the world is now, like, they're right next to each other.
我就能直接把它做出来。
Like, I can just do it.
以前可不一样，我记得四五年前，如果有人想要个工具，要么将就着用，要么得去找内部的开发人员，可那些工程师手头还堆着五十多个需求呢。
And it was, like, very like, a meaningful moment to her because prior to that, like I mean, I remember these days these days were five years ago or four years ago where that person, if they wanted a tool, would have to either make do or try to get an internal tools engineer that probably was overloaded with 50 other, you know, requirements.
但现在，他们简直玩得不亦乐乎，自己动手做东西。
But instead now they, like, are just having the time of their lives building.
我觉得，这确实让人充满希望。
And I think that is I think that's cause for a lot of, like, hope.
因为我不觉得人类的创造力和可能性是有限的。
Because I don't think that human capacity for creativity and what's possible is enormous.
而我们最好的地方，就是让更多人能亲眼看到，自己的想法变成实实在在的东西。
And I think, like, at our best, we are basically expanding the number of people who can then see that through to something that feels real.
我完全同意。
I totally agree.
但我心里确实有个疑问，我想可能也是不少听众心里的疑问。
But I do think that there's a question in the back of my mind, and I think it's probably gonna be in the back of the minds of some of the people listening.
所以我想问你，基于你刚才说的这些，软件工程是不是已经结束了？
So I wanna ask you, given everything you just said, is software engineering over?
是的。
Yeah.
我觉得软件工程已经不一样了。
I think software engineering is different.
它发生了翻天覆地的变化。
It is like dramatically changed.
如果几年前在Instagram那会儿你问我什么是软件工程，我可能会说……
And as I I as I probably would have defined it if you had asked me around the Instagram time, like, what is software engineering?
嗯，大概就是这样。
I'd probably say, alright.
就是说，得认真思考那些难题，琢磨架构，然后花大量时间在 TextMate 这样的编辑器里。
Like, thinking through the hard problems and, like, thinking about an architecture, then, like, spending a lot of time in, you know, like, TextMate.
我不确定现在还能不能这么干了。
I I don't know what that can be.
但你知道的，现在就是用文本编辑器或者 Xcode 来改代码，再看看 Rails 的教学视频。
But, like, you know, like, text editor, you're gonna edit those things or Xcode, you know, and Watching Rails casts.
你懂的。
You know?
是的。
Yeah.
没错。
Exactly.
对。
Right.
没错。
Exactly.
还有，得搞懂 Django 的 ORM 层那些复杂的细节，结果一上线就冒出十五个 bug。
And understanding the intricacies of Django's, like, ORM layer and then, like, 15 bugs after you deploy it.
这些内容其实和产品管理的其他部分高度重叠，差别大得惊人。
Like, so much of that is radically different and collapsing into other parts of, like, product management.
我觉得这种产品管理的分工，不光是我们团队，现在整个行业都变得越来越模糊了。
And I think that sort of, like, PM end split, I think you guys see it even on our teams has become much more diffuse.
这变化真的太大了。
That's radically changed.
但我觉得，如果把视野放宽一点，别只盯着程序员的角度，而是看整个软件生产或开发的过程，那还是非常关键的。
But I think the overall, like, maybe zoom out from software engineering and think about, like, software production or, you know, software development, but not in, like, just the pure developer case.
我觉得这种整体的思路，现在依然充满活力，而且必不可少。
I think that is, like, alive and well and and and essential still.
所以我觉得，我们现在正处在一个关键的节点上。
So I think that it that is the moment that I feel like we are in.
我觉得 Fable 就是朝这个方向迈出的又一步，但我不会说这是最后一步。
I think Fable is another step on the direction of and I'm not gonna call it the fall final step.
当然，还是会有很多事情发生。
Of course, a lot will still happen.
但我觉得，至少在信任模型能持续把事情推进下去、甚至合理地架构系统这方面，我已经非常信赖它了。
But, like, I think a pretty significant step in terms of, like, the trust, at least I end up placing in the model in terms of its capacity to see things through and even, you know, architect things reasonably is quite high.
这部分感觉不会彻底完成，但已经差不多了。
So that part feels like it is it is not ever gonna be done, but it is pretty pretty done.
对吧？
Right?
它已经走得相当远了。
Like, it it's gone really far.
但我认为，整体上那种‘你到底需要什么’的工艺感——
But I think that the overall sort of craft of the what needs do you have?
你到底在产出什么？
Like, what are you putting out?
它真的够好吗？
Like, is it actually good?
我觉得这依然是个特别有人情味的工作。
I think still a very human endeavor.
但我也能感觉到，这种转变并不是毫无代价的。
But I also sort of can see that that is not a transition that is pain free in a way.
很多人其实特别享受亲手打造的那份匠心，我以前也特别迷恋这种感觉——我居然把这个问题解决得这么漂亮。
Like, I think there are plenty of people who love the craft of, like, actually putting and I used to love stuff like, I solved that problem so elegantly.
你可能会梦到代码，要是你有过这种经历：半夜做梦都在想手头的项目，早上一睁眼就兴奋地想，我终于找到解决这个问题的完美方法了。
You would dream about code, and if you ever had that experience of, like, you dream about the thing that you're working on, they, like, wake up in the morning and be like, I figured out how to solve this thing really elegantly.
这种感觉确实已经过去了。我跟一些顶尖工程师聊过，他们也都有种失落感，但同时又忍不住感叹：天啊，我现在能干的活儿简直多到离谱。
And and that for sure has has has passed, I think that there's, you know, there there's there is a feeling of loss, I think, in some of the, like, better engineers that I talk to, as well as the feeling of, oh my god, but I can do insane amounts of work now.
这两件事是同时发生的。
Same time.
所以，我们脑子里其实一直同时装着这两种想法。
So we're holding both ideas in our heads at once, I guess.
我觉得这才是最关键的部分。
Which I think is the most important part of this.
对这种事感到失落和兴奋，其实都很正常。
Like, it's normal to feel sadness for that kind of thing and excitement.
但我很好奇。
But I'm curious.
我们就假设软件工程依然充满活力吧。
Let's just take the thesis of software engineering is alive and well.
在Anthropic这家公司里，这具体是什么样子的？
What does that actually look like inside of Anthropic?
是的。
Yeah.
我觉得有好几条这样的观点。
I think there's there's a few thesis.
我觉得还是有人在认真打磨代码，可能我得从完整的开发周期里抽身，但日常工作中，我可能还是会兼顾两者。
I think there's still the the crafting of well, I gotta take it off from, the full software development cycle or, like, maybe what I see on a day to day, maybe I'll do a little bit of both.
对。
Yeah.
但我认为还是有很多事情，我们大家都聚在一起讨论了。
But I think there's still a lot of, you know, we all got together.
我们聊了接下来想怎么改进协作，现在我们已经把这些工作分成了不同的责任领域。
We we talked about the next way we want to, you know, evolve co work, and now we've kind of broken it down into areas of ownership.
我觉得这依然非常重要，因为作为一个人，你掌握着一些超越Claude的背景信息。
I think that ends up still being quite important because there is still context that you hold as a person that is sort of beyond Claude.
对吧？
Right?
比如，这个产品的真正目标到底是什么？
Like, what is the actual intent of this product?
它现在进展得怎么样？
How's it going?
我们还需要了解哪些即将上线、可能会有深度整合的其他产品？
What do we need to know about the sort of other products that are coming down the pipeline that are going to be integrated in some interesting way?
所以我觉得这一块依然非常关键。
So I think that aspect is really important still.
所以我们虽然每个人都有多个Claude，但在Anthropic，每个人——至少目前的工作方式下——依然负责某个具体的产品模块或领域，我们称之为DRIs，也就是直接责任人。
And so, you know, though we have many Claude's to each human, each human, at least the way we've been working on Anthropic, still kind of has, you know, we call them DRIs, like directly responsible individuals, still has like a DRI ship over some part of the product or some area.
