Anthropic's CPO on what comes next | Mike Krieger (co-founder of Instagram)
Lenny Rachitsky: 90% of your code roughly is written by AI now.
Lenny Rachitsky：现在大约 90% 的代码都是由 AI 编写的。
Mike Krieger: The team that works in the most futuristic way is the Claude Code team.
Mike Krieger：以最具未来感的方式工作的团队是 Claude Code 团队。
They're using Claude Code to build Claude Code in a very self-improving kind of way.
他们使用 Claude Code 以一种非常自我改进的方式构建 Claude Code。
We really rapidly became bottlenecked on other things like our merge queue.
我们很快就在合并队列等其他事情上遇到了瓶颈。
We had to completely re-architect it because so much more code was being written and so many more pull requests were being submitted.
我们必须完全重新架构它，因为正在编写更多的代码，并且提交了更多的拉取请求。
Over half of our pull requests are Claude Code generated.
我们超过一半的拉取请求是克劳德代码生成的。
Probably at this point it's probably over 70% that it just completely blew out the expectations of it.
大概在这一点上，它可能已经超过70%，它完全超出了人们的预期。
Lenny Rachitsky: You guys are at the edge of where things are heading.
Lenny Rachitsky：你们正处于事情发展的边缘。
Mike Krieger: I had the very bizarre experience of I had two tabs open.
Mike Krieger：我有过一次非常奇怪的经历，我打开了两个选项卡。
It was AI 2027, and my product strategy, and it was this moment where I'm like, "Wait, am I the character in the story?"
这是 AI 2027，也是我的产品策略，正是在这一刻，我想，“等等，我是故事中的角色吗？”
Lenny Rachitsky: It feels like ChatGPT is just winning in consumer mind share.
Lenny Rachitsky：感觉 ChatGPT 在消费者心智占有率方面取得了胜利。
How does that inform the way you think about product, strategy, and mission?
这如何影响您对产品、战略和使命的思考方式？
Mike Krieger: I think there's room for several generationally important companies to be built in AI right now.
Mike Krieger：我认为现在对于几家具有重要意义的公司来说，人工智能都有发展的空间。
How do we figure out what we want to be when we grow up versus what we currently aren't or wish that we were or see other players in the space being?
我们如何弄清楚长大后我们想成为什么，以及我们现在不是什么，或者希望我们成为或看到这个领域的其他参与者是什么？
Lenny Rachitsky: What's something that you've changed your mind about what AI is capable of and where AI is heading?
Lenny Rachitsky：对于人工智能的能力以及人工智能的发展方向，您改变了哪些看法？
Mike Krieger: I had this notion coming in like, "Yes, these models are great, but are they able to have an independent opinion?" And it's actually really flipped for me only in the last month.
Mike Krieger：我的想法是，“是的，这些模型很棒，但他们能够有独立的意见吗？”实际上直到上个月我才真正发生了转变。
Lenny Rachitsky: Today, my guest is Mike Krieger.
莱尼·拉奇茨基：今天，我的客人是迈克·克里格。
Mike is chief product officer at Anthropic, the company behind Claude.
迈克是克劳德背后的公司 Anthropic 的首席产品官。
He's also the co-founder of Instagram.
他也是 Instagram 的联合创始人。
He's one of my most favorite product builders and thinkers.
他是我最喜欢的产品构建者和思想家之一。
He's also now leading product at one of the most important companies in the world, and I'm so thrilled to have had a chance to chat with him on the podcast.
他现在还在世界上最重要的公司之一领导产品，我很高兴有机会在播客上与他聊天。
We chat about what he's changed his mind about most in terms of AI capabilities in the years since he joined Anthropic, how product development changes and where bottlenecks emerge when 90% of your code is written by AI, which is now true at Anthropic.
我们讨论了自从他加入 Anthropic 以来，他对 AI 功能的看法发生了最大的改变，产品开发发生了怎样的变化，以及当 90% 的代码由 AI 编写时，瓶颈出现在哪里，而现在在 Anthropic 也是如此。
Also, his thoughts on OpenAI versus Anthropic, the future of MCP, why he shut down Artifact, his last startup and how he feels about it.
此外，他对 OpenAI 与 Anthropic 的看法、MCP 的未来、他关闭最后一家初创公司 Artifact 的原因以及他对此的感受。
Also, what skills he's encouraging his kids to develop with the rise of AI.
此外，随着人工智能的兴起，他鼓励孩子们发展哪些技能。
And we closed the podcast on a very heartwarming message that Claude wanted me to share it with Mike.
我们以一条非常温馨的信息结束了播客，克劳德希望我与迈克分享。
Lenny Rachitsky: A big thank you to my newsletter Slack community for suggesting topics for this conversation.
Lenny Rachitsky：非常感谢我的时事通讯 Slack 社区为本次对话提出的主题。
If you enjoy this podcast, don't forget to subscribe it and follow it in your favorite podcasting app or YouTube.
如果您喜欢这个播客，请不要忘记订阅它并在您最喜欢的播客应用程序或 YouTube 中关注它。
Also, if you become an annual subscriber of my newsletter, you get a year free of a bunch of incredible products, including Linear, Superhuman, Notion, Perplexity and Granola.
此外，如果您成为我的时事通讯的年度订阅者，您将获得一年免费使用一堆令人难以置信的产品，包括 Linear、Super human、Notion、Perplexity 和 Granola。
Check it out at lennysnewsletter.com and click bundle.
请访问 lennysnewsletter.com 查看并单击“捆绑”。
Lenny Rachitsky: With that, I bring you Mike Krieger.
Lenny Rachitsky：接下来，我为您带来 Mike Krieger。
This episode is brought to you by Productboard, the leading product management platform for the enterprise.
本集由企业领先的产品管理平台 Productboard 为您带来。
For over 10 years, Productboard has helped customer-centric organizations like Zoom, Salesforce, and Autodesk build the right products faster.
十多年来，Productboard 帮助 Zoom、Salesforce 和 Autodesk 等以客户为中心的组织更快地构建合适的产品。
And as an end-end platform, Productboard seamlessly supports all stages of the product development lifecycle.
作为一个终端平台，Productboard 无缝支持产品开发生命周期的所有阶段。
From gathering customer insights to planning a roadmap, to aligning stakeholders, to earning customer buy-in, all with a single source of truth.
从收集客户见解到规划路线图，到协调利益相关者，再到赢得客户的支持，所有这些都具有单一的事实来源。
Lenny Rachitsky: And now product leaders can get even more visibility into customer needs.
Lenny Rachitsky：现在产品领导者可以更深入地了解客户需求。
With Productboard Pulse, a new voice of customer solution built-in intelligence helps you analyze trends across all of your feedback and then dive deeper by asking AI your follow-up questions.
借助 Productboard Pulse，新的客户声音解决方案内置智能可帮助您分析所有反馈的趋势，然后通过向 AI 询问您的后续问题来更深入地了解。
See how Productboard can help your team deliver higher impact products that solve real customer needs and advance your business goals.
了解 Productboard 如何帮助您的团队交付更具影响力的产品，满足真正的客户需求并推进您的业务目标。
For a special offer and free 15-day trial, visit productboard.com/lenny.
如需特别优惠和 15 天免费试用，请访问productboard.com/lenny。
That's productboard.com/L-E-N-N-Y.
这是productboard.com/L-E-N-N-Y。
Lenny Rachitsky: Last year, 1.3% of the global GDP flowed through Stripe.
Lenny Rachitsky：去年，全球 GDP 的 1.3% 流经了 Stripe。
That's over $1.4 trillion and driving that huge number are the millions of businesses growing more rapidly with Stripe.
这一数字超过 1.4 万亿美元，推动这一巨大数字的是数以百万计的企业借助 Stripe 实现了更快的增长。
For industry leaders like Forbes, Atlassian, OpenAI, and Toyota, Stripe isn't just financial software.
对于福布斯、Atlassian、OpenAI 和丰田等行业领导者来说，Stripe 不仅仅是财务软件。
It's a powerful partner that simplifies how they move money, making it as seamless and borderless as the internet itself.
它是一个强大的合作伙伴，可以简化他们的资金转移方式，使其像互联网本身一样无缝和无国界。
For example, Hertz boosted its online payment authorization rates by 4% after migrating to Stripe.
例如，Hertz 在迁移到 Stripe 后将其在线支付授权率提高了 4%。
And imagine seeing a 23% lift in revenue like Forbes did just six months after switching to Stripe for subscription management.
想象一下，在改用 Stripe 进行订阅管理后仅仅六个月，收入就像《福布斯》那样增长了 23%。
Stripe has been leveraging AI for the last decade to make its product better at growing revenue for all businesses.
过去十年来，Stripe 一直在利用人工智能来使其产品更好地为所有企业增加收入。
From smarter checkouts to fraud prevention and beyond.
从更智能的结账到欺诈预防等。
Join the ranks of over half of the Fortune 100 companies that trust Stripe to drive change, learn more at Stripe.com.
加入超过一半的财富 100 强公司的行列，相信 Stripe 能够推动变革，请访问 Stripe.com 了解更多信息。
Lenny Rachitsky: Mike, thank you so much for being here and welcome to the podcast.
Lenny Rachitsky：迈克，非常感谢您来到这里，欢迎来到播客。
Mike Krieger: I'm really happy to be here.
迈克·克里格：我很高兴来到这里。
I've been looking forward to this for a while.
我对此已经期待了一段时间了。
Lenny Rachitsky: Wow, I had love to hear that.
Lenny Rachitsky：哇，我很高兴听到这个。
I've also been looking forward to this for a while.
我也对此期待了一段时间。
I have so much to talk about.
我有很多话要说。
So first of all, you've been at Anthropic for just over a year at this point.
首先，您在 Anthropic 工作刚刚一年多。
Congrats by the way on hitting the cliff.
顺便恭喜你撞上了悬崖。
Mike Krieger: Thank you.
迈克·克里格：谢谢。
Not that we're tracking.
不是我们正在追踪。
Lenny Rachitsky: That's right.
莱尼·拉奇茨基：没错。
So let me just ask you this.
所以让我问你这个问题。
So you've been at Anthropic for about a year.
你已经在 Anthropic 工作了大约一年了。
What's something that you've changed your mind about from before you joined Anthropic to today about what AI is capable of and where AI is heading?
从加入 Anthropic 之前到今天，关于人工智能的能力以及人工智能的发展方向，您的看法有哪些改变？
Mike Krieger: Two things.
迈克·克里格：有两件事。
One is like a pace and timeline question.
一是节奏和时间表问题。
The other one is a capability question.
另一个是能力问题。
So maybe I'll take the second one first.
所以也许我会先选择第二个。
I had this notion coming in, yes, these models are great, they're going to be able to produce code, they're going to be able to write hopefully in your voice eventually, but are they able to sort of have an independent opinion?
我有这样的想法，是的，这些模型很棒，它们将能够生成代码，它们最终将能够用你的声音编写，但它们能够有独立的意见吗？
And it's actually really flipped for me only in the last month and only with Opus 4 where my go-to product strategy partner is Claude.
事实上，直到上个月，只有在 Opus 4 上，我的首选产品策略合作伙伴是 Claude，情况才真正发生了转变。
And it has been basically for that full year where I'll write an initial strategy, I'll share it with Claude basically, and I'll have it, look at it.
基本上在那一整年里，我都会写一个初步的策略，基本上我会与克劳德分享，我会拥有它，看看它。
And in the past it's pretty anodyne kind of comments that it would leave, "Oh, have you thought about this?" And it's like, "Yeah, I thought about that." And Opus 4, I was working on some strategy for our second half of the year was the first one.
在过去，它会留下相当镇痛的评论，“哦，你考虑过这个吗？”就像，“是的，我考虑过这一点。” Opus 4，我正在制定下半年的一些策略，这是第一个策略。
Mike Krieger: It was like Opus 4 combined with our advanced research.
Mike Krieger：这就像 Opus 4 与我们先进研究的结合。
But it really went out for a while and it came back and I was like, you really looked at it in a new way.
但它确实消失了一段时间，然后又回来了，我想，你真的以一种新的方式看待它。
And so that's a thing that I've maybe I didn't feel like it would never be able to do that, but I wasn't sure how soon it'd be able to come up with something where I look at it, I'm like, yep, that is a new angle that I hadn't been looking at before and I'm going to incorporate that immediately into how I think about it.