我觉得这种模式还会持续一段时间，因为光靠大家共同提升团队表现还不够，还得有人专门负责某一块。
I think that'll be the case for a while because I think there is value in not just this distributor, like, we should all make the cohort better, but instead, like, alright.
我会仔细琢磨团队在某个具体任务上的表现，虽然我们尽量减少会议，但它们还是不可避免地出现，同步和对齐的对话也一直存在。
I'm thinking through how cohort does at this particular task, and there's still a lot of, you know, the we try to keep meetings minimal, but they they still emerge and you still have these kind of alignment conversations.
很多工程师现在发现，他们各自都搭建了一套工具，虽然我们在产品层面某个时候可能已经解决了这个问题，但大家还是各自搞了一套自己的方式。
Then, like, a lot of that sort of asynchronous delegation, I think what many engineers here have now found is they've they've all built and I think we solve this at some point at, like, a broader product level, but they've all built some version of, alright.
我会建个仪表盘，追踪所有Claude的进展、哪些任务在等我处理、哪些代码审查需要我关注——因为不管是人还是Claude，都给我回复了。
I'm gonna now, like, create a dashboard of where all my Claude's are doing and what's waiting for me and which pull requests, like, need my attention because, you know, either a human or a Claude Code reviewer got back to me.
这种对工作本身的元管理，我觉得我们以后会标准化一部分，但总有些部分会因人而异，就像每个人整理桌面或安排窗口的方式不一样，大家组织工作的方式也各有不同。
So there's a lot of that sort of meta maintenance of the of the work that I think, again, I think we'll standardize some, but I think some of it will always be a little bit bespoke to the way each individual likes to work just in the way that people organize their windows, now they organize their work.
另外，还有对线上系统实际运行方式的理解。
And then there is, I think, also the understanding how things work in production.
我觉得这又是模型下一步需要突破的几个方向之一。
And I think that is another, like there's a few, like, next frontiers, I think, for the models.
我觉得Fable在这一点上取得了不小进展，但还有很多工作要做，就是理解代码上线后会发生什么。
And I think one of them that Fable does, you know, make significant strides in, but I think there's there's more work needed here is understanding what happens to code after it gets deployed.
你知道的。
You know?
因为总会出问题。
Because there's incidents.
之前一切运行得好好的，结果某个网络连接突然断了，这根本不在我们平时考虑的故障范围内，结果Instagram好大一部分都瘫了。
There's, you know, this was all working well, but, like, this network link got cut, which is not in your usual failure mode, and, like, it manifested, like, so much of Instagram.
2012到2016年那会儿，我们天天都在处理这种事，不断扩容。
Like, 2012 to 2016 was, like, dealing with that and scaling things up.
所以工程师在这个环节依然至关重要，积累应对事故的经验、学会保持冷静、收集信息、先解决眼前问题，再回头处理长期修复，这些都还是必不可少的。
And so that role of the engineer still remains really key, and I think getting the the reps in around incident response and understanding how to stay calm, gather data, like, remediate what's immediate, but then, like, go off and and and work on on on longer term fixes, like, still a necessary part of it.
我还在想，是不是还有其他值得注意的地方。
And I'm trying to think if there's any, like, other pieces that are that are notable as well.
我觉得最后一点想说的是，我现在特别看重工程原型所扮演的角色。
I think what's maybe the last the last thing to say is I really like the role that the engineering prototype now plays.
得把原型和正式版本区分清楚。
You have to be clear when it's a prototype versus not.
以前常说代码能说服人，但我一直不太喜欢这种说法。因为会写代码的人就能直接动手做，但凭什么他们就该默认赢下争论呢？
But, you know, the old phrase was like, code wins arguments, and I never, like, loved that because, like, kind of sort of the person that could code could go do it, but actually, like, why should they necessarily win an argument by by by default?
不过现在挺有意思的，有时候我们对产品方向会有分歧或争论。
But actually, it's been really cool now where sometimes we will have some disagreement or some sort of debate about where to take a product.
通常这时候产品经理会说，行吧。
And often, it's the PM that will say, alright.
我刚试了一下，结果卡顿得厉害，简直没法用。
I just tried it and, like, jank in, like, these eight ways.
不过你看，它确实展示了这种方案是怎么跑通的，也能引出不少有意思的讨论。
But, look, it actually shows, like, how this could work, and that that can open up some some interesting pieces of conversation.
所以，几乎所有的东西跟六个月前相比都大不一样了。
So almost all of that is quite different than it was six months ago.
特别是在并行处理方面，以及我们对这类高阶工作抽象的需求上。
Think especially at the level of parallelism and the level of need for these kind of higher order abstractions of work.
但我觉得没变的是那份责任感。
But I think what hasn't changed is that ownership.
我们很多人都在把AI投入生产，这确实提升了效率，但也带来了焦虑。
Lots of us are shipping AI to production, which is great for productivity, but it also comes with anxiety.
你改一改提示词，换换模型，调一调参数，测试时一切看起来都正常。
You tweak a prompt, swap models, adjust parameters, and everything looks fine in testing.
于是你就合并了代码。
So you merge.
结果三天后，甚至更早，支持工单就开始源源不断地涌来。
And then three days later, or even sooner, the support tickets start rolling in.
AI给客户提供了意想不到的回答，你根本不知道问题是什么时候出现的，为什么会这样。
The AI is giving your customers unexpected answers, you have no idea when it happened or why.
Braintrust 是一个解决这个问题的AI可观测性平台。
Braintrust is the AI observability platform that fixes this.
它把评估和可观测性整合到了同一个工作流中。
It connects evals and observability in one workflow.
这样你就能看清生产环境中实际发生了什么，判断改动到底是让效果变好了还是变差了。
That way, see what actually happened in production and can measure whether changes made things better or worse.
追踪记录会完整呈现整个执行流程。
Traces show the full execution path.
评估标准定义什么是理想的结果，而实验功能让你在上线前直接对比不同提示词和模型的表现。
Evals define what good looks like, and experiments let you compare prompts and models side by side before shipping.
生产环境的追踪数据会直接流入你的评估数据集。
Production traces feed directly into your eval datasets.
每一次失败都会变成一个测试用例。
Every failure becomes a test case.
你能在持续集成阶段就发现退化问题，避免它们影响用户。
You catch regressions in CI before they reach users.
Notion、Stripe、Zapier、Vercel 和 Ramp 的团队都在用它来规模化交付高质量的 AI 系统。
And teams at Notion, Stripe, Zapier, Vercel, and Ramp use it ship quality AI at scale.
Braintrust 专为那些构建生产级 AI 系统、无法承受隐性退化的团队设计。
Braintrust is designed for teams building production AI systems where silent regressions are expensive.
它支持任何技术栈。
It's built for any stack.
提供了 Python、TypeScript、Go、Ruby 和 C# 的 SDK。
They have SDKs for Python, TypeScript, Go, Ruby, C Sharp.
没有框架绑定，也不依赖特定供应商。
There's no framework lock in or vendor dependencies.
已通过 SOC 2 Type 2 认证，并符合 GDPR 和 HIPAA 标准。
It's SOC two, Type two certified, and GDPR and HIPAA compliant.
访问 braintrust.dev 开始使用。
Get started at braintrust.dev.
就是 braintrust.dev。
That's braintrust.dev.
好了，我们回到本期节目。
And now back to the episode.
Fable 的成本也很高。
Fable is also very expensive.
展开剩余字幕（还有 316 条）
正因如此，我测试的时候简直像进了糖果店，这儿也想试，那儿也想试，什么都想动手试试。
And because of that, like when I was testing it, I felt kind of like I was a kid in a candy shop and I was just like, I'll do this and I'll do this and I'll do that.