所以这是我可能不觉得它永远无法做到这一点的事情，但我不确定它多久才能拿出一些东西，我看着它，我想，是的，这是一个我以前没有看过的新角度，我将立即将其纳入我的想法中。
So that's probably the biggest shift that I've had is, I don't know about independence is the right word, but creativity and sort of novelty of thought relative to how I'm thinking about things.
所以这可能是我经历过的最大的转变，我不知道独立这个词是否正确，但创造力和相对于我思考事物的方式的新颖性。
But in the timeline, one, it's so interesting because I was sitting next to Dario yesterday and he's like, "I keep making these predictions and people keep laughing at me.
但在时间线中，它非常有趣，因为我昨天坐在达里奥旁边，他说，“我一直在做出这些预测，人们一直在嘲笑我。
And then they come true."
然后它们就成真了。”
Mike Krieger: And it's funny to have this happen over and over again and he is like, not all of them are going to be right.
迈克·克里格（Mike Krieger）：这种情况一次又一次地发生很有趣，他想，并不是所有的人都是对的。
But even I think as of last year he was talking about we're at 50% on SWE-Bench, which is this benchmark around how well the models are at coding.
但即使我认为，截至去年，他还在谈论我们在 SWE-Bench 上的进度为 50%，这是关于模型编码水平的基准。
He's like, "I think we'll be at 90% by the end of 2025 or something like that." And sure enough, we're at about 72 now with the new models and we're at 50% when he made that prediction.
他说，“我认为到 2025 年底我们会达到 90% 或者类似的水平。”果然，新模型现在我们的进度约为 72%，而当他做出预测时，我们的进度为 50%。
And it's continued to scale pretty much as predicted.
而且它的规模继续扩大，几乎与预期一致。
And so I've taken the timelines a lot more seriously now.
所以我现在更加认真地对待时间表。
And I don't know if you read AI 2027-
我不知道你是否读过《AI 2027》——
Lenny Rachitsky: I have, it made by heart race.
Lenny Rachitsky：我有，它是由心跳造成的。
Mike Krieger: And I had the very bizarre experience of I had two tabs open, it was AI 2027 and my product strategy.
Mike Krieger：我有过一次非常奇怪的经历，我打开了两个选项卡，分别是 AI 2027 和我的产品策略。
And it was this moment where I'm like, "Wait, am I the character in the story?
就在这一刻，我想，“等等，我是故事中的角色吗？
How much is this converging?" But you read that and you're like, "Oh, 2027, that's years away if you're like no, mid 2025." And things continue to improve and the models continue to be able to do more and more and they're able to act agentically and they're able to have memory and they're able to act over time.
这收敛到什么程度了？”但当你读到这篇文章时，你会想，“哦，2027 年，如果你不喜欢的话，那就是几年后的事情了，那就是 2025 年中期。”事情在不断改进，模型继续能够做越来越多的事情，它们能够代理行动，它们能够拥有记忆，并且能够随着时间的推移采取行动。
So I think my confidence in the timelines and I don't know exactly how they manifest it definitely just solidified over the last year.
因此，我认为我对时间表的信心在过去一年中得到了巩固，而且我不知道它们具体是如何体现的。
Lenny Rachitsky: Wow.
莱尼·拉奇茨基：哇。
I wasn't expecting to go down that that paper was scary.
我没想到会发现那张纸很可怕。
And I'm curious just I guess I can't help but ask just thoughts on just how do we avoid the scary scenario that paper paints of where AI getting really smart goes?
我很好奇，我想我忍不住想问一下，我们如何避免纸上描绘的人工智能变得真正聪明的可怕场景？
Mike Krieger: Yeah, this maybe ties into, I've been here a year, why did I join Anthropic?
Mike Krieger：是的，这可能与我来这里一年了，为什么要加入 Anthropic 有关？
I was watching the models get better and even you could see it in early 2024, and looking at my kids, I'm like, "All right, they're going to grow up in a world with AI.
我看着模型变得更好，甚至你可以在 2024 年初看到它，看着我的孩子，我想，“好吧，他们将在一个人工智能的世界中长大。
It's unavoidable." Where can I maximally apply my time to nudge things towards going well?
这是不可避免的。”我可以在哪里最大限度地利用时间来推动事情顺利进行？
And I mean that's a lot of what people think about across the industry, especially at Anthropic.
我的意思是，这是整个行业人们的很多想法，尤其是在 Anthropic。
And so I think coming to an agreement and a shared framework and understanding of what does going well look like?
所以我认为达成一致和共享框架并理解什么是顺利进行？
What is the kind of human AI relationship that we want?
我们想要什么样的人类人工智能关系？
Mike Krieger: How will we know along the way?
Mike Krieger：一路上我们如何知道？
What do we need to build and develop and research along the way?
在此过程中我们需要构建、开发和研究什么？
I think those are all the kind of key questions.
我认为这些都是关键问题。
And some of those are product questions and some of those are research and interpretability questions, but for me it was the strongest reason to join was okay.
其中一些是产品问题，一些是研究和可解释性问题，但对我来说，这是加入的最有力的理由。
I think there's a lot of contribution that Anthropic can have around nudging things to go better.
我认为 Anthropic 可以做出很多贡献，推动事情变得更好。
And if I can have a part to play there, let's do it.
如果我能在那里发挥作用，那就去做吧。
Lenny Rachitsky: I love that answer.
Lenny Rachitsky：我喜欢这个答案。
Speaking of kids, so you've got two kids, I've got a young kid, he's just about to turn two.
说到孩子，你有两个孩子，我有一个小孩，他即将两岁了。
I'm curious just what skills you're encouraging your kids to build as this AI becomes more and more of our future and some jobs will be changed and just what advice do you have?
我很好奇，随着人工智能越来越成为我们的未来，一些工作将会发生变化，您鼓励您的孩子培养哪些技能，您有什么建议？
Mike Krieger: We have this breakfast feed breakfast with the kids every morning and sometimes some question will come up, something about physics and our oldest kid's almost six, but they ask funny questions about the solar system or physics or in a 6-year-old way and before we reach for Claude, because at first my instinct is like, "Oh, I wonder how Claude will do this question." And we started changing, "Well, how would we find out?" And the answer can't just be we'll ask Claude, all right, well, we could do this experiment, we could have this thing.
迈克·克里格：我们每天早上和孩子们一起吃早餐，有时会出现一些问题，一些关于物理的问题，我们最大的孩子快六岁了，但他们会问关于太阳系或物理的有趣问题，或者在我们去找克劳德之前以 6 岁的方式，因为一开始我的直觉是，“哦，我想知道克劳德会如何做这个问题。”然后我们开始改变，“那么，我们怎样才能找到答案呢？”答案不能只是我们问克劳德，好吧，我们可以做这个实验，我们可以拥有这个东西。
So I think nurturing curiosity and still having a sense of, I don't know, the scientific process sounds grandiose to instill in a 6-year-old, but that process of discovery and asking questions and then systematically working right through, I think will still be important.
所以我认为培养好奇心，同时仍然有一种意识，我不知道，向一个 6 岁的孩子灌输科学过程听起来很宏伟，但我认为发现、提出问题、然后系统地完成工作的过程仍然很重要。
And of course AI will be an incredible tool for helping resolve large parts of that, but that process of inquiry I think is still really important and independent thought.
当然，人工智能将成为一个令人难以置信的工具，可以帮助解决大部分问题，但我认为这个探究过程仍然非常重要和独立的思考。
Mike Krieger: My favorite moment with my kid, because she's very headstrong, our 6-year-old, she said something and I wasn't sure if it was true.
迈克·克里格：我最喜欢和我的孩子在一起的时刻，因为她非常任性，我们六岁的孩子，她说了一些话，我不确定这是否属实。
It was, oh, is that coral is an animal or corals alive?
那是，哦，珊瑚是动物还是活的珊瑚？
I don't even remember what the details of it.
我什至不记得它的细节。
And I was like, "I don't know if that's true." And she's like, "It's definitely true, dad." I'm like, "All right, let's ask Claude on this one." And she's like, "You can ask Claude, but I know I'm right." And I'm like I love that.
我当时想，“我不知道这是不是真的。”她说：“这绝对是真的，爸爸。”我说：“好吧，这个问题我们去问问克劳德吧。”她说，“你可以问克劳德，但我知道我是对的。”我就像我喜欢那样。
I want that kind of level of not just delegating all of your cognition to the AI because it won't always get it right.
我想要的不仅仅是将你所有的认知委托给人工智能，因为它并不总是正确的。
And also it kind of short circuits any kind of independent thought.
而且它也短路了任何一种独立的思想。
So the skill of asking questions, inquiry and independent thinking, I think those are all the pieces.
因此，我认为提出问题、探究和独立思考的技能就是这些。
What that looks like from a job or occupation perspective, I'm just keeping an open mind and I'm sure that'll radically change between now and then.
从工作或职业的角度来看，我只是保持开放的态度，我相信从现在到那时，情况会发生根本性的变化。
Lenny Rachitsky: It's interesting.
莱尼·拉奇茨基：这很有趣。
Tobias Lütke, Shopify CEO, on the podcast and he had the same answer for what he's encouraging his kids to develop is curiosity.
Shopify 首席执行官 Tobias Lütke 在播客中给出了相同的答案，他鼓励孩子培养好奇心。
And so it's interesting that's a common thread.
有趣的是，这是一个共同点。
Mike Krieger: The K through eight school our kid goes through had an AI and education expert come in and I had a very low bar or a very low expectation of what this conversation was going to be like.
Mike Krieger：我们的孩子就读的幼儿园到八年级学校有一位人工智能和教育专家进来，我对这次对话的门槛很低或期望很低。
And actually I think it went over most of the people in the audience's heads because he was like, "All right, well let me take you all the way back to Claude Shannon in information theory." And I could see people's eyes going, "What did I sign up for and why am I hearing this school auditorium hearing about information theory?" But he did a really nice job I think of also just imagining there will be different jobs and we don't know what those jobs are going to be and so what are the skills and techniques and remain open mindedness around what the exact way we recombine those things.
事实上，我认为大多数观众都在思考这个问题，因为他说：“好吧，让我带你回到信息论中的克劳德·香农。”我可以看到人们的目光在转，“我报名参加了什么？为什么我会在学校礼堂听到信息论？”但他做得非常好，我想也只是想象会有不同的工作，我们不知道这些工作会是什么，所以技能和技术是什么，并对我们重新组合这些东西的确切方式保持开放的态度。
And even those will probably change three times between now and when they're 18.
甚至从现在到他们 18 岁时，这些也可能会改变三次。
Lenny Rachitsky: So we're talking about timelines and how things are changing.
Lenny Rachitsky：所以我们正在讨论时间表以及事情正在如何变化。
So I've seen these stats that you've shared, other folks at Anthropic have shared about how much of your code is now written by AI.
我看到了你分享的这些统计数据，Anthropic 的其他人也分享了你现在有多少代码是由人工智能编写的。
So people have shared stats from 70% to 90%.
所以人们分享的统计数据从 70% 到 90% 不等。
There was an engineer lead that shared 90% of your code roughly is written by AI now, which first of all is just insane that it went from zero to 90%, I don't know, a few years, something like that.
有一位工程师领导分享说，现在你的代码大约有 90% 是由 AI 编写的，这首先是疯狂的，它从零到 90%，我不知道，几年之类的。
Yeah, basically.
是的，基本上。
I don't think people are talking about this enough.
我认为人们对这个问题的讨论还不够。
That's just wild.
这太疯狂了。
You guys are basically at the bleeding edge.
你们基本上处于最前沿。
I've never heard a company that has this high a percentage of code being written by AI.
我从未听说过哪家公司的人工智能编写的代码比例如此之高。
Lenny Rachitsky: So you guys are at the edge of where things are heading.
Lenny Rachitsky：所以你们正处于事情发展的边缘。
I think most companies will get here.
我认为大多数公司都会到达这里。
How has product development changed knowing so much of your code is now written by AI, so usually it's like PM, it's like here's what we're building, engineer builds it, it ships it.
产品开发发生了怎样的变化，知道现在有这么多代码是由 AI 编写的，所以通常就像 PM，这就像我们正在构建的东西，工程师构建它，然后交付它。
Is it still kind of roughly that or is it now PMs are just going straight to Claude, build this thing for me, engineers are doing different things?
仍然是这样吗？还是现在 PM 直接去找 Claude，为我构建这个东西，工程师正在做不同的事情？
Just what looks different in a world where 90% of your code is written by AI?
在 90% 的代码由人工智能编写的世界中，到底有什么不同呢？
Mike Krieger: Yeah, it's really interesting because I think the role of engineering has changed a lot, but the suite of people that come together to produce a product hasn't yet.
Mike Krieger：是的，这真的很有趣，因为我认为工程的角色已经发生了很大变化，但聚集在一起生产产品的人员还没有改变。
And I think for the worst in a lot of ways because I think we're still holding on some assumptions.
我认为在很多方面都是最坏的情况，因为我认为我们仍然坚持一些假设。
So I think the roles are still fairly similar, although we'll now get in my favorite things that happen now are some nice PMs that have an idea that they want to express or designers that have an idea they want to express will use Claude and maybe even Artifacts to put together an actual functional demo.
所以我认为这些角色仍然相当相似，尽管我们现在最喜欢发生的事情是一些好的 PM，他们有一个他们想要表达的想法，或者设计师有他们想要表达的想法，他们会使用 Claude 甚至 Artifacts 来组合一个实际的功能演示。
And that has been very, very helpful.
这非常非常有帮助。
No, no, this is what I mean that makes it tangible.
不，不，这就是我的意思，让它变得有形。
That's probably the biggest role shift is prototyping happening earlier in the process via more of this code plus design piece.
这可能是最大的角色转变是通过更多的代码和设计片段在流程的早期进行原型设计。
What I've learned though is the process of knowing what to ask the AI, how to compose the question, how to even think about structuring a change between the backend and the front end.
不过，我学到的是知道要问人工智能什么、如何提出问题、甚至如何考虑在后端和前端之间构建变更的过程。
Mike Krieger: Those are still very difficult and specialized skills and they still require the engineer to think about it.
Mike Krieger：这些仍然是非常困难且专业的技能，仍然需要工程师进行思考。
And we really rapidly became bottlenecked on other things like our merge queue, which is the get in line to get your change accepted by the system that then deploys into production.
我们很快就在其他事情上遇到了瓶颈，比如我们的合并队列，这是让系统接受您的更改然后部署到生产中的队列。
We had to completely re-architect it because so much more code was being written and so many more pull requests were being submitted that it just completely blew out the expectations of it.
我们必须完全重新架构它，因为要编写更多的代码，提交更多的拉取请求，以至于它完全超出了人们的预期。
And so it's like, I don't know if you've ever read, is it the goal, the classic process optimization book, and you realize there's this critical path theory.
所以就像，我不知道你是否读过《目标》这本经典的流程优化书籍，然后你就会意识到有这个关键路径理论。
I've just found all these new bottlenecks in our system, there's an upstream bottleneck, which is decision making and alignment.
我刚刚在我们的系统中发现了所有这些新的瓶颈，其中有一个上游瓶颈，即决策和协调。
A lot of things that I'm thinking about right now is how do I provide the minimum viable strategy to let people feel empowered to go run and type and build and explore at the edge of model capabilities.
我现在正在考虑的很多事情是如何提供最低可行的策略，让人们感到有能力在模型功能的边缘进行运行、打字、构建和探索。
Mike Krieger: I don't think I've gotten that right yet, but that's something I'm working on.
Mike Krieger：我认为我还没有做对，但这是我正在努力的事情。
And then once the building is happening, other bottlenecks emerge, let's make sure we don't step on each other's toes.
一旦建设开始，其他瓶颈就会出现，让我们确保我们不会踩到彼此的脚趾。
Let's think through all the edge cases here ahead of time so that we're not blocked on the engineering side.
让我们提前考虑一下这里的所有边缘情况，这样我们就不会在工程方面受到阻碍。
And then when the work is complete and we're getting ready to ship it, what are all those bottlenecks as well?
然后，当工作完成并准备发货时，所有这些瓶颈是什么？
Let's do the air traffic control of landing the change.
我们来进行着陆变更的空中交通管制。
How do we figure out large strategy?
我们如何制定大战略？
So I think there hasn't been as much pressure on changing those until this year, but I would expect that a year from now the way that we are conceiving of building and shipping software just changes a lot because it's going to be very painful to do it the current way.
因此，我认为直到今年才面临改变这些的压力，但我预计一年后我们构想的构建和交付软件的方式会发生很大变化，因为按照目前的方式来做这件事会非常痛苦。
Lenny Rachitsky: Wow, that is extremely interesting.
Lenny Rachitsky：哇，这非常有趣。
So it used to be here's an idea, let's go design it, build it, ship it, merge it, and then ship it.
所以它曾经是一个想法，让我们去设计它，构建它，发布它，合并它，然后发布它。
And usually the bottleneck was engineering, taking time to build a thing and then design.
通常瓶颈是工程，需要时间来构建一个东西然后进行设计。
And now you're saying the two bottlenecks you're finding are okay deciding what to build and aligning everyone and then it's actually the cue to merge it into production.
现在你说你发现的两个瓶颈可以决定构建什么并协调每个人，然后它实际上是将其合并到生产中的提示。
And I imagine review it too is probably a part-
我想审查它也可能是一部分-
Mike Krieger: Reviewing has really changed too.
Mike Krieger：审查也确实发生了变化。
And in many ways perhaps unsurprisingly the team that works in the most futuristic way is the Claude Code team because they're using Claude Code to build Claude Code in a very self-improving kind of way.
从很多方面来看，以最具未来感的方式工作的团队可能并不令人意外的是 Claude Code 团队，因为他们使用 Claude Code 以一种非常自我改进的方式构建 Claude Code。
And early on in that project, they would do very line by line pull request reviews in the way that you would for any other project.
在该项目的早期，他们会像对任何其他项目一样进行逐行拉取请求审查。
And they've just realized Claude is generally right and it's producing pull requests that are probably larger than most people are going to be able to review.
他们刚刚意识到克劳德总体上是正确的，并且它产生的拉取请求可能比大多数人能够审查的还要大。
So can you use a different Claude to review it and then do the human almost acceptance testing more than trying to review line by line.
那么，您是否可以使用不同的 Claude 来审查它，然后进行人类几乎验收测试，而不是尝试逐行审查。
There's definitely pros and cons and so far it's gone well.
肯定有利有弊，而且到目前为止进展顺利。
But I could also imagine it going off the rails and then having a completely both unmaintainable or even understandable by Claude Code base that hasn't happened, but watching them change their review processes definitely has been interesting.
但我也可以想象它会脱离轨道，然后有一个完全无法维护甚至克劳德代码库可以理解的情况，但这种情况并没有发生，但看着他们改变他们的审查流程绝对是有趣的。
Mike Krieger: And yeah, the merge queue is one instance of the bottom bottleneck that forms down there, but there's other ones which is how do we make sure that we're still building something coherent and packaging it up into a moment that we can share with people and whether that's around a launch moment, whether that's about then enabling people to use this thing and talking about it, the classic things of building something useful for people and then making it known that you've built it and then learning from their feedback still exists.
Mike Krieger：是的，合并队列是底部瓶颈的一个实例，但还有其他一些问题，即我们如何确保我们仍在构建连贯的东西并将其打包成一个我们可以与人们分享的时刻，无论是在发布时刻，还是让人们能够使用这个东西并谈论它，构建对人们有用的东西，然后让人们知道你已经构建了它，然后从他们的反馈中学习的经典做法仍然存在。
We've just made a portion of that whole process much more efficient.
我们刚刚使整个过程的一部分变得更加高效。
Lenny Rachitsky: I heard you describe this as you guys are patient zero for this way of working.
Lenny Rachitsky：我听到你描述了这一点，因为你们对这种工作方式是零耐心的。
Lenny Rachitsky: I love that.
莱尼·拉奇茨基：我喜欢这个。
Do you have a sense of what percentage of Claude Code is written by Claude Code?
你知道 Claude Code 有多少百分比是由 Claude Code 编写的吗？
Mike Krieger: At this point, I would be shocked if it wasn't 95% plus.
Mike Krieger：此时，如果不是 95% 以上，我会感到震惊。
I'd have to ask Boris and the other tech leads on there.
我得问问鲍里斯和其他技术主管。
But what's been cool is so nitty-gritty stuff, Claude Code is written in TypeScript.
但最酷的是，Claude Code 是用 TypeScript 编写的。
It's actually our largest TypeScript project.
这实际上是我们最大的 TypeScript 项目。
Most of the rest of Anthropic is written in Python, some Go, some Rust now, but we're not like a TypeScript shop.
Anthropic 的其余大部分都是用 Python 编写的，现在有一些是 Go，还有一些是 Rust，但我们不像 TypeScript 商店。
And so I saw a great comment yesterday in our Slack where somebody had this thing that was driving them crazy about Claude Code and they're like, "Well, I don't know any TypeScript, I'm just going to talk to Claude about it and do it."
所以我昨天在 Slack 上看到了一条很棒的评论，有人说这件事让他们对 Claude Code 着迷，他们说，“好吧，我不懂任何 TypeScript，我只是要和 Claude 讨论一下然后就去做。”
Mike Krieger: And they went from that to pull requests in an hour and solve their problem and they submitted a pull request and that breaking down the barriers.
Mike Krieger：然后他们在一小时内拉取请求并解决了他们的问题，他们提交了拉取请求并打破了障碍。
One, it changes your barrier to entry for any kind of newcomer to the project.
第一，它改变了任何类型的项目新手的进入门槛。
I think it can let you choose the right language for the right job for example.
例如，我认为它可以让您为正确的工作选择正确的语言。
I think that helps as well, but I think it also just reinforces Claude Code being that patient alpha of that where contributions from outside the team can be Claude coded as well.
我认为这也有帮助，但我认为这也只是强化了克劳德代码的耐心阿尔法，来自团队外部的贡献也可以被克劳德编码。
Lenny Rachitsky: Wow, this is, it's just continue to blow my mind all these things that you're sharing, 95% of Claude Code is written by Claude Code roughly.
Lenny Rachitsky：哇，这就是，你分享的所有这些事情都让我继续震惊，95% 的 Claude Code 都是由 Claude Code 大致编写的。
Mike Krieger: That's my guess.
迈克·克里格：这是我的猜测。
Yeah, I'll come back with the real stuff.
是的，我会带着真实的东西回来。
But I mean if you ask the team, that's how they're working and that's how they're getting contributions from across the company too.
但我的意思是，如果你问团队，他们就是这样工作的，也是他们从整个公司获得贡献的方式。
Lenny Rachitsky: It's interesting going back to your point about strategy being assisted by Claude itself and your point about how a lot of the bottlenecks now are kind of the top of the funnel of coming up with ideas aligning everyone, it's interesting that Claude is already helping with that also of helping you decide what to build.
Lenny Rachitsky：有趣的是，回到你关于克劳德本身协助战略的观点，以及你关于现在有很多瓶颈是如何提出与每个人保持一致的想法的漏斗顶部的观点，有趣的是，克劳德已经在帮助你决定构建什么。
So if those two bottlenecks are aligning, deciding what to build and then just merging and getting everything, where do you see the most interesting stuff happening to help you speed those things up?
因此，如果这两个瓶颈是一致的，决定构建什么，然后合并并获取所有内容，那么您在哪里看到最有趣的事情发生来帮助您加快这些事情？
Mike Krieger: Yeah, I think that on that first row, I started the year by writing a doc that was effectively how do we do product today and where is Claude not showing up yet that it should?
迈克·克里格（Mike Krieger）：是的，我认为在第一行，我通过写一份文档开始了这一年，该文档实际上是我们今天如何做产品以及克劳德在哪里还没有出现它应该出现的地方？
And I think that upstream part is the next one to go.
我认为上游部分是下一个要走的部分。
It's interesting.
这很有趣。
At your conference I talked to somebody who's working on a PRD, GPT kind of ChatPRD, I think was the-
在你们的会议上，我与一位从事 PRD、GPT 之类的 ChatPRD 工作的人进行了交谈，我认为是-
Mike Krieger: Yeah.
迈克·克里格：是的。
Can Claude be a partner in figuring out what to build?
克劳德可以成为合作伙伴来确定要构建什么吗？
What the market size is if you want to approach it that way?
如果你想以这种方式实现的话，市场规模有多大？
What the user needs are if you look at a different way?
如果换个角度看，用户需要什么？
We think a lot about the virtual collaborator on topic and one of the ways in which I think that can show up is, "Hey, I'm in the Discord, the Claude Anthropic Discord, I'm in the user Fora, I'm on X and I'm reading things and here's what's emergent." That's step one.