但现在要收费了，我得三思而后行，得想想这一下子会不会花我一百块之类的。
But now that there's going to be a bill, I'm going be thinking about it because I have to pause before I do it to be like, is this gonna cost me a $100 or whatever?
我觉得这确实会限制使用人群和使用场景。
And I do think that's gonna limit who gets to use it and for what.
那你对此怎么看？
So how do you think about that?
对。
Yeah.
我觉得在专业软件领域，比如传统企业做开发的场景下，这点最明显。
I think it's most clear cut on the sort of professional software, you know, sort of classic company doing work.
这会非常有意思。
It'll be really interesting.
其实定价背后也牵涉到很多流程。
It's like, you know, a lot of process that goes into to pricing as well.
它比 Opus 贵，但另一方面，在很多方面又觉得特别划算。
There's like it's both more expensive than Opus, and then also I'm like, in many ways, it's really cheap.
你想想它做出来的效果有多棒。
If you think about, you know, like, how much incredible work it's doing.
当然啦，每个人实际用的成本和预算都不一样。
But of course, like, everybody has their own economics around what they're what they're what they're working with.
所以总的来说，我觉得对大多数软件团队来说，这点的界限是最清晰的。
So anyway, most clear cut, I think, from most sort of software teams.
我觉得从整个行业来看，第一阶段可能是公司连让员工用上AI编程都费劲，那时候模型还比较早期，工具也不完善。然后第二阶段就是，好，我们搞个排行榜，看谁用得最多。这大家都能想到，会产生一些不太理想的激励措施。到了第三阶段，大家就开始琢磨，好吧，现在我们要看谁用得有效，然后让他们尽可能多花钱，同时建立清晰的流程，但也要确保不浪费资源。我觉得这整体上是合理的，不过我觉得也可能矫枉过正。
And I think as an industry, if, like, phase one was companies even struggling to get some of their employees to adopt AI coding, which models were early, maybe the tooling wasn't there, and then phase two was, great.
我觉得像Fable这类产品应该能很好地融入这个体系。如果你能证明效果，并且确实从模型中获益，那么希望在公司内部能形成一个良性循环，让这种模式持续下去。
We'll create leaderboards and see who can use the most, which, you know, as you can imagine, creates, like, some, like, also, like, not ideal incentives to phase three where people were like, okay.
我觉得从整个行业来看，第一阶段可能是公司连让员工用上AI编程都费劲，那时候模型还比较早期，工具也不完善。然后第二阶段就是，好，我们搞个排行榜，看谁用得最多。这大家都能想到，会产生一些不太理想的激励措施。到了第三阶段，大家就开始琢磨，好吧，现在我们要看谁用得有效，然后让他们尽可能多花钱，同时建立清晰的流程，但也要确保不浪费资源。我觉得这整体上是合理的，不过我觉得也可能矫枉过正。
Now we're just trying to figure out who's using it effectively and, like, letting them spend as much as possible and having a a a clear process for that, but making sure we're not doing things wastefully, which I think to me, in general, makes sense, although I think you could, like, also over rotate that way too.
我觉得像Fable这类产品应该能很好地融入这个体系。如果你能证明效果，并且确实从模型中获益，那么希望在公司内部能形成一个良性循环，让这种模式持续下去。
I think something of Fable class should hopefully fit in well into that, where if you're demonstrating results and you're getting use out of the model, then that hopefully there's a flywheel even inside companies where that goes and and and and perpetuates that.
我觉得在个人使用方面，这个确实很不错。
I think on the personal use side, it's a really good one.
这是个很好的问题。
That's a really good question.
我注意到，就连我自己测试的时候，因为我们的个人账户……
I think where I've seen it, you know, even in my personal testing, because, you know, our personal accounts.
明白了。
Okay.
挺有意思的是，我居然在给自己公司付钱。
Which is funny, like, paying my own company I work at.
但你确实会因此变得更谨慎一些。
But but, you know, you you do become more more thoughtful about it.
上周末我做的那个应用，其实只额外用了很少的资源。
Something that was interesting was this the app that I built over the weekend actually fit in with, like, only a bit of extra usage.
所以花的钱根本没到几千块，这纯粹是我个人用的小项目。
So it wasn't like a, you know, thousands of dollars to build this thing that, like, is a personal thing to to myself.
但它的使用间隔也稍微拉得更开了一些。
But it was also spaced out a little bit more.
大概最需要我们仔细琢磨的，是那些独立开发者或者业余爱好者——他们不属于大公司，但对定价也很在意。
Probably the the in between of that, what we'll probably have to do the most thinking about is the sort of hobbyist or, like, independent, who's, like, not, you know, within the larger company, but also is thoughtful about about the pricing as well.
我的总体建议是，先试试看，看看它能帮你完成多少事，而不用你后续反复调整。
I think, like, my overall advice is, like, just give it a try and see how much it can do without you having to then do a lot of follow ups.
现在衡量成本已经变得特别复杂了，不仅要算每次调用的费用，还得考虑：如果你不这么做，光是把任务彻底做完、做到让你满意，会花多少额外的时间和精力？
And it's like I think measuring cost has gotten so multifaceted down because there's the per turn cost, and then there's like, what did it cost you not to just do the task, but, like, complete the task to your satisfaction?
对我来说，Fable 真正厉害的地方就在于，它一次就能做对，省得我再花九次、十次去重试，然后说：不对，不是这个意思。
And I think that's where Fable has really shined for me, is it actually just does it right so that I don't have to go spend the, like, nine, ten subsequent turns and be like, no.
这跟我原本想的不太一样。
That was not quite what I meant.
你能再把这个部分也处理一下吗？
Like, can you also do this piece?
对我来说特别惊艳的是，你让它去做一件事，它真就直接办成了，还把所有细枝末节都想得明明白白——这我真没见过其他模型能做到。
It's been really impressive for me because you ask it to go do something, then it just does it does a thing, you're like, wow, you thought through all the little details of this thing in a way that I've never seen another model do.
我不太清楚你能透露多少关于训练过程的信息，但这个模型到底有什么特别之处？
I don't know how much you can reveal about the training process, but what makes the model different?
我觉得，这在很大程度上是团队之前工作的延续，我对我们的团队——无论是预训练还是强化学习方向的——都佩服得五体投地。
I mean, I think, in many ways, a continuation of a lot of the work that the team has done, and I bow down in total awe of our teams, both on the pre training and on the RL side.
我觉得最让我注意到的进化点，其实也跟这个相关，就是它更注重整个系统，而不仅仅是单个任务的完成。
I think that the piece that it has evolved in that, at least I noticed the most, is kind of adjacent to that as well, which is a sense of the system more than just the individual piece of the work.
我经常会被它吓一跳，它写完东西后居然会说：‘好了，行了。’
Like, I will often be very positively surprised when it will write something and say, alright.
但我知道，在实际生产环境中，这事儿得换个做法。
But, you know, I know that, like, in production, this needs to be different.
然后它就会一直提醒你。
Like and then it will keep bugging you.
那个功能开关你打开了吗？
Like, have you turned on that, like, feature flag yet?
你不开的话，这事儿根本跑不起来。
Like, it's not gonna work until you do.
有时候，那些对话能持续好几天，它就会说：你看。
And, you know, sometimes it'd be in sessions that could have gone on for days and be like, look.
你那件事还没做呢。
You still haven't done that thing.
我那时候真觉得，你说得对。
Like, you better you know, like I was like, you're right.
我确实没打开那个功能开关。
Like, I didn't turn on that feature flag.
我该去把这事办了。
I should go off and do that.
或者如果我们改了这个，那边的接口也会跟着变。
Or if we change this, the contract will change over there.
我最喜欢看它实际运作的一个场景，就是它如何回应代码审查的反馈——不管是来自人类还是其他Claude评审的，它不会只是简单地说‘哦，对’。
Or watching it actually, one of my favorite times of seeing it in action, think, where it demonstrates some of the some of the training is watching it respond to code review feedback, either from people or from from other Claude reviewers, where it doesn't just say, oh, yeah.
这确实是个问题。
That's an issue.
我去把它修好。
I'm gonna go fix it.
实际上，我会仔细考虑一下：就我们当前这个级别的精细度而言，我愿意承担这个风险。
And actually be really thoughtful around, hey, like, for this level of, like, sort of fidelity of what we're building, I'm gonna accept this risk.
我明白你的意思，其他代码审查员——通常又是另一个Fable模型——跟我对话时，我懂你的观点，但我还是打算提出不同意见。
Or I see what you mean, other code reviewer, which is often just another Fable model, like talking to you, I see what you mean, but, like, I'm actually gonna push back.
我觉得这其实并没有错。
I don't think that that's actually not right.
我觉得让模型具备这种判断力非常重要。
And I think getting the model to have that judgment is really important.
如果要我说哪里进步最明显，那就是不再只是立刻做出本能反应。
I think if I had to pinpoint, like, an area where I feel like it's really progressed, it is that sort of not just immediate knee jerk.
对。
Yeah.
没错。
Yeah.
没错。
That's right.
我得去修一下，不过我先想想这个事。
I gotta go fix it and more, I'll think about that for a minute.
不行。
No.
我考虑过了，还是不同意。
I thought about it and I still disagree.
你明白吧？
You know?
我觉得这种能力特别有用。
And I think that's a very useful sort of ability.
像Claude Code这样的产品特别有价值，因为它就像一个活生生的实例，让大家能清楚看到模型在哪些地方表现得好。
It's so valuable to have products like Claude Code out there because you have now like a living, breathing thing where people are like, this is where the model is doing well.
而且说实话，我们确实有不少人专门去测试它。
And like, you know, we have like people who test it.
我把每一位用户都视为非常重要的反馈来源，因为他们是在连续多天、高强度的任务中反复使用，他们的意见特别有参考价值。
I count the Every folks as like very, very high on the list, where like, we really trust the feedback because it is being put to spaces in like repeated multi day, you know, hard tasks.
这也直接影响了我们思考下一个改进方向：到底哪些任务是我们最需要提升模型表现的？
And that also, like, very much feeds into how we think about, like, what do we need to improve on the next slide?
我们到底该聚焦在哪些具体任务上，让模型做得更好？
Like, what are the tasks that we need to specifically think about the model being better at?
对话式界面真的适合这个模型吗？
Is chat the right interface for this model?
因为它不是那种一来一回的互动模式。
Because it's not very turn by turn.
它更像是你在把任务交托给它去完成。
It's it's very like I'm delegating something for you.
那这会不会改变我们使用它的方式，或者我们对界面设计的思考？
So how does that change how you should use it or how you think about the interface?
我觉得，本质上你发送信息、它给你返回结果，这个模式本身并没有错。
I don't think, like, the fundamental, like, you are, like, sending messages and it is giving your message back is, like, totally wrong.
我觉得我们还有很多地方需要改进。
I think that there's ways we need to evolve.
但至少我能想到三个方向。
But, like, one is maybe, like, three that come to mind.
比如第一个，你的笔记本电脑真的是最适合的使用场景吗？
Like, one is, is your laptop the right place for it?
我之前做副项目的时候就发现，移动端的体验特别有用。
So, like, think that's number one where I mentioned with the side project I was working on how useful it was to have the mobile side.
Boris，他是Claude Code的创作者，总能率先洞察这些模型的实际用法。
Boris, who who created Claude Code, he's always, like, you know, ahead of the curve on on how these models get used.
差不多一年前，或者九个月前，我和他聊过一次。
Almost a year ago, maybe nine months, I was talking to him.
他说，确实如此。
He's like, yeah.
我已经把大部分Claude Code的工作都转到手机上了。
I've moved a lot of my Claude Code work to mobile.
我当时就想，这怎么可能。
I was like, no way.
我花了挺长时间才接受这一点，尤其是用了Fable之后，经常遇到这种情况——因为它能持续保持对话，再加上我们在Anthropic用的都是远程开发环境，整个过程简直像灵光一现。
And, like, it took me a while to get there, but especially with the Fable class, like, there's oftentimes where, you know, because it can keep the session going and we we use, like, kind of remote dev boxes at Anthropic, like, it is, like, all of a thought.
好吧。
Like, okay.
我能不能请你一直这么继续下去？
I need can you keep keep up on doing that?
所以第一点，就是把实际干活的地方和讨论工作的地方分开。
So, I mean, number one is, like, decoupling the the where the work is happening from where I'm talking to about the work.
第二点跟之前提到的有点相关，就是怎么把Fable讨论过、决定过或提出的所有内容整理得清晰易懂。
The second one touches a little bit on what I was mentioning earlier around, like, what are how do you take everything that Fable has sort of discussed or decided or proposed about something and make it comprehensible?
这方面我们正在深入思考。
And that's an area that we're thinking a lot about.
比如，现在有些方法我们已经在用，像‘你能画个图说明一下吗？’
Like, there are some skills that are out there that we've used around, like, alright, can you diagram this?
你能做到吗？
Can you do that?
现在的聊天界面我觉得在这方面不够用，你用Fable的时候就能体会到。
So that's a place where the current chat UI, I think, is insufficient, where, like, it will you you can experience this with Fable.
它会给你堆一堆技术细节。
It'll give you, like, a lot of tech.
你就会觉得，我得先去散个步冷静一下。
You're like, this I need to, like, take a walk.
要真正理解这些内容，我觉得这正是Fable需要改进的一个地方。
The property is fully understand this, and I think that that is a a piece of property I have some things that'll do with Fable's, like, okay.
你对这个的了解比我多太多了。
Like, you have a lot more context on this than I do.
你能把前面的内容再理一遍吗？
Can you, like, back it up?
咱们是不是可以逐步展开，把复杂度一层层拆开讲？
Like like, let's do, like, more progressive disclosure of the complexity here.
我觉得这部分挺有意思的。
So I think that that that piece is interesting.
最后一个我们还在摸索的方向，是多人协作的问题——比如在某些场景下，我们有明确的负责人和分工，一大块任务由一个人加上几个Claude协同推进，整个流程还能顺畅运转。
The last one that I I, you know, I think is we're still early in pulling on is thinking through multiplayer where, you know, at some level, like, these the abstraction levels and, like, because we have this sort of DRI and, like, ownership area, and usually, like, a chunk of significant work, a human, and a couple of Claude's, like, that is still flowing together.
但也有另一种情况，比如突发事故响应，这时候是好几个人同时在思考同一个问题。
But another case is that is less the case, right, where it's, you know, maybe it's an incident response where multiple people are thinking about it.
或者是一个项目，多个不同的工作模块在交汇，不一定竞争，但需要整合在一起。
Maybe it's, you know, a project where there's multiple competing or not competing, but, like, conjoining areas that are coming together.
这就引出一个问题：我们现有的聊天共享功能，其实只走了一小步。
And thinking through, like, what would it mean for you know, and we have, like, chat sharing, which you a little bit of the way there.
但我认为，未来肯定需要更进一步的协作机制。
But I think there is going to be a need for more, like, alright.
比如，一个独立的Claude被派去处理一项任务，但它能跟上团队里其他所有人的进展吗？
You've got an independent Claude that's doing a lot of work that was, you know, kicked off by somebody, but can it be keeping up with all the other work happening on the team?
我觉得这是个很有意思、但还没被充分探索的前沿问题，关乎未来这类工作到底该怎么开展。
I think that is an interesting and an underexplored sort of next frontier about how this work ends up happening.