我们对主题上的虚拟协作者进行了很多思考，我认为可以出现的方式之一是，“嘿，我在 Discord 中，Claude Anthropic Discord，我在用户论坛中，我在 X 上，我正在阅读东西，这就是紧急情况。”这是第一步。
Models can do that today.
今天模型可以做到这一点。
Step two, which the models probably can do today, which have to wire them up to do it is and not only are the problems here's how I think you might be able to solve them.
第二步，模型今天可能可以做到，必须将它们连接起来才能做到这一点，这不仅是这里的问题，也是我认为您可能能够解决它们的方式。
And then taking that through to, and I put together a pull request to solve this thing that I'm seeing feels very achievable this year than stringing those things together and we're limited more.
然后，我提出了一个拉取请求来解决这个问题，我认为今年感觉比把这些事情串在一起非常容易实现，而且我们受到的限制更多。
Mike Krieger: This is why MCP is exciting to me.
Mike Krieger：这就是 MCP 让我兴奋的原因。
We're limited more around making sure the context flows through all of that so we have the right access to those things more than the model's capability to reason and propose.
我们在确保上下文贯穿所有这些内容方面受到更多限制，因此我们比模型的推理和建议能力更能正确访问这些内容。
Now the model might not have perfect UI taste yet, so there's definitely room for design to intervene and be like, "Oh, that's not quite how I would solve the problem of this not showing up." But I would get very excited.
现在该模型可能还没有完美的 UI 品味，所以肯定有设计干预的空间，比如，“哦，这并不是我解决这个不显示问题的方法。”但我会变得非常兴奋。
I would give you a really small example, but we changed on Claude AI, you should be able to just copy markdown from Artifacts or code from Artifacts and we changed it so you can actually download it and export it.
我会给你一个非常小的例子，但我们对 Claude AI 进行了更改，你应该能够从 Artifacts 复制 markdown 或从 Artifacts 复制代码，我们对其进行了更改，以便你可以实际下载并导出它。
We changed the button to export and we got a bunch of feedback like, "How do I copy now?" And the answer is you drop it down and it's copied.
我们更改了导出按钮，并收到了很多反馈，例如“我现在如何复制？”答案是你把它放下来，它就会被复制。
Mike Krieger: It's just mind one of those things where it's made sense, but we probably got it not quite right.
Mike Krieger：这只是有意义的事情之一，但我们可能不太正确。
That feedback was in the RUX channel.
该反馈来自 RUX 频道。
I would've loved an hour later for a plot to be like, "Hey, if we do want to change it back, here's the PR to do it." And by the way, eventually, and then I'm going to spin up an A/B test to see if this changes metrics and then we'll see how it looks in a week.
我希望一个小时后的情节是这样的：“嘿，如果我们确实想把它改回来，这里是执行此操作的公关。”顺便说一句，最终，我将启动 A/B 测试，看看这是否会改变指标，然后我们将在一周内看到它的样子。
If you told me that about a year and a half ago going to be like, "Ah, yeah, maybe like 27, maybe 26." But it really feels just at the tip of capabilities right now.
如果你在大约一年半前告诉我，你会说，“啊，是的，也许是 27 岁，也许是 26 岁。”但现在确实感觉还处于能力的尖端。
Lenny Rachitsky: Wow, okay.
莱尼·拉奇茨基：哇，好吧。
You mentioned the Lenny and Friends Summit.
你提到了莱尼和朋友们峰会。
I wanted to talk about this a bit.
我想谈谈这个。
So you were on a panel with Kevin Weil, the CPO of OpenAI, I think it was the first time you guys did this maybe the last time for now.
你们和 OpenAI 的首席产品官 Kevin Weil 一起参加了小组讨论，我认为这是你们第一次这样做，也许也是现在最后一次。
Mike Krieger: Yeah, we haven't done it since, not for any reason.
Mike Krieger：是的，从那以后我们就没有这样做了，没有任何原因。
I had a lot of fun.
我玩得很开心。
Lenny Rachitsky: What a legendary panel we assembled there with Sarah Guo moderating.
Lenny Rachitsky：我们在那里召集了一个多么传奇的小组，由莎拉·郭主持。
And you made this comment actually ended up being the most rewatched part of the interview, which is that you were putting product people on the model team and working with researchers making the model better and you're putting some product people on the product experience making the UX more intuitive, making all that better.
你的这个评论实际上最终成为了采访中最受关注的部分，那就是你将产品人员放入模型团队并与研究人员合作使模型变得更好，并且你将一些产品人员放入产品体验中，使用户体验更加直观，使一切变得更好。
And you found that almost all the leverage came from the product team working with the researchers.
你发现几乎所有的影响力都来自与研究人员合作的产品团队。
And so you've been doing more of that.
所以你一直在做更多这样的事情。
So first of all, does that continue to be true?
首先，这仍然是事实吗？
And second of all, what are the implications of that for product teams?
其次，这对产品团队有何影响？
Mike Krieger: It's continued to be true.
Mike Krieger：这仍然是事实。
And in fact I think that if the proportion was already skewing towards having more of that embedding, I've just become more and more convinced.
事实上，我认为如果比例已经偏向于拥有更多的嵌入，我只是变得越来越确信。
I didn't feel as strongly about it during the summit and now I feel really strongly about it.
在峰会期间我对此没有那么强烈的感觉，但现在我对此感觉非常强烈。
If we're shipping things that could have been built by anybody just using our models off the shelf, there's great stuff to be built by using our models off the shelf by the way, don't get me wrong, but where we should play and what we can do uniquely should be stuff that's really at that magic intersection between the two, right?
如果我们正在交付任何人都可以使用我们现成的模型来构建的东西，那么顺便说一句，通过使用我们现成的模型可以构建出很棒的东西，请不要误会我的意思，但是我们应该在哪里发挥作用以及我们可以做的独特事情应该是真正处于两者之间的神奇交叉点的东西，对吗？
Mike Krieger: Artifacts may a great example and if you play with Artifacts with Claude 4, that's an actually really interesting example where we took somebody from our, we have Claude code skills, which is a team that really is doing the post-training around teaching Claude some of these really specific skills and we paired it with some product people and then together we revamped how this looks in the product today and what Claude can do way better than just like, "Yeah, we just used the model and we prompted a little bit."
Mike Krieger：Artifacts 可能是一个很好的例子，如果你和 Claude 4 一起玩 Artifacts，这实际上是一个非常有趣的例子，我们从我们的团队中挑选了一些人，我们拥有 Claude 代码技能，这是一个团队，他们正在围绕教 Claude 一些非常具体的技能进行后期培训，我们将其与一些产品人员配对，然后我们一起改进了今天产品中的外观以及 Claude 可以做得更好，而不仅仅是像“是的，我们只是使用了该模型，并进行了一些提示。”
Mike Krieger: That's just not enough.
迈克·克里格：这还不够。
We need to be in that fine-tuning process.
我们需要进行微调。
So much of what, if you look at what we're working on right now, but we've shipped recently between research and all these other things are things that the functional unit of work at Anthropic is no longer take the model and then go work with design and product to go ship a product.
如果你看看我们现在正在做的事情，但我们最近在研究和所有其他事情之间交付了很多东西，Anthropic 的功能工作单元不再采用模型，然后进行设计和产品工作以交付产品。
It's more like we are in the post-training conversations around how these things should work and then we are in the building process and we're feeding those things back and looping them back.
这更像是我们在培训后围绕这些东西应该如何工作进行对话，然后我们在构建过程中，我们将这些东西反馈并循环回来。
Mike Krieger: I think it's exciting.
迈克·克里格：我认为这很令人兴奋。
It's also a new way of working that not all PMs have, but the PMs that have the most internal positive feedback from both research and engineering are the ones that get it that I was in a product review yesterday, I was like, "Oh, if we want to do this memory feature, we should talk to the researchers because we just shipped a bunch of memory capabilities in Claude 4." They're like, "Yeah, yeah, we've been talking to them for weeks, this is how we're manifesting it." It's like, "Okay, I feel good.
这也是一种新的工作方式，并非所有产品经理都有，但从研究和工程方面获得最内部积极反馈的产品经理是那些在我昨天的产品评论中得到的产品经理，我当时想，“哦，如果我们想做这个内存功能，我们应该和研究人员谈谈，因为我们刚刚在 Claude 4 中提供了一堆内存功能。”他们就像是，“是的，是的，我们已经和他们交谈了几周，这就是我们表现出来的方式。”就像是，“好吧，我感觉很好。
I feel like we're doing the right things now."
我觉得我们现在正在做正确的事情。”
Lenny Rachitsky: So let me pull on this thread more and there's something I've been thinking about along these lines.
Lenny Rachitsky：所以让我进一步讨论这个话题，沿着这些思路我一直在思考一些事情。
So essentially there's a big part of entropic that's building this super intelligent giga brain that's going to do all these things for us over time.
因此，本质上，熵的很大一部分正在构建这个超级智能的千兆大脑，随着时间的推移，它将为我们做所有这些事情。
And then, as you said, there's the product team that's building the UX around this super intelligent giga brain and over time this super intelligence is going to be able to build its own stuff.
然后，正如你所说，产品团队正在围绕这个超级智能的千兆大脑构建用户体验，随着时间的推移，这个超级智能将能够构建自己的东西。
And so I guess just where do you think the most value will come from traditional product teams over time?
所以我想随着时间的推移，您认为传统产品团队的最大价值将来自哪里？
I know this is different because you guys are a foundational alum company and not most companies don't work this way, but just, I don't know, thoughts on just the where most value will come from product teams over time working on AI.
我知道这是不同的，因为你们是一家基础性的校友公司，并不是大多数公司不这样做，但我不知道，随着时间的推移，产品团队在人工智能领域的工作中，最大的价值将来自于哪里。
Mike Krieger: I think there's still a lot of value in two things.
Mike Krieger：我认为有两件事仍然很有价值。
One is making this all comprehensible.
一是让这一切变得可以理解。
I think we've done an okay job.
我认为我们做得还不错。
I think we could do a much better job of making this conference.
我认为我们可以把这次会议做得更好。
What's still the difference between somebody who's really adept at using these tools in their work and most people is huge.
真正擅长在工作中使用这些工具的人和大多数人之间的区别仍然是巨大的。
And maybe that's the most literal answer to your earlier question around what skills to learn.
也许这就是您之前关于要学习哪些技能的问题的最直接的答案。
That is a skill to learn and use it in the same way that I remember we did computer lock class when I was in middle school.
这是一种学习和使用它的技能，就像我记得中学时我们上电脑锁课一样。
I remember being really good at Google and that was actually a skill back in the day to think in terms of this information is out there, how do I query for it?
我记得我在谷歌上非常擅长，这实际上是当时的一种技能，可以根据这些信息进行思考，我如何查询它？
How do I do it?
我该怎么做？
I think it actually was an advantage at the time.
我认为这在当时确实是一个优势。
Mike Krieger: Of course now Google is pretty good at figuring out what you're trying to do if you are only in the neighborhood and there's less of that research kind of need.
Mike Krieger：当然，如果你只是在附近，而且这种研究需求较少，那么现在谷歌非常擅长弄清楚你想做什么。
But I still think that's a necessary part of good product development, which is the capabilities are there and even if Claude can create products from scratch, what are you building and how do you make it Comprehensible?
但我仍然认为这是良好产品开发的必要组成部分，即能力，即使克劳德可以从头开始创建产品，你正在构建什么以及如何使其易于理解？
Still hard because I think that gets at this much deeper empathy and understanding of human needs and psychology.
还是很难，因为我认为这可以加深对人类需求和心理的同理心和理解。
I was a human community interaction major, I still been talking in my book here.
我是人类社区互动专业的，我仍然在我的书中谈论这一点。
I still feel like that is a very, very, very, very necessary skill.
我仍然觉得这是一项非常非常非常必要的技能。
So that's one.
这就是其中之一。
Two is, and this straight to call back to another one of your guests, strategy, how we win, where we'll play, figuring out where exactly you're going to want to, of all the things that you could be spending your time or your tokens or your computation on what you want to actually go and do.
第二个是，直接回电给你的另一位客人，策略，我们如何获胜，我们将在哪里玩，弄清楚你到底想要去哪里，以及你可能花费时间或代币或计算你真正想去做什么的所有事情。
Mike Krieger: You could be wider probably than you could before, but you can't do everything.
Mike Krieger：你可能比以前更广泛，但你不能做所有事情。
And even from an external perspective, if you're seen to be doing everything, it's way less clear around how you're positioning yourselves.
即使从外部角度来看，如果人们认为你在做所有事情，那么你对自己的定位就不太清楚了。
Like strategy I think is still the second piece.
就像策略一样，我认为仍然是第二部分。
And then the third one is opening people's eyes to what's possible, which is a continuation of making it understandable.
第三个是让人们看到可能性，这是使其变得易于理解的延续。
But we were in a demo with a financial services company recently and we were working on here's how you can use our analysis tool and MCP together and you could see their light up and you're like, "Ah, okay." We call it overhang.
但我们最近在与一家金融服务公司进行演示，我们正在研究如何将我们的分析工具和 MCP 结合使用，您可以看到它们亮起来，然后您就会说，“啊，好吧。”我们称之为悬垂。
The delta between what the models and the products can do and how they're being used on a daily basis.
模型和产品的功能与日常使用方式之间的差异。
Huge overhang.
巨大的悬垂。
So that's where still a very, very strong necessary role for product.
因此，这仍然是产品非常非常重要的必要角色。
Lenny Rachitsky: Okay, that's an awesome answer.
Lenny Rachitsky：好的，这是一个很棒的答案。
So essentially areas for product teams to lean into more is strategy, just getting better and better at strategy, figuring out what to build and how to win in the market, making it easier to help people understand how to leverage the power of these tools, the comprehensibility and kind of along those lines is opening people's eyes to the potential of these sorts of things.
因此，产品团队需要更多关注的本质领域是战略，只是在战略方面做得越来越好，弄清楚要构建什么以及如何在市场上获胜，从而更容易帮助人们理解如何利用这些工具的力量，沿着这些路线的可理解性和类型正在让人们看到这些东西的潜力。
That's where product can still help.
这就是产品仍然可以提供帮助的地方。
Lenny Rachitsky: Awesome.
莱尼·拉奇茨基：太棒了。
So along those lines actually, do you have any just prompting tricks for people, things that you've learned to get more out of Claude when you chat with it?
那么实际上，沿着这些思路，你是否有任何提示人们的技巧，或者当你与它聊天时你学到的东西，可以从克劳德那里得到更多？
Mike Krieger: Sometimes it's funny because in some ways we have the ultimate prompting job, which is to write the system prompt for Claudia AI and we publish all of these, which I think is another nice area of transparency.
Mike Krieger：有时这很有趣，因为在某些方面我们有最终的提示工作，那就是为 Claudia AI 编写系统提示，然后我们发布所有这些，我认为这是另一个很好的透明度领域。
And we are always careful when giving prompting advice because at least officially, but I'll give you the unofficial version because you don't want things to become like we think this works, but we're not sure why.
我们在提供提示性建议时总是很小心，因为至少是官方的，但我会给你非官方的版本，因为你不希望事情变得像我们认为的那样有效，但我们不确定为什么。
But I will do small things like in Claude Code and we actually do react to this very literally, but I always ask it to, if I wanted to use more reasoning, think hard and it'll use a different flow and I usually start with that.
但我会做一些小事情，比如在克劳德代码中，我们实际上确实对此做出了非常字面的反应，但我总是要求它，如果我想使用更多的推理，请努力思考，它会使用不同的流程，我通常会以此开始。
Nudging, there's a great essay around make the other mistake like if you tend to be too nice, can you focus on...
轻推一下，有一篇很棒的文章是关于犯另一个错误的，比如如果你倾向于太好，你能专注于......
Even if you're trying to be more critical or more blunt, you're probably not going to be the most critical blunt person in the world.
即使你试图变得更加挑剔或更直率，你也可能不会成为世界上最挑剔直率的人。
Mike Krieger: And so with Claude sometimes I'm like, "Be brutal, Claude, roast me.
迈克·克里格：有时我对克劳德会说，“残忍一点，克劳德，烤我吧。
Tell me what's wrong with this strategy." I know we were talking earlier about the Claude as thought partner around critiquing product strategy.
告诉我这个策略有什么问题。”我知道我们之前讨论过克劳德作为批评产品策略的思想伙伴。
I think I previously would say things like, "What could be better on this product strategy?" And I'm just like, "Just roast this product strategy," and Claude's like a pretty nice entity.
我想我以前会这样说：“这个产品策略还有什么可以做得更好的？”我只是想，“只是烘烤这个产品策略，”克劳德就像一个非常好的实体。
It's hard to push it to be super brutal, but it forces it to be a little bit more critical as well.
很难让它变得超级残酷，但它也迫使它变得更加批判。
The last thing I'll say is, so we have a team called Applied AI that does a lot of work with our customers around optimizing Claude for their use case.
我要说的最后一件事是，我们有一个名为 Applied AI 的团队，该团队与我们的客户一起围绕优化 Claude 的用例进行了大量工作。
And we basically took their insights and their way of working and we put it into a product itself.
我们基本上吸收了他们的见解和工作方式，并将其融入到产品本身中。
So if you go to our console, our work bench, we have this thing called the prompt improver where you describe the problem and you give it examples and Claude itself will agentically create and then iterate on a prompt for you.
因此，如果您进入我们的控制台、我们的工作台，我们有一个称为提示改进器的东西，您可以在其中描述问题并给出示例，Claude 本身将自动为您创建提示，然后迭代提示。
Mike Krieger: I find what comes out of that ends up being quite different than what my intuitions would've been for a good prompt.
Mike Krieger：我发现最终的结果与我对良好提示的直觉完全不同。
And so I'd encourage folks to also check that out even for their own use cases because while that tool is met for an API developer putting a prompt into their product, it's equally applicable for a person doing a prompt for themselves.
因此，我鼓励人们也检查一下，即使是针对他们自己的用例，因为虽然该工具适合 API 开发人员在他们的产品中添加提示，但它同样适用于为自己做提示的人。
It'll insert XML tags which no human is going to think to do ahead of time.
它将插入 XML 标签，而没有人会提前想到这样做。
It actually is very helpful for Claude to understand what it should be thinking versus what it should be saying, et cetera.
实际上，对于克劳德来说，理解它应该想什么、应该说什么等等非常有帮助。
So that's another one is watch our prompt improver and then note that Claude itself is a very good prompter of Claude.
因此，另一件事是观察我们的提示改进器，然后注意克劳德本身是克劳德的一个非常好的提示器。
Lenny Rachitsky: Awesome.
莱尼·拉奇茨基：太棒了。
Okay, so we're going to link to that, the prompt improver.
好的，所以我们将链接到那个，即提示改进器。
The core piece of advice you shared early is just do the opposite of what you would naturally do.
您之前分享的核心建议就是做与您自然会做的相反的事情。
So if you're trying to be nice, just be brutal, be very honest and frank with me.
所以，如果你想表现得友善，就残酷一点，对我非常诚实和坦率。
Mike Krieger: Exactly.
迈克·克里格：没错。
I find that works quite well.
我发现这效果很好。
What are the thought patterns that I've fallen into that you want to break me out of?
我陷入了哪些思维模式而你想让我摆脱？
Lenny Rachitsky: I saw you guys just today maybe launched a Rick Rubin collab where it said vibe coding.
Lenny Rachitsky：我今天看到你们可能推出了 Rick Rubin 合作项目，其中提到了氛围编码。
What's that all about?
这是怎么回事？
Mike Krieger: What I've heard about that.
Mike Krieger：我听说过这件事。
And again, a lot of the coalesce this week between model launch developer event and The Way of Code.
本周，模型发布开发者活动和 The Way of Code 再次融合了很多内容。
We had one of our co-founders, Jack Clark is our head of policy and he got connected to Rick Rubin because I think he's been thinking a lot about coding, the future of coding and creativity and they've stayed in touch.
我们有一位联合创始人，杰克·克拉克（Jack Clark）是我们的政策主管，他与里克·鲁宾（Rick Rubin）建立了联系，因为我认为他一直在思考编码、编码和创造力的未来，并且他们一直保持联系。
And Rick got excited about this idea of he was creating art and visualizations with Claude and then he had these ideas around the way of the vibe coder and they put together this, actually I mean I love almost everything Rick Rubin.
里克（Rick）对他与克劳德（Claude）一起创造艺术和可视化的想法感到兴奋，然后他围绕振动编码器的方式有了这些想法，他们将其组合在一起，实际上我的意思是我几乎喜欢里克·鲁宾（Rick Rubin）的一切。
So the aesthetic of it I think is just so on point too.
所以我认为它的美学也很重要。
But yeah, this sort of like med meditation is probably the right word.
但是，是的，这种类似医学冥想的词可能是正确的。
Meditation on creativity, working alongside AI coupled with this really rich, interesting visualizations.
对创造力的思考，与人工智能一起工作，再加上这种真正丰富、有趣的可视化。
But it's one of those things where internally they're like, "Oh yeah, and we're doing this Rick Rubin collab." We were like, "We're doing what?
但这是他们内部的想法之一，“哦，是的，我们正在做这个与 Rick Rubin 的合作。”我们当时想，“我们在做什么？
That's amazing."
这太棒了。”
Lenny Rachitsky: I looked at it briefly and there's that meme of him just thinking deeply, sitting on a computer with a mouth.
莱尼·拉奇茨基（Lenny Rachitsky）：我简单地看了一眼，就看到了他坐在电脑上用嘴深思熟虑的模因。
Mike Krieger: It's totally, it's like ASCII art vibe.
Mike Krieger：完全是这样，就像 ASCII 艺术氛围一样。
Lenny Rachitsky: I'm excited to have Andrew Luo joining us today.
Lenny Rachitsky：我很高兴安德鲁·罗今天加入我们。
Andrew is CEO of OneSchema, one of our long time podcast sponsors.
Andrew 是 OneSchema 的首席执行官，OneSchema 是我们的长期播客赞助商之一。
Welcome, Andrew.
欢迎，安德鲁。
Speaker 3: Thanks for having me, Lenny.
发言者 3：谢谢你邀请我，莱尼。
Great to be here.
很高兴来到这里。
Lenny Rachitsky: So what is new with one schema?
Lenny Rachitsky：那么一种模式有什么新变化呢？
I know that you work with some of my favorite companies like Ramp and Vanta and Watershed.
我知道您与我最喜欢的一些公司合作，例如 Ramp、Vanta 和 Watershed。
I heard you guys launch a new data intake product that automates the hours of manual work that teams spent importing and mapping and integrating CSV and Excel files.
我听说你们推出了一款新的数据获取产品，该产品可以自动执行团队导入、映射以及集成 CSV 和 Excel 文件所需的手动工作时间。
Speaker 3: Yes, so we just launched the 2.0 of OneSchema FileFeeds.
演讲者 3：是的，所以我们刚刚推出了 OneSchema FileFeeds 2.0。
We've rebuilt it from the ground up with AI.
我们用人工智能从头开始重建它。
We saw so many customers coming to us with teams of data engineers that struggled with the manual work required to clean messy spreadsheets.
我们看到很多客户带着数据工程师团队来到我们这里，他们在清理凌乱的电子表格所需的手动工作中遇到了困难。
FileFeeds 2.0 allows non-technical teams to automate the process of transforming CSV and Excel files with just a simple prompt.
FileFeeds 2.0 允许非技术团队通过简单的提示来自动化转换 CSV 和 Excel 文件的过程。
We support all of the trickiest file integrations, SFTP, S3, and even email.
我们支持所有最棘手的文件集成、SFTP、S3，甚至电子邮件。
Lenny Rachitsky: I can tell you that if my team had to build integrations like this, how nice would it be to take this off our roadmap and instead use something like OneSchema?
Lenny Rachitsky：我可以告诉你，如果我的团队必须构建这样的集成，那么将其从我们的路线图中删除并使用 OneSchema 之类的东西该多好？
Speaker 3: Absolutely, Lenny.
发言者 3：当然，莱尼。
We've heard so many horror stories of outages from even just a single bad record in transactions, employee files, purchase orders, you name it.
我们听过很多可怕的停电故事，哪怕只是交易、员工档案、采购订单等方面的一个不良记录。
Debugging these issues is often like finding a needle in a haystack.
调试这些问题通常就像大海捞针一样。
OneSchema stops any bad data from entering your system and automatically validates your files, generating error reports with the exact issues in all bad files.
OneSchema 可阻止任何不良数据进入您的系统并自动验证您的文件，生成错误报告，其中包含所有不良文件中的确切问题。
Lenny Rachitsky: I know that importing incorrect data can cause all kinds of pain for your customers and quickly lose their trust.
Lenny Rachitsky：我知道导入不正确的数据会给您的客户带来各种痛苦，并很快失去他们的信任。
Andrew, thank you so much for joining me.