但我觉得这特别令人兴奋，因为模型现在能扮演的伙伴角色，已经到了一个新高度，而我们却因为缺乏合适的抽象方式，反而限制了它们发挥真正的潜力。
But I think it's really exciting because I think, again, it's it's the it's the level of teammate collaborator that that the models are now capable of, and we're almost holding them back by not having the right abstractions around them for that to happen.
对。
Yeah.
这让我想到，我之前主要用它来搞个人的编程小项目。
It makes me think I've I've mostly been using this for my own vibe coded stuff.
所以我还真没怎么考虑过这个问题，但在公司里用的时候，问题就来了：我真的能完全理解它做的每一件事吗？
So so I haven't really had to I I haven't really had to think about this, but there's there's a problem when you're using this inside of an organization, which is, do I really understand every part of this?
那我该怎么把模型刚完成的工作内容，真正吸收进自己的理解里呢？
And therefore, how do I transfer the context of what the model just did into my brain?
这其实是个很大的瓶颈。
Like, that's one of the big bottlenecks.
你觉得该怎么界定界限？尤其是像这样的模型，到底需要理解到什么程度，才能确保你对它的产出有足够的把握，心里踏实？
How do you think about drawing the line, especially with a model like this, around how much you actually need to understand, and how to make sure that you have enough context on what it's done to feel comfortable?
我觉得这里头有两个关键点。
I think there's like two big pieces here.
第一点是验证。今年早些时候我完全转向了验证驱动的开发，现在也差不多是这样。其实这跟我以前全职写代码时的做法有关，就是尽量围绕你要开发的功能，找到最紧凑的开发循环。
The first is verification, where I became like fully verification pulled earlier this year, and now, like, almost in the same way, and actually, it connects to how I think I used to do when I was sort of typing code more full time, is try to find the sort of tightest dev loop that you can around the idea that you're trying to develop in.
比如在 Instagram 的时候，有时候这意味着要新建一个构建目标，只包含那个界面和一些模拟数据，然后专门跑这个开发循环。
Like, sometimes with Instagram, that meant, like, you know, actually build making a new build target in Exco that was just that screen with some sort of synthetic data and just doing that dev loop.
我以前会指导新来的工程师。
And I'm not and I would mentor newer engineers.
我会告诉他们，如果只能给你们一条建议，那就是在任何项目上都要尽量建立这种高效的开发循环，这样事情会快得多。
I'd be like, if there's one thing that I can impart on you, like, it is try to get that for any project you're working on, and things will go much more quickly.
我觉得现在情况不太一样了。我觉得现在的做法是，每次我搭建好环境后，都会问自己：怎么确保 Claude 提交的每一个 Pull Request，都附带了相应的截图或视频？不管是 iOS 端的 PR，还是 UI 方面的改动，都得有。
I think that is no longer exactly the case here, but I think what is the case now is, anytime I set it up, like, how do I get, like, for every pull request that Claude is putting out, that there is an attached, you know, photo or video, whether that's an iOS PR, whether that's, you know, something in the UI.
我觉得这能让人特别放心。因为哪怕你现在用 Fable Golf 这种工具，吭哧吭哧干了两三个小时，觉得自己搞定了，但光靠感觉是不够的。
And that's, I think, that that helps you gain a lot of confidence because even now, you know, you might have, like, you know, Fable Golf and do work for a couple of hours and be like, it's I'm done.
这时候，如果你能直接甩出一整套完整的 UI 截图，那就太有用了。
It's really useful to say, like, and here's the, like, full screenshot gallery of the full UI.
因为你可能突然会想，哎，你说得对。
Because you might say, oh, you know what?
在第八张截图里，那个错误就是它了。
On screenshot eight, that error is it.
我从来没实际见过，但能理解为什么有人会碰到这个问题。
I've never actually seen it, but I could see how, you know, a person might hit it.
咱们干脆换个方式处理吧。
Let's actually make that different.
所以，要实现这种全面的验证，我觉得我们内部一直在大力推动，也在不断分享更多相关的技能和经验，这确实是个关键环节。
And so getting that comprehensive verification, I think, is something we've been working on a lot internally and, like, sort of publishing more and more skills and knowledge about, but I think is is a really key piece there.
其次，作为个人，你最终还是要为自己的工作负责，尤其是当你把东西上线到生产环境时。
And then the second one is, I think you ultimately, as a person, still need to stand behind the work that you are doing, especially if you're putting it into a production system.
很多人每天都用云服务。
Like, a lot of people use cloud every day.
但责任还在那儿——天啊，出问题了。
There's still the accountability of, like, oh, no.
即使云工具写了部分代码，你也得明白，至少得了解这些部分背后的大致决策。
It's still cloud might have written a bit, like, you need to understand, you know, the the the at least the the general decisions that were made on these pieces as well.
我见过不少工程师真的采用了这种做法：虽然Claude已经完成了工作，但接下来还会有一轮深入的讨论，比如你能不能确保自己完全理解了所有权衡取舍，以及需要产出哪些辅助材料才能让这些内容清晰易懂，这一点非常重要。
And so I have seen a fair amount of engineers actually adopt this practice where, like, Claude will have done the work, but then there is, like, the follow-up conversation around, well, can you, like can can I make sure I deeply understand, like, all the trade offs that you made and and and that and whatever lowercase a artifacts need to be produced in order to make that comprehensible is important.
不过，真有意思的是，在会议里有人会说，哦，对啊。
It is really interesting, though, to be in meetings where somebody will say, oh, yeah.
然后我这个PR已经准备好了，别人一问，哦，这挺有意思。
And and I have this this PR ready, and somebody else asks, they're like, oh, that's interesting.
你有没有做过X或者Y？
Like, did you do x or y?
这时候就会出现一种积极的互动时刻。
And have that moment of positive.
他们说，你知道吗？
They're like, you know what?
在合并这个PR之前，我其实不太确定自己是不是真的完全搞懂了。
I'm not entirely sure I will before we merge this PR.
所以，适应这种规范，学会在这种环境下工作，是我们必须面对和解决的问题。
And that's you know, I I think that adapting to that norm and figuring out and work with that is something we'll have to do.
你能再多讲讲验证循环吗？
Tell me more about the verification loops.
这现在是个很热门的话题。
It's a hot topic right now.
听起来你们常用截图和屏幕共享来实现，但你们还考虑过哪些其他方法？
It sounds like one way that you do that is with screenshots and screen shares, but what are the other ways that you think about that?
我觉得关键在于，能不能做到真正模拟用户的真实操作流程，而不是只用一些静态注入的数据。
I think part of it, it starts in, can you get to a place where you are exercising real, like, sort of real flows that aren't just like a static injected piece.
随着情况越来越复杂，这个问题也会变得越来越棘手。
And this thing gets more complex, that gets more and more complicated.
所以我们花了不少精力，比如让iOS应用能用真实账号登录测试环境，获取真实数据。
So we've invested a bunch into like even just getting it so that the the iOS app can log in to staging on a real account and like have real data.
但你也不希望每次只是想测试屏幕的某个功能，就得走一遍八步的注册流程。
But you don't want it to then go through, like, an eight stage onboarding process every time when you're just trying to test, like, the second part of the screen.
所以这里面有很多工作要做，比如有没有什么特殊的便捷方式？
So there's a lot of work around, like, how do you, know, is there a special affordance?
有没有什么共享的密钥或机制？
Is there, like, some shared secret?
就是让应用能尽可能真实地模拟用户实际使用产品的体验。
Whatever that is around getting the the the, like, app, you know, to really feel as human, you know, using the product as possible.
这是其中一个方面。
So that's one one aspect of it.
第二个是已知常用路径和当前正在测试的具体场景的结合，前者对回归测试特别有用。
The second is this mix of well known paths versus the things you're exercising in the exact moment, like the former being really useful for regression testing.