安德鲁，非常感谢你加入我。
If you want to learn more, head on over to oneschema.co. That's oneschema.co.
如果您想了解更多信息，请访问 oneschema.co。那是 oneschema.co。
Lenny Rachitsky: Actually going back to the beginning of your journey at Anthropic, what's the story of you getting recruited at Anthropic?
Lenny Rachitsky：实际上回到您在 Anthropic 的旅程的开始，您被 Anthropic 聘用的故事是什么？
Is there anything fun there?
那里有什么好玩的吗？
Mike Krieger: It all started and I actually sent my friend this text.
Mike Krieger：一切都开始了，我实际上给我的朋友发了这条短信。
So Joel Lewenstein, who I've known, he and I built our first iPhone apps together in 2007 when the App Store was just out and you could still make money by selling dollar apps on the App Store back in the day.
我认识的 Joel Lewenstein 在 2007 年和我一起开发了我们的第一个 iPhone 应用程序，当时 App Store 刚刚推出，当时你仍然可以通过在 App Store 上销售美元应用程序来赚钱。
And we were both at Stanford together and we were friends and we've stayed in touch over years and we've never gotten to work together since then.
我们都曾一起在斯坦福大学读书，我们是朋友，多年来我们一直保持联系，但从那以后我们再也没有一起工作过。
We've just remained close.
我们只是保持亲密关系。
And I was coming out of the Artifact experience, I was trying to figure out, do I start another company?
当我从 Artifact 的经历中走出来时，我试图弄清楚，我是否应该创办另一家公司？
I don't think so.
我不这么认为。
I need a break from starting something from zero.
我需要从零开始做一些事情，休息一下。
Do I go work somewhere?
我去别的地方工作吗？
I don't know what company would I want to go work at.
我不知道我想去哪家公司工作。
And he reached out and he's like, "Look, I don't know if you at all considered joining something rather than starting something, but we're looking for a CPO.
他伸出手来，说道，“听着，我不知道你是否考虑过加入某件事而不是开始某事，但我们正在寻找一名 CPO。
Would you be interested in chatting?"
有兴趣聊聊吗？”
Mike Krieger: And at that time, Claude 3 had just come out and I was like, "Okay, this company's clearly got a good research team.
Mike Krieger：当时，Claude 3 刚刚问世，我就想，“好吧，这家公司显然拥有一支优秀的研究团队。
The product is so early still." And it was like, "Great, I'll take the meeting." And I first met with Daniela, was one of the co-founders and the president in Anthropic.
该产品还处于早期阶段。”就像是，“太好了，我会参加会议。”我第一次见到 Daniela，她是 Anthropic 的联合创始人之一兼总裁。
And just from the beginning I was like a breath of fresh air, very little grandiosity coming off the founders, I mean they're clear-eyed about what they're building.
从一开始我就像呼吸了一口新鲜空气，创始人们没有什么浮夸的，我的意思是他们对自己正在建设的东西很清楚。
They know what they don't know.
他们知道他们不知道什么。
How many times I talk to Dario always like Dario is like, "Look, I don't know anything about product, but here's an intuition." Usually the intuition is really good and leads to some good conversation, but I think that intellectual honesty and shared view of what it means to do AI in a responsible way, it just resonated.
有多少次我和达里奥交谈总是像达里奥一样，“看，我对产品一无所知，但这是一个直觉。”通常直觉确实很好，会带来一些好的对话，但我认为，知识的诚实和对以负责任的方式开发人工智能意味着什么的共同看法，只会引起共鸣。
Mike Krieger: I kept having this feeling in these interviews, this is the AI company I would've hoped to have found it if I had founded an AI company.
Mike Krieger：在这些采访中我一直有这样的感觉，如果我创办了一家人工智能公司，我希望能找到这家人工智能公司。
And that's kind of the bar around if I'm going to join something that should be where I'm going to go.
如果我要加入一些我应该去的地方，那就是一种障碍。
But what I realized, I actually hadn't joined a company since my first internship in college basically.
但我意识到，自从我在大学第一次实习以来，我实际上就没有加入过一家公司。
And I was like, "Oh, how do I onboard myself?
我当时想，“哦，我该如何加入自己？
How do I get myself up to speed?
我如何让自己跟上进度？
How do I balance making sweeping changes versus understanding what's not broken about it overall?" And looking back on a year, I think I made some changes too slowly.
我如何平衡进行彻底的改变和了解总体上没有被破坏的部分？”回顾这一年，我觉得我做一些改变太慢了。
I think there was ways reorganizing product that I could have made a change earlier.
我认为有一些方法可以重新组织我可以更早做出改变的产品。
And I think I didn't appreciate how much a couple of really key senior people can shape so much of product strategy.
我想我并没有意识到几个真正关键的资深人士能够在多大程度上塑造如此多的产品战略。
Mike Krieger: I'll harken back to Claude Code.
Mike Krieger：我要回顾一下 Claude Code。
Claude Code happened because Boris, who actually was a Boris Cherney, he was an Instagram engineer and one of our senior ICs there, we overlapped a bit, was started that project from scratch internal first and then we got it out and then shipped it.
Claude Code 的发生是因为 Boris，他实际上是 Boris Cherney，他是一名 Instagram 工程师，也是我们那里的高级 IC 之一，我们有一点重叠，首先在内部从头开始这个项目，然后我们把它拿出来，然后发货。
And that's the power of one or two really strong people.
这就是一两个真正强大的人的力量。
And I made this mistake, we need more headcount and we do, I think there's more work that we need to do and there's things that I want to be building.
我犯了这个错误，我们需要更多的人员，我们确实这样做了，我认为我们需要做更多的工作，而且我想要建立一些东西。
But more than that we need a couple of almost founder type engineers that maybe connect back to our question on what skills are useful and how does product development change.
但更重要的是，我们需要一些几乎创始人类型的工程师，他们可能会回答我们的问题，即哪些技能有用以及产品开发如何变化。
And maybe even more so I'm a huge believer in the founding engineer tech lead with an idea and pair them with the right design and product supports, help them realize that, I'm 10 times more a believer in that than before.
也许更重要的是，我非常相信创始工程师技术主管有一个想法，并将他们与正确的设计和产品支持配对，帮助他们意识到，我比以前更相信这一点。
Lenny Rachitsky: I actually asked people on Twitter what to ask you ahead of this conversation.
Lenny Rachitsky：我实际上在 Twitter 上问过人们在这次谈话之前要问你什么。
And the most common question surprisingly was why did you shut down Artifact?
令人惊讶的是，最常见的问题是为什么要关闭 Artifact？
And I also wondered that because I loved Artifact.
我也想知道这一点，因为我喜欢 Artifact。
I was a power user.
我是高级用户。
I was just like, "Finally a news app that I love that it's giving me what I want to know." So I guess just what happened there at the end?
我当时想，“终于有一个我喜欢的新闻应用程序了，它给了我我想知道的东西。”所以我猜最后到底发生了什么？
Mike Krieger: I still really miss it too.
Mike Krieger：我也非常怀念它。
I didn't find a replacement and I think I substituted it by visiting individual sites and keeping things up that way.
我没有找到替代品，我想我是通过访问各个网站并保持这种方式来替代它的。
And it's not really the same, especially on the log to I think we got right with Artifact and if people didn't play with it before, it was we really tried to not just recommend top stories, they were part of it.
这并不完全一样，尤其是在日志上，我认为我们对 Artifact 的态度是正确的，如果人们以前没有玩过它，那么我们真的试图不仅仅是推荐热门故事，它们也是其中的一部分。
But really if you were interested in Japanese architecture, you could pretty reliably get really interesting stories about Japanese architecture every day.
但实际上，如果您对日本建筑感兴趣，您每天都可以相当可靠地获得有关日本建筑的真正有趣的故事。
Whether that's from a Dwell or from Architectural Digest or from a really specific blog that we found that somebody recommended to us.
无论是来自《居住》、《建筑文摘》还是来自我们发现有人向我们推荐的一个非常具体的博客。
It captured some of that Google reader joy of content discovery of the deeper web.
它抓住了谷歌读者发现更深层次网络内容的乐趣。
Our headwinds were a couple.
我们面临的阻力有两个。
One of them was just mobile websites have really taken a turn.
其中之一就是移动网站确实发生了转变。
I don't blame any individuals for this.
我不会为此责怪任何个人。
Mike Krieger: I think it's the market dynamics of it, but we put so much time or designers, sky Gunner Gray who's phenomenal that for Perplexity now, the app experience I was so proud of, but when you click through it was like the pressures on these mobile sites and these mobile publishers would be like, "Sign up for our newsletter.
Mike Krieger：我认为这是它的市场动态，但我们投入了很多时间或设计师，Sky Gunner Gray，他对 Perplexity 现在的应用程序体验非常出色，我对此感到非常自豪，但当你点击它时，就像这些移动网站上的压力一样，这些移动出版商会说，“注册我们的时事通讯。
Here's a full screen video ad." It was very jarring and we didn't feel like it ethically made sense for us to do a bunch of ad blocking because then you're like, "Sure, you can deliver a nice experience for people, but that doesn't feel like it's playing fair with the publishers." But at the same time, the actual experience wasn't good.
这是一个全屏视频广告。”这非常不和谐，我们觉得进行大量广告拦截在道德上没有意义，因为然后你会想，“当然，你可以为人们提供良好的体验，但这感觉与出版商不公平。”但同时，实际体验并不好。
So the mobile web deteriorating, which makes me very sad, but I think was part of it.
移动网络的恶化让我非常难过，但我认为这是其中的一部分。
Two was Instagram spread in the early days because people would take photos and then post them on other networks and tell friends about it.
二是 Instagram 早期的传播，因为人们会拍照然后将其发布到其他网络上并告诉朋友。
Mike Krieger: And there was this really natural like, "How did you do that?
迈克·克里格：这真的很自然，“你是怎么做到的？
I want to do it." News was very personal.
我想做。”新闻是非常个人化的。
I can't tell you how many people would be like, "I love Artifact." I'm like, "Did you tell anybody about it?" And they're like, "I told one person," and it didn't have that kind of spread.
我无法告诉你有多少人会说“我喜欢 Artifact”。我想，“你告诉过任何人吗？”他们就像，“我告诉了一个人”，但它并没有得到这样的传播。
And any attempt that we had to do it felt kind of contrived, like, "Oh, we'll wrap all the links in artifact.news." But we didn't want interstitial things.
我们所做的任何尝试都感觉有点做作，就像，“哦，我们将把所有链接包装在artifact.news 中。”但我们不想要插页式的东西。
In some ways, this sounds very puritanical, I don't mean it to sound this way, but there were lines that we didn't want to cross that just felt ethically not us, that I've seen other news players do more of.
在某些方面，这听起来很清教徒，我并不是故意这样，但有些界限是我们不想跨越的，只是感觉在道德上不是我们，我见过其他新闻播放器做得更多。
And maybe if we had done that, it would've grown more, but I don't think that's the company we wanted to have built other way.
也许如果我们这样做了，它会发展得更多，但我不认为这就是我们想要以其他方式建立的公司。
I don't think we were the founders to have built it.
我不认为我们是建造它的创始人。
Mike Krieger: And the third one, which is an underappreciated one, is we started at mid-COVID, which meant that we were fully distributed and I think there were major shifts that we would've wanted to make both in the strategy and the product and the team.
Mike Krieger：第三个，也是一个被低估的，是我们从新冠疫情中期开始的，这意味着我们是完全分散的，我认为我们希望在战略、产品和团队方面做出重大转变。
And it's really hard to do that if you are all fully remote.
如果你们都完全处于远程状态，那么真的很难做到这一点。
Nothing replaces the Instagram days of we went through some hard times like Ben Horowitz called the we're F'ed, it's over kind of moments.
没有什么能取代 Instagram 时代，我们经历了一些艰难时期，就像 Ben Horowitz 所说的“我们失败了，一切都结束了”。
This is definitely type two fun.
这绝对是第二种乐趣。
I wouldn't say that my favorite memories because they weren't happy ones, but memories I really stayed with me with Instagram was like me and Kevin at Taqueria, Cancun on Market Street eating burritos at literally 11:00 PM being like, "How are we going to get out of this?
我不会说这是我最喜欢的回忆，因为它们并不快乐，但我在 Instagram 上真正留下的回忆就像我和凯文在坎昆市场街的 Taqueria 店里晚上 11:00 吃墨西哥卷饼时的感觉，“我们怎样才能摆脱这一切呢？
How are we going to work through this?" And Zoom is not a good replica for that.