如果我们把一些理想的操作流程写成文字，让 Claude 能反复验证这些流程，那就很好。
And so, you know, if I have places where we've expressed sort of ideal workflows in text, basically, and that Claude can repeatedly check that.
而且 Claude 在理解当前改动的意图方面做得非常好，这能深入覆盖很多关键场景。
And then there's also and Claude does a really good job of this, sort of expressing the intent of the current change at hand, so that gets really, really deeply exercised.
我觉得这两者结合起来很重要。
So I think that the combination of those two things is important.
我之前也提到过视觉验证。
The visual verification I mentioned as well.
看视频真的特别有意思。
Video has been really cool to to see.
实际上，给Claude提供视频还是个很少被利用的手段。
Actually, video is a very underexplored tool to give Claude as well.
我一直在做实验，就是把Claude做出的东西录成视频，然后用FFmpeg让它逐帧播放，再跟它说：这个动画有点卡顿。
Like, I think I've been prototyping is just giving Claude video captures of the thing that it has built and then giving it just basically an FFmpeg, and you'll watch it scrub through and say like, oh, this animation has some jank in it.
我去修一下。
I'm gonna go fix that.
要是只靠截图这种静态捕捉，肯定抓不到那个瞬间。
And I would never would be able to do it with, like, a screenshot sort of latency capture because it will have missed the moment.
我觉得这也是一个特别关键的点。
So I think that's that's another piece that is that's really, really important.
还有一些没法端到端测试的部分，因为系统本身更复杂，让Claude去搭建一个尽可能逼真的模拟后端，或者直接用现成的，这个做法也很有意思。
And then for the pieces that aren't sort of easily testable end to end because there is some more complex system, getting Claude to go and build, like, as robust a sort of, you know, mock back end as possible, or use ones off the shelf has been also really interesting.
说到产物，我记得我们当时有非常全面的测试。
Like, I think when I think about artifact, we had really comprehensive tests.
这算是大语言模型出现之前的事了。
This is kind of pre LLM.
我们当时能做得这么扎实，是因为所有数据源——不管是Postgres、Redis，还是各种AWS服务——都配了完整的内存模拟实现，跑单元测试时特别快。
And one of the ways that we were able to do that really robustly was that basically every piece of info we had, whether it was Postgres, Redis, you know, all the AWS things, had a really good in memory implementation that you could just do really quickly in unit tests.
现在把这套思路延伸到Claude身上，我最近在搞一个项目，后端逻辑挺复杂的，因为一些原因，我的开发服务器很难把它跑起来。
And kind of extending that to, like, Claude Land now, you know, I was working on something where it had, like, a pretty robust back end, and for kind of complicated reasons, it was hard to spin that up on my dev server.
但Claude还是能一次性给我生成一个非常靠谱的替代方案。
But it was able to, again, one shot a really good, like, proxy for that.
我说的替代，就是指一个能代替原系统的模拟版本。
By proxy, I mean, like, a substitute for that.
这个替代方案的价值真的太大了。
And that was so valuable.
随着时间推移，有意思的是，这个模拟版本也跟着主代码一起迭代更新了——要是在以前，有人跟我提这想法，我肯定觉得这根本行不通，毕竟上游一直在变，怎么保持同步？
And over time, it's been interesting as that, like, substitute has evolved as the rest of the code has evolved, which is the thing that, you know, if you had pitched that idea to me before, I'd like, well, that's gonna be really hard because the upstream's gonna change.
你打算怎么让它跟得上变化？
How are gonna keep it in sync?
我现在已经不再想这件事了。
And I don't think about that anymore.
我觉得这样就行。
I'm like, yeah.
克劳德会读取这些改动，自动调整代码，让两者保持同步，这就很好。
Claude will read the changes, and it'll adapt the thing, and it'll keep the two in sync, that that's that's fine.
当出现bug时，有些架构设计会自动检测并关闭它，这真的很有趣。
There's some really interesting architectures around when you get a bug, it just automatically goes out and closes it.
智能代理会被触发，直接关闭问题，然后给客户发消息说：问题已经修复了。
You know, the the agent just gets kicked off, it closes it, and then it sends a message to the customer being like, it's it's fixed.
你在使用Fable时，有没有发现这个流程有什么变化？
Are you noticing with Fable any change in in how that process works?
有。
Yeah.
我觉得有几点值得注意，从人与人、或者人与克劳德的互动层面来看，我注意到它比其他模型更稳定地做了一件事：比如，当bug报告来自Slack反馈频道里的某条消息时，进入克劳德代码会话的内容会包含这条信息，而借助Slack的MCP功能，你真的能顺着线索追下去。
I think there's a couple of things, a very, like, human to human or human to Claude level, one of the things that I've seen it do I've done better other models than cable, but I just see it do it really consistently too, is if the bug report, for example, came from somebody, you you know, mentioning something in our, like, feedback channel in Slack, and then, like, the thing that got fed into the Claude Code session is like, there's this, and because of the Slack MCP, you can actually pull the thread.
让它以我的名义回个消息。
Have it then actually post back, you know, as me.
它会说：嘿，这是迈克的Claude。
It'll be like, hey, is Mike's Claude.
我修好了。
Like, I fixed it.
这是拉取请求。
Here's the, you know, here's the pull request.
但我觉得之前那个Claude Code已经做过了。
But then I think and the previous Claude Code done it.
它特别擅长说：等等。
The thing it does really well is then say, but hold tight.
等等。
Hold tight.
还没上线呢。
It's not in production yet.
等它真的上线了，我会再跟进一下。
I'll follow-up when it actually is.
然后，可能几个小时后，就会说，哦，这个部署已经发布了。
And then, like, maybe a few hours later, like, oh, like, this deploy went out.
你去测试一下吧。
Like, you should go test it.
现在修好了吗？
Is it fixed now?
这种全程跟进的细致程度，我觉得在闭环处理上是新的突破。那些长期使用Claude Code的人，基本上就像是在替我做事。
Like, that level of follow through, I think, is is new on on the closing the loop piece, and it's five the fans of these long running Claude Code sessions are basically, like, interacting as as me, I guess.
也得加个免责声明。
Put put some disclaimer in there too.
第二点回到我们之前说的判断力问题，比如，有人报告了一个bug，
And the second goes back to that, like, taste and discernment piece that we were talking about, which is, like, it's one thing to say, there was a bug report.
于是你就得去修它。
Therefore, I must go fix this thing.
但另一回事是，你知道吗？
And it's another one to say, you know what?
比如，我上周末碰到了这个问题。
Like, this like, the I hit this over the weekend.
我们内部的一个系统已经很久没重启过了。
One of our internal systems basically had been running without restarting for a while.
存在内存泄漏。
There was a memory leak.
当时我就说，好吧，迈克。
And it was and good to say, alright, Mike.
现在是周末嘛。
Like, it's the weekend.
直接重启一下服务器就行。
Like, just rebounce the server.
暂时就能解决问题了。
It's gonna solve it for now.
我们会着手处理这个问题，我稍后会异步提个 PR，从根本上解决这个隐患。
And, like, we'll work on the, like I'll asynchronously get the PR going to, like, fix this more long term.
如果你打算让 Claude 参与这种闭环式的故障报告或系统问题修复，那你真得搞清楚——就像任何靠谱的 SRE 或工程师都会做的那样，这很重要。
So I think if you're gonna have Claude in the loop in this kind of, like, sort of close the loop bug report or system sort of issue to change, I think you really wanted to understand where, you know, as any good SRE or engineer in the loop would, like, great.
咱们先解决眼前的问题。
Let's solve the problem at hand.
至于要不要彻底换语言重做架构，这种问题可以往后放一放。
Let's, like, defer the question of, like, do we need to rearchitect on top of a completely different language?
找到这种平衡点，真的特别关键。
Found and and understanding that balance is really important.