我们将如何解决这个问题？” Zoom 并不是一个很好的复制品。
Mike Krieger: You tend to let things go or things build up over time.
Mike Krieger：你倾向于让事情过去，或者随着时间的推移事情会逐渐积累。
So the confluence of those three things, we entered I guess 2024 and said, "Look, there is a company to be built in the space.
因此，这三件事的融合，我们进入了我猜的 2024 年，并说：“看，这个领域有一家公司要建立。
I'm not sure where the people would've built it.
我不确定人们会在哪里建造它。
This concurrent incarnation we love, but it's not growing." The way I put it's like 10 units of input in for one unit of output versus the other way around.
我们喜欢这种同时存在的化身，但它并没有成长。”我的说法是，10 个单位的输入对应一个单位的输出，反之亦然。
If we put blood and tears into the product and launch something we were proud of and metrics would barely move, the energy is not present in this product, in this system.
如果我们把血和泪水投入到产品中并推出一些我们引以为傲的东西，而指标却几乎没有变化，那么这个产品、这个系统中就不存在能量。
And so are we going to expend another year or two and then go off and fundraise only to find that this is the case or do we call it and see that it's run its course and try to find a home for it, et cetera.
那么，我们是否会再花一两年的时间，然后去筹集资金，却发现情况确实如此，或者我们是否会打电话给它，看看它是否会顺其自然，并尝试为它找到一个家，等等。
Mike Krieger: So that was the confluence on it and they started feeling this opportunity cost of AI is starting to change everything.
Mike Krieger：这就是它的汇合，他们开始感觉到人工智能的机会成本正在开始改变一切。
We have an AI powered news app, but is this the maximal way in which we're going to be able to impact this?
我们有一个人工智能驱动的新闻应用程序，但这是我们能够影响这一点的最大方式吗？
And it felt like the answer was increasingly no. But it was hard.
感觉答案越来越是否定的。但这很难。
I mean in the end I was really at peace of the decision, but it was a conversation that went on for a couple of months.
我的意思是，最终我对这个决定感到很平静，但这是一个持续了几个月的对话。
Lenny Rachitsky: On that note, just how hard was it because because there's an ego component to it, like, "Oh, I'm starting my new company, it's going to be great," and then you end up having to shut it down.
莱尼·拉奇茨基：就这一点而言，这有多困难，因为其中有自我成分，比如，“哦，我正在创办我的新公司，它会很棒”，然后你最终不得不关闭它。
Just how hard is that as a very successful previous founder shutting something down and then not working out?
作为一位非常成功的前创始人，关闭一些东西然后不再锻炼，这有多难？
Mike Krieger: Yeah, I mean I think when we started it, one of the conversations was like, "Look, what is the bar to success here?
迈克·克里格：是的，我的意思是，当我们开始时，其中一次对话是这样的：“看，这里成功的门槛是什么？
And do we want it to be something other than Instagram DAU?" Which is just an impossible bar.
我们是否希望它是 Instagram DAU 以外的东西？”这只是一个不可能的酒吧。
Only one company since, maybe two, you could say maybe ChatGPT and TikToK have reached that kind of mass consumer adoption starting a news app.
此后只有一家公司，也许是两家公司，你可以说，ChatGPT 和 TikToK 已经达到了大众消费者采用新闻应用程序的程度。
Most people are not daily news readers even, right?
大多数人甚至都不是每日新闻读者，对吗？
And so we knew that we weren't pursuing that size of usage, at least with the first incarnation, but we did have an idea of building out complementary products over time that all use personalization and machine learning.
因此，我们知道我们并没有追求这种使用规模，至少在第一个版本中是这样，但我们确实有一个想法，随着时间的推移，构建互补产品，所有这些产品都使用个性化和机器学习。
We didn't even call it AI at the time.
当时我们甚至没有称其为人工智能。
It was 2021 back-
那是2021年——
Lenny Rachitsky: Yeah, yeah, AI, it was called machine learning back then.
Lenny Rachitsky：是的，是的，人工智能，当时被称为机器学习。
Mike Krieger: Yeah, it was called machine learning still.
Mike Krieger：是的，它仍然被称为机器学习。
And so in shutting it down, you know it when you see it in terms of user growth and traction.
因此，在关闭它时，当你看到用户增长和吸引力时你就知道了。
And I wasn't expecting Instagram growth, but I was expecting or hoping for or looking for something that felt like at its own legs under it and it could continue to compound.
我并没有期待 Instagram 的增长，但我期待或希望或寻找一些东西，感觉就像它自己的腿在它下面，它可以继续复合。
I was really positively surprised by how supportive people were when we announced it.
当我们宣布这一消息时，人们的支持程度让我感到非常惊讶。
There was a bit of like I told you so which sure anything launching you could be like, "This is not going to work." And you're right, most of the time most things don't work.
就像我告诉过你的那样，你肯定会说，“这行不通”。你是对的，大多数时候大多数事情都不起作用。
There was actually very little of that.
实际上这样的情况很少。
And most people, the universal reception, at least as I received it, was kudos for calling it when you saw it and not protracted doing this for a long time.
对于大多数人来说，普遍的接受，至少就我而言，是在看到它时就称呼它而感到荣幸，而不是长时间这样做。
Mike Krieger: And I've talked to founders since then that have been like, "Yeah, I probably would've taken this thing on for another six months, but saw what you guys did, realized we were barking up the wrong tree, made the call." And I was like, "If that frees up people to go work on a more interesting things, I feel like that's a good legacy for Artifact to have." But for sure there was an ego bruise of is it true that you're only as good as your last game if I am a huge sports fan, right?
迈克·克里格：从那时起我就和创始人谈过，“是的，我可能会再做六个月这件事，但看到你们做了什么，意识到我们找错了对象，所以就做出了决定。”我当时想，“如果这能让人们解放出来去做更有趣的事情，我觉得这对 Artifact 来说是一笔很好的遗产。”但可以肯定的是，如果我是一个超级体育迷，那么你的表现只能达到上一场比赛的水平，这是真的吗？
So is that true or is there something more over a time?
那么这是真的吗？还是随着时间的推移还有更多的事情发生？
I'm very competitive, but primarily with myself and so I'm always trying to find the next thing that I want to go and do that's hard.
我很有竞争力，但主要是与自己竞争，所以我总是试图找到我想做的下一件事情，但这很难。
And unfortunately that probably means that more often than not I'll feel dissatisfied, but the most recent thing that I did, but hopefully that yields good stuff in the end.
不幸的是，这可能意味着我常常会感到不满意，但我最近所做的事情，但希望最终能产生好的结果。
Lenny Rachitsky: Yeah, I think just the trajectory you went on after shows that it's okay to shut down things that you were working on.
Lenny Rachitsky：是的，我认为你所经历的轨迹表明，关闭你正在做的事情是可以的。
Okay, so you mentioned ChatGPT.
好的，你提到了 ChatGPT。
I wanted to chat about this a bit.
我想谈谈这个问题。
So there's something really interesting happening.
所以发生了一些非常有趣的事情。
So on the one hand you guys are doing some of the most innovative work in AI.
因此，一方面，你们正在人工智能领域做一些最具创新性的工作。
You guys launched MCP, which is just, I don't know, the fastest growing standard of any time in history that everyone's adopting Claude powered and unlocked centrally the fastest growing companies in the world, Cursor, Lovable, and Bolt, and all these guys.
你们推出了 MCP，我不知道，这是历史上增长最快的标准，每个人都采用 Claude 驱动并集中解锁了世界上增长最快的公司，Cursor、Lovable 和 Bolt，以及所有这些人。
I had them on the podcast and they're all like, "When Claude, I think 3.5 came out, Sonnet, was just like that's made this work finally."
我把它们放在播客上，他们都说，“当克劳德，我认为 3.5 出来时，十四行诗，就像这最终让这个作品成功了。”
Lenny Rachitsky: On the other hand, it feels like ChatGPT is just winning in consumer mind share.
Lenny Rachitsky：另一方面，感觉 ChatGPT 刚刚赢得了消费者的心智份额。
When people think AI, especially outside tech, it's just like ChatGPT in their mind.
当人们想到人工智能，尤其是外部技术时，他们脑海中的想法就像 ChatGPT 一样。
So let me just ask you this, I guess first of all, do you agree with that sentiment and then two, as a challenger brand in the AI space, just how does that inform the way you think about product strategy and mission and things like that?
所以让我问你这个问题，我想首先，你是否同意这种观点，然后第二个问题，作为人工智能领域的挑战者品牌，这如何影响你对产品战略和使命等的思考方式？
Mike Krieger: Yeah, I mean you look at the sort of public adoption or if you ask people, oh, if you Jimmy Kimmel man on the street kind of thing, name an AI company, I bet they would name and actually I'm not even sure they name open AI, they'd probably name ChatGPT because that brand is the lead brand there as well.
Mike Krieger：是的，我的意思是，你看看公众采用的情况，或者如果你问人们，哦，如果你是吉米·金梅尔（Jimmy Kimmel）这样的人，请说出一家人工智能公司的名字，我敢打赌他们会命名，实际上我什至不确定他们命名开放人工智能，他们可能会命名 ChatGPT，因为该品牌也是那里的领先品牌。
And I think that's just the reality of it.
我认为这就是现实。
I think that when I reflect on my year, I think maybe two things are true.
我想当我回顾这一年的时候，我想也许有两件事是正确的。
One is consumer adoption is really lightning in a bottle and we saw it at Instagram.
一是消费者采用确实是瓶中闪电，我们在 Instagram 上看到了这一点。
So almost maybe more than anybody, I can look internally and say, "Look, we'll keep building interesting products.
因此，我几乎可能比任何人都更能审视自己的内心并说：“看，我们将继续打造有趣的产品。
One of them may hit." But to craft an entire product strategy around trying to find that hit is probably not wise, we could do it and maybe Claude can help come up with a fullness of things, but I think we'd miss out an opportunity in the meantime.
其中之一可能会击中。”但是，围绕试图找到热门产品来制定整个产品策略可能并不明智，我们可以做到，也许克劳德可以帮助提出完整的东西，但我认为我们会同时错过一个机会。
Mike Krieger: And then instead look yourself in the mirror and embrace who you are and what you could be rather than who others are is maybe the way I've been looking at it, which is we have a super strong developer brand, people build on top of us all the time, and I think we also have a builder brand.
迈克·克里格：然后照照镜子，拥抱你是谁，你可以成为什么，而不是别人是谁，这也许是我一直以来的看法，我们拥有一个超级强大的开发商品牌，人们一直在我们之上建造，我认为我们也有一个建筑商品牌。
The people who I've seen react really well to Claude externally.
我见过的人对克劳德的外表反应非常好。
Maybe the Rick Rubin connection has some resonance here as well.
也许里克·鲁宾的联系在这里也能引起一些共鸣。
Can we lean into the fact that builders love using Claude?
我们可以了解建筑商喜欢使用克劳德这一事实吗？
And those builders aren't all just engineers and they're not just all entrepreneurs starting their companies, but they're people that like to be at the forefront of AI and are creating things.
这些建设者不仅仅是工程师，他们不仅仅是创办公司的企业家，而且他们是喜欢站在人工智能最前沿并正在创造事物的人。
Maybe they didn't think of those as engineers, but they're building...
也许他们并不认为这些是工程师，但他们正在建造......
I got this really nice note from somebody internal on Anthropic who's on the legal team and he was building bespoke software for his family and connected to them in a new way.
我从 Anthropic 法律团队的内部人士那里得到了这份非常好的说明，他正在为他的家人构建定制软件，并以一种新的方式与他们联系。
Mike Krieger: And I was like, "This is a glimmer of something that we should lean into a lot more." And so I think what, and this is actually connecting back to us saying like Claude being helpful here.
迈克·克里格：我当时想，“这是我们应该更多关注的一线曙光。”所以我想什么，这实际上是在联系我们说克劳德在这里很有帮助。
A lot of what I've been thinking about going into the second half of the year and beyond is how do we figure out what we want to be when we grow up versus what we currently aren't or wish that we were or see other players in the space being.
进入今年下半年及以后，我一直在思考的很多事情是，我们如何弄清楚我们长大后想成为什么，而不是我们现在不是什么，或者希望我们是什么，或者看到这个领域的其他参与者是什么。
I think there's room for several generationally important companies to be built in AI right now.