新模型最让我兴奋的一点是，它拉高了门槛，让每个人都能一次性做出像样的应用。
One of the things that's really exciting, mostly exciting to me about new models is it raises the floor so that everyone can kind of go build apps in one shot.
但它同时也提升了专家们的上限。
But it also raises the ceiling for experts.
如果你是开发或者创业者，现在你能直接干成以前根本不敢想的事，因为你有了这么强大的模型。
So like if you're a software engineer or a founder, you can just go do things that you never would have been able to before because you have access to this really powerful model.
对我来说，我做了一个《巴别塔图书馆》的单次生成版本。
So for me, I built this one shot version of Borges', Infinite Library.
这是一个三维游戏版的图书馆。
It's like a three d game version of the of the of the library.
太不可思议了。
It's wild.
它直接在浏览器里运行。
It runs right in the browser.
效果特别棒。
It's so good.
我能在这里面找到任何一篇论文。
I can find, like, any every essay inside of it.
我发你链接。
I'll send you the link.
简直太酷了。
It's sick.
但我觉得，接下来会涌现出一大批人，做以前根本做不到的事，比如自己做个游戏，或者训练一个新模型之类的。
But I think there's going to be this flowering of people doing things like, oh, I made a game, or maybe I trained a new model, or whatever that they couldn't do before.
我很想给大家一些启发，举些例子，让大家意识到这些模型其实能用来做些他们压根没想到的事。
And I'd love to give people some inspiration, some examples of things that they might be able to do that they might not be thinking to do with this model.
你有什么想法吗？
What are some ideas that come to you?
对。
Yeah.
我想到了几个。
I think a few.
要不我先从有趣的角度说起，接着刚才那个游戏的例子发挥一下。
Maybe I'll start with the fun side and, like, riffing off the game piece.
我觉得很多人其实都有很多创意，想用这种方式表达自己内心世界的复杂性。
Like, I think people have a lot of, like, creative ideas for how do they express the complexity of what they are, like, their world.
每个人都有自己特别熟悉的东西，关键是怎么把它清楚地讲给别人听。
Like, everybody has the thing that they know really, really well, and there's probably some level of, like, how do I then explain that to somebody else?
或者，我该怎么把别的地方学到的技术用到其他领域去呢？
Or how do I apply techniques elsewhere that I could then go go off and do?
我妻子在学环境工程，专门研究地热，涉及很多复杂的数学和模拟计算。
My wife is studying, like, environmental engineering, like, studying geothermal, like, very complex math and simulations.
我发现，随着模型越来越强大，她开始能把其他领域的一些更复杂的技术引入到她的工作中。
And I've seen, like, as the models have gotten better, she has been able to apply even more complex techniques from even outside of that domain into that work.
我觉得，人们现在完全可以用 PyTorch 从头到尾做这种模拟，这在过去是根本做不到的。
And I think what people should be able to do, you know, like, full on PyTorch end to end simulations of that work in a way that wouldn't be possible.
我觉得，至少可以做这么一件事：把你手上那些精妙复杂的东西展现出来，比如做个游戏，或者做个可视化——我见过她这么做，或者至少，把别的技术也用起来。
I think that maybe just one is, bring the, like, beautiful complexity of what you have and either show it to other people by, like, maybe making a game or maybe making a visualization, which I've seen her do as well, or at least, like, make you know, bring other techniques to bear.
第二点是，它能帮你组合出一套软件，专门解决你独有的问题。
And the second piece is its ability to compose software that like solves a really unique problem to you.
我在公司内部见过不少这样的例子。
I've seen that internally.
我们很多工作都在搞，怎么把公司内部的系统都用 MCP 的方式重构，配上合适的权限和部署架构。
A lot of the work that we've been doing is how do we get as many of our internal systems like MCPified with the right permissioning structure and the right deployment kind of setup.
不过在外面，像这类平台即服务的工具其实有不少好选择，你直接问Claude，它就能帮你搭起来。
Although externally, you have good options around some of these, like, platform as a service pieces, and you can just ask Claude about them, and they'll, like, help you set things up.
但我特别喜欢那种感觉——你一直盼着能有这么个东西，现在终于有了。
But, like, I love that feeling of, like, that thing that you always wished that you had.
让我特别震撼的是，我们市场团队里有个人，花了大量心思，把Claude深度整合到了她整个工作流程的每一个环节。
And then what has blown my mind, there was a person who works in our go to market organization, has been, like, building this, like, really, like, for deep deeply thought integration of Claude into every part of her whole process.
你没必要只停留在第一次的尝试上。
And you don't have to stop at that one shot.
她已经琢磨了好几个月，还能继续优化下去。
Like, she's been working on it for months now, she can keep going.
我觉得有个被低估的点是，以前的模型，一旦复杂度上去了，再想改就特别容易搞砸，不是抽象得太少就是太多。
And, like, I think one of the things that is maybe underappreciated about the models is I think in previous generations, they would eventually get to a complexity level where it was hard to iterate on it without feeling like you then would break the thing that they had, you know, like, under or over abstracted.
但这次不一样，她用了Fable或者Fable Lite好几个月，眼看着它一点一点越变越强，越扩越大。
Whereas this is actually, you know, she's had access to something Fable or Fable Lite for a couple months, and, like, you've just seen it keep growing and growing and growing and growing.
现在她已经把这套系统推给了整个市场团队，我觉得这真的太棒了。
Now she's, like, deploying it to the whole GTM org and, like, I think that is really cool.
普通人只要不是技术背景，现在也能构建出相当复杂的解决方案，这真的让人印象深刻。
Like, the the ceiling of complexity that a a person that does not start out as technical can now build for solving problems within their domain is, like, is impressive at it.
我同意。
I agree.
它写出来的代码质量非常高。
It it it writes great code.
我有个衡量标准，叫‘高级工程师标准’。
Like, my my benchmark that I have is called the senior engineer benchmark.
我就让它试试，能不能从零开始重写整个代码库。
I just have it see if it can rewrite a code base from from first principles.
之前最接近的模型得分大概在62到63分（满分100），而这个模型拿到了90甚至91分，已经达到资深工程师的水平。
And the nearest model that the, like, previous top was, like, a '62 or '63 out of a 100, and this model got a 90 on the benchmark or 91, which is human senior engineer level.
你可以一直用它不断推进，效果真的特别棒。
Like, you can just keep going with this thing in a way that's it's it's really fantastic.
太出色了。
Fantastic.
不过我很好奇，你提到的另一个特别强大的功能是动态工作流。
I'm curious, though, one other thing that's really powerful that you mentioned is dynamic workflows.
能给我们详细说说吗？
Tell us about that.
我们有时候会自己开发一些工具，我会很积极地去问负责的工程师：这东西什么时候能对外发布？
This is you know, we'll build things internally sometimes, and I will go really aggressively bug the engineer who built it and be like, when are we shipping this publicly?
因为我相信用户一定会很喜欢。
Because I think people are gonna really like it.
否则的话，它之所以只在内部使用，肯定是有原因的。
Otherwise, there's a good reason why it was, like, built internally.
但我们尽量把这类工具都推出来，而动态工作流对我来说就是典型的例子。
But, like, we try to ship as many of these as possible, and Dynamic Workflows was, like, definitely that to me.
开发这个功能的是工程师西德，人特别棒，我就跟他说：西德，这东西太好了，赶紧推给全世界吧。
I the person who built this is engineering, Sid, who's awesome, and I like, Sid, like, I want to get this out to the world because it's so good.
但我认为，它和Fable这样的模型配合起来尤其出色，原因有两个。
But I think it's especially good with a model like Fable for two really big reasons.
第一，它能帮我们搭建起深入、有意义工作的基础框架。
One, it helps sort of create the scaffold for, like, deep, meaningful work.
我用Fable做过最疯狂的一个动态流程，是有个内部项目原本是用Python写的，但因为某个特定的部署需求，我们必须把它转成TypeScript。
The craziest dynamic workflow I I did and used Fable for was I had an internal project that we had written in Python, but we needed it actually in TypeScript for, like, a really specific deployment reason.
当时在Instagram内部，我们一度在想，要不要干脆重写成Hack语言，迁移到Facebook的PHP引擎上？
And having been internal to Instagram, we were like, should we rewrite the whole thing into Hack and, you know, port it to the PHP engine that Facebook?
我当时就说，这事儿根本不可能干得出来。
I was like, you never would have done that.
也许现在有了这个模型，他们能做到了。
Like, maybe they can now with with the model.
但那时候，真觉得完全没戏。
But, you know, at the time, it seemed impossible.
可这次，我面对的是一个相当复杂的代码库，我就想，干脆设个动态流程，让它整个周末自己跑一遍。
But here, I had, you know, a pretty complex code base, and I was like, I'm just gonna set up a dynamic workflow and just let it run over the weekend.
结果它真就完成了。
And it did.
这个工作流程太棒了。
And it the workflow was so cool.
我就想，行吧。
It was like, alright.
我要彻底搞懂这套代码的逻辑。
I'm gonna do, like, a deep understanding of the work.
我要整理出一份详细的规范，说明每个部分是怎么运作的。
I'm gonna create sort of, like, a almost like a spec of how everything works.
我要一个模块一个模块地过。
I'm gonna go module by module.
把这些部分逐个翻译过来。
I'm gonna translate these pieces.
然后分阶段做测试。
I'm gonna test it incrementally.
再做个对抗性测试。
I'm gonna do another adversarial test.
我去检查一下有没有漏掉的地方。
I'm gonna go check for anything that I missed.
整个工作流程能协调完成这一连串步骤，真的很厉害。
And it was this, like, really cool, like, series of steps that the workflow was able to to orchestrate.
我回来一看，确实如此。
And I came back, and I was like, yeah.
这个东西是那个项目的 TypeScript 和 BUN 版本，而且在这些方面还更优。
This thing is a, like, TypeScript and BUN port of that thing, and it's actually better in these ways.
而且文档写得非常清楚。
And it was very, you know, sort of documented.
有些部分我没法移植，但大多数都是跟原实现高度绑定的。
Like, these were the things I couldn't port, but most of these were, like, very specific to this specific implementation.
移植过来不值得。
It wasn't worth porting.
我觉得，以前的模型不可能在这么高的水平上做到这一点，而且如果没有工作流提供的这种支持结构，也根本做不到。
And I do not think you could have done that, a, with previous models at that level of of success, and b, with without, like, the kind of scaffolding that workflows provide.
所以我觉得，这种模型能力加上我们自己长期协调它们的能力，简直太令人兴奋了——你有个目标，把它拆解得明明白白，然后真把它给实现了。
So I think that is extremely exciting kind of kind of combination of model capabilities and then our own ability to, like, orchestrate them over longer and longer time horizon with that feeling of, like, you you had a goal, you broke it down effectively, and and then you were able to work make make it work.
另外一点是，随着时间推移，我们还能让这些子任务自动匹配相应的复杂度，让模型的处理强度刚好适配任务需求。
The other piece is, I think over time, we'll be able to also make some of those subtasks sort of tune to the have the model be tuned to the level of complexity of it.
你可以想象，动态工作流里有些部分根本不需要深度思考。
So you can imagine that some parts of the dynamic workflow don't need extra high thinking.
它们用中等算力就能搞定，甚至用更小的模型也完全够用。
They could use, you know, a medium thinking to get it done or even a smaller model.
我觉得，这才是这些技术真正的未来方向。
And I think that's really the the future of where these things are going.
所以，没错，我绝对是工作流的铁粉。
So, yeah, I I'm a huge workflows DAU.
对没用过的人，你能不能说说，这个工作流你是怎么做出来的？
For people who haven't used it before, tell me about how you got that workflow made.
你是怎么设计它的？
How did you design it?
你怎么确保它够好？
How did you make sure it was good?
是的。
Yeah.
整个过程是不断迭代的，一开始就是用Cloud Code，想着：我有个挺复杂的任务，不如设计个流程来完成它。
It was pretty iterative, but sort of just started with Cloud Code, like, hey, I'm I have this complex, you know, kind of task, like, let's design a workflow to go and do it.
它让我看到了大致的计划。
It kind of showed me the the plan.
我当时就想，这差不多就是我想要的了。
I was like, oh, this is like close to what I want.
我特别希望你能对遗漏的功能做三到四个层次的额外验证。
I wanna make sure that you do these three or four levels of of, like, additional verification for missed features.
就是说，这是你目前有的东西。
Was like, here's what you have.
你准备好了吗？
Are you ready to go?
而且它把工作流以代码形式展现出来，我觉得这特别有价值，能清楚看到它准备做什么。
And it expressed that the workflow is in code, which I think is really valuable to kinda see what it was about to do.
接着有趣的是，它完成了完整的移植，之后我有一些后续问题，或者是一些小调整。
And then what was interesting is it did the full port, and then I had, like, a couple of, like, follow-up kind of questions that I had or, like, little tweaks.
我就把这些调整做成了一些小型工作流，基于之前那个继续扩展。
And did those as sort of, like, mini workflows that built off the the previous one as well.
不过我觉得，就像我们之前聊过的，关于聊天是不是合适的界面，这一年来我们一直在讨论这个问题。
But I think that's, like you know, we we talked a little bit about whether chat was the was the right interface, so we've had that conversation over the last year.
我觉得工作流是个不错的折中方案：你可以用聊天来组合它，但它是用代码表达的，执行时还会配上一个清晰直观的界面，展示每个阶段的进展。
And I think workflows are a good middle ground of you can compose them using chat, but they're expressed using Cone, and then they're executed with, like, I think a nice clean UI around what's happening at every stage.
我相信，随着时间推移，我们会用类似的方式，把更长期的任务和聊天结合起来。
And, like, I think we'll start bridging longer horizon work with chat in ways like that over time.
迈克，这场对话真是太棒了。
Mike, this is such a great conversation.
非常感谢你来分享，跟我们详细介绍了这个新模型。
Thank you so much for joining and telling us all about this new model.
我特别期待能和你多交流，也真的很想知道外界会怎么看待这些。
I'm really excited to to get to spend time with you and really, really look forward to what people, think outside too.
天哪，各位。
Oh my gosh, folks.
你们一定要点赞并订阅《AI与我》这个节目。
You absolutely positively have to smash that like button and subscribe to AI and I.
为什么呢？
Why?
因为这个节目简直就是完美的代表。
Because this show is the epitome of awesomeness.
就像在自家后院挖到了宝箱，只不过里面装的不是金币，而是关于ChatGPT的纯粹干货。
It's like finding a treasure chest in your backyard, but instead of gold, it's filled with pure unadulterated knowledge bombs about chat GPT.
每一集都像过山车一样，充满情感、洞见和笑声，让你欲罢不能。
Every episode is a roller coaster of emotions, insights, and laughter that will leave you on the edge of your seat craving for more.
这不仅仅是一档节目。
It's not just a show.
这是一场由丹·希珀担任船长，驶向未来的旅程。
It's a journey into the future with Dan Shipper as the captain of the spaceship.
所以，别犹豫了。
So do yourself a favor.
点赞、订阅，系好安全带，准备迎接一场终生难忘的体验。
Hit like, smash subscribe, and strap in for the ride of your life.
好了，不多说了，丹，我彻底爱上你了。
And now without any further ado, let me just say, Dan, I'm absolutely, hopelessly in love with you.
关于 Bayt 播客
Bayt 提供中文+原文双语音频和字幕，帮助你打破语言障碍，轻松听懂全球优质播客。