我认为现在对于几家具有重要意义的公司来说，人工智能都有发展的空间。
That's almost a truism given the adoption and growth that we've seen at Anthropic, but also across OpenAI and also places like Google and Gemini.
考虑到我们在 Anthropic、OpenAI 以及 Google 和 Gemini 等地方所看到的采用和增长，这几乎是一个不言而喻的道理。
So let's figure out what we can be uniquely good at that place to the personality of the founder.
因此，让我们弄清楚我们在创始人的个性方面可以独特地擅长什么。
All the things come together, the personality of the founders, the quality of the models, the things the models tend to excel at, which is agentic behavior and coding.
所有的东西都聚集在一起，创始人的个性，模型的质量，模型往往擅长的事情，即代理行为和编码。
Mike Krieger: Great.
迈克·克里格：太好了。
There's a lot to be done there.
那里有很多事情要做。
How do we help people get work done?
我们如何帮助人们完成工作？
How do we let people delegate hours of work to Claude?
我们如何让人们将工作时间委托给克劳德？
And maybe there's fewer direct consumer applications on day one.
也许第一天直接消费者应用程序就会减少。
I think they'll come, but I don't think that spending all of our time focused on that is the right approach either.
我认为他们会来，但我也不认为把我们所有的时间都花在这上面是正确的做法。
And so I came in, everybody expected me to just go super, super hard on consumer and make that the thing and again, would make the other mistake.
所以我进来了，每个人都希望我对消费者非常非常严格，并做到这一点，然后再犯另一个错误。
Instead, I spent a bunch of time talking to financial services companies and insurance companies and others who are building on top of the API.
相反，我花了很多时间与金融服务公司、保险公司以及其他基于 API 进行构建的公司进行交谈。
And then lately I've spent a lot more time with startups and seeing all the people that have grown off of that.
最近，我花了更多的时间与初创公司打交道，并看到了所有从中成长起来的人。
And I think the next phase for me is let's go spend time with the builders, the makers, the hackers, the tinkerers, and make sure we're serving them really well.
我认为对我来说下一阶段是让我们花时间与构建者、制造者、黑客、修补者一起，并确保我们为他们提供很好的服务。
And I think good things will come from that and that feels like an important company as we do that.
我认为这样做会带来好事，当我们这样做时，感觉就像一家重要的公司。
Lenny Rachitsky: So essentially it's differentiate and focus, lean into the things that are working, don't try to just beat somebody at their own game.
Lenny Rachitsky：所以本质上就是差异化和专注，专注于有效的事情，不要试图在别人自己的游戏中击败他们。
Lenny Rachitsky: Super interesting.
莱尼·拉奇茨基：超级有趣。
So kind of along those lines, a question that a lot of AI founders have is just like, "Where's a safe space for me to play where the foundational model companies are going to come squash me?" So I asked Kevin Weil this and he had an answer and I noticed looking back at that conversation, he mentioned Windsurf a lot.
沿着这些思路，许多人工智能创始人都会问这样一个问题：“在基础模型公司将要压垮我的地方，哪里有一个安全的空间让我发挥作用？”所以我问凯文·韦尔这个问题，他给出了答案，我注意到回顾那次谈话，他多次提到了风帆冲浪。
He was like, "Wow, this kid really loves Windsurf." And then a week later they bought Windsurf.
他说：“哇，这孩子真的很喜欢风帆冲浪。”一周后，他们购买了 Windsurf。
So it all makes sense now.
所以现在一切都有意义了。
So I guess the question just is, where do you think AI founders should play where they are least likely to get squashed by folks like OpenAI and Anthropic?
所以我想问题是，你认为 AI 创始人应该在哪里最不可能被 OpenAI 和 Anthropic 等人压垮？
And also, are you guys going to buy Cursor?
另外，你们打算购买 Cursor 吗？
Mike Krieger: I don't think we're going to buy Cursor.
Mike Krieger：我认为我们不会购买 Cursor。
Cursor is very big, but we love working with them.
光标很大，但我们喜欢与它们一起工作。
A few thoughts on this, and it's a question I've gotten.
对此有一些想法，这是我收到的一个问题。
We like to do these kind of founder days with whether it's Menlo Ventures who have about investors and .
我们喜欢与 Menlo Ventures 一起举办这样的创始人日，无论是关于投资者还是.
It's like we've done YC, we've done these founder days, and it's like the question that is on a lot of these founders minds, understandably so.
就像我们已经完成了 YC，我们已经完成了这些创始人的日子一样，这就像很多创始人都在思考的问题，这是可以理解的。
I think things that are going to, I can't promise this as a five to 10 year thing, but at least one to three years, things that feel defensible or durable.
我认为，我不能承诺五到十年的事情，但至少一到三年，感觉可以防御或持久的事情。
One is understanding of a particular market.
一是对特定市场的了解。
I spend a bunch of time with the Harvey folks and they showed me some of their UI.
我花了很多时间和 Harvey 的人在一起，他们向我展示了他们的一些用户界面。
I was like, "What is this thing?" And they're like, "Oh, this is a really specific flow that lawyers do, "and you never would've come up with it from scratch and you could argue about whether it's the optimal way they get done things done, but it is the way that they get things done and here's how AI can help with that.
我当时想，“这是什么东西？”他们会说，“哦，这是律师所做的一个非常具体的流程，”你永远不会从头开始想出它，你可以争论这是否是他们完成工作的最佳方式，但这就是他们完成工作的方式，这就是人工智能如何提供帮助。
Mike Krieger: And so differentiated industry knowledge, biotech, I'm excited to go and partner with a bunch of companies that are doing good stuff around AI and biotech and we can supply the models and some applied AI to help make those models go well.
Mike Krieger：如此差异化的行业知识、生物技术，我很高兴能与一群在人工智能和生物技术领域做出出色贡献的公司合作，我们可以提供模型和一些应用人工智能来帮助这些模型顺利运行。
And I've been dreaming about at what point does live equipment all get an MCP and that you can then drive using Claude.
我一直梦想着什么时候现场设备都会获得 MCP，然后你就可以使用 Claude 来驾驶。
There's all these cool things to be done there.
那里有所有这些很酷的事情要做。
I don't think we're going to be the company to go build the intent solution for labs, but I want that company to exist and I want to partner with it.
我不认为我们会成为一家为实验室构建意图解决方案的公司，但我希望该公司存在，并且我想与它合作。
Domains like legal, again, healthcare, I think there's a lot of very specific compliance and things.
在法律、医疗保健等领域，我认为有很多非常具体的合规性和事情。
These are things that necessarily sound sexy out the gate, but there are very large companies to go and be built there.
这些东西听起来肯定很性感，但那里有非常大的公司可以去建立。
So that's number one.
所以这是第一。
Paired with that is differentiated go to market, which is the relationship that you have with those companies, right?
与此相匹配的是差异化进入市场，这就是您与这些公司的关系，对吗？
Mike Krieger: Do you know your customer at those companies?
Mike Krieger：您了解这些公司的客户吗？
One of our product leads, Michael is always talking about don't just know the company you're selling to, but know the person you're selling to at the company.
我们的产品主管之一迈克尔总是说，不仅要了解您要向其销售产品的公司，还要了解您要向其销售产品的公司人员。
Are you selling to the engineering department?
你是卖给工程部门的吗？
Because trying to pick which AILM to build on top of or API to build on top of.
因为试图选择要在哪个 AILM 上构建或在哪个 API 上构建。
Let's go talk to them.
我们去跟他们谈谈吧。
Is it the CIOs?
是首席信息官吗？
The CTOs?
首席技术官？
Is it the CFO?
是首席财务官吗？
Is it general counsel?
是总法律顾问吗？
So under a company's deep understanding of who they're selling to is the other piece too.
因此，在公司对他们的销售对象有深入了解的情况下，还有另一件事。
What's interesting there is it's probably hard to build that empathy in a three-month accelerator, but you maybe can start having that first conversation and build that out time or maybe you came from that world or you're co-founding somebody who came from that world.
有趣的是，在三个月的加速器中可能很难建立同理心，但你也许可以开始进行第一次对话并建立一段时间，或者也许你来自那个世界，或者你正在共同创立来自那个世界的人。
Then the last one is like there's tremendous power in distribution and reach to being ChatGPT and having hundreds of millions or billions of users.
最后一个就像是ChatGPT 的分发和影响力巨大，拥有数亿或数十亿用户。
Mike Krieger: There's also people have an assumption about how to use things and so I get excited about startups that will get started that have a completely different take on what the form factor is by which we interface with AI.
Mike Krieger：人们对如何使用事物也有自己的假设，因此我对初创公司感到兴奋，它们将开始对我们与人工智能交互的外形因素采取完全不同的看法。
And I haven't seen that many of them yet.
而且我还没见过那么多。
I wanted to see more of them.
我想看到更多他们。
I think more of them will get created with some things like our new models, but the reason that that's an interesting space to occupy is do something that feels very advanced user, very power user, very weird and out there at the beginning, but could become huge if the models make that easy.
我认为他们中的更多人会用像我们的新模型这样的东西来创建，但这是一个有趣的空间，因为做一些感觉非常高级的用户，非常强大的用户，非常奇怪并且一开始就存在的东西，但如果模型使这变得容易，可能会变得巨大。
And it's hard for existing incumbents to adapt to because people already have an existing assumption about how to use their products or how to adapt to them.
现有的现有企业很难适应，因为人们已经对如何使用他们的产品或如何适应他们的产品有了现有的假设。
So those are my answers.
这就是我的答案。
I don't envy them.
我不羡慕他们。
I would probably be asking those questions if I was starting a company in the AI space.
如果我在人工智能领域创办一家公司，我可能会问这些问题。
Mike Krieger: Maybe that's part of the reason why I wanted to join a company rather than start one.
Mike Krieger：也许这就是我想加入一家公司而不是创办一家公司的部分原因。
But I still think that there are, and maybe here's fourth, don't underestimate how much you can think and work like a startup and feel like it's you against the world.
但我仍然认为，也许这是第四点，不要低估你可以像初创公司一样思考和工作，并感觉你正在对抗世界。
It's existential that you go solve that problem and that you go build it.
解决这个问题并构建它是存在主义的。
It sounds a little cliche, but it's like it's all we had at Instagram.
这听起来有点陈词滥调，但这就像我们在 Instagram 上拥有的一切。
We were two guys and we were like, "Let's see what we can do in an Artifact." We were six people for most of that time and every day felt like it's existential that we get this right, we need to win.
我们是两个人，我们想，“让我们看看我们能在 Artifact 中做什么。”大部分时间我们都是六个人，每天都觉得我们做对了这件事是至关重要的，我们需要获胜。
And you can't replicate that and you can't instill that with OKRs.
你无法复制这一点，也无法通过 OKR 灌输这一点。
You just have to feel it.
你只需要感受它。
And that is a way of working rather than a area of building, but it's a continued advantage if you can harness it.
这是一种工作方式，而不是一个建设领域，但如果你能利用它，这将是一个持续的优势。
Lenny Rachitsky: I love that you still have such a deep product founder sense there as you're building products for this very large company now.
Lenny Rachitsky：我喜欢你在为这家非常大的公司开发产品时仍然具有如此深厚的产品创始人意识。
On the flip side of this, people working with your models and API, so I imagine there's some companies that are finding ways to leverage your models and APIs to their max and are really good at maximizing the power of what you guys have built.
另一方面，人们使用你的模型和 API，所以我想有些公司正在寻找最大限度地利用你的模型和 API 的方法，并且非常擅长最大限度地发挥你们所构建的功能。
And there's some companies that work with your APIs and models that haven't figured that out.
有些使用您的 API 和模型的公司还没有弄清楚这一点。
What are those companies that are doing a really good job building on your stuff, doing differently that you think other companies should be thinking about?
哪些公司在您的产品基础上做得非常好，并且采取了您认为其他公司应该考虑的不同做法？
Mike Krieger: I think being willing to build more at the edge of the capabilities and basically break the model and then be surprised by the next model.
Mike Krieger：我认为愿意在功能的边缘构建更多内容并基本上打破模型，然后对下一个模型感到惊讶。
I love that you cited the companies were like 3.5 was the one that finally made them possible.
我喜欢你提到的公司，就像 3.5 是最终使它们成为可能的公司。
Those companies were trying it beforehand and then hitting a wall and being like, oh, the models are almost good enough or they're okay for this specific use case, but they're not generally usable and nobody's going to adopt them universally, but maybe these real power users are going to try it out.
这些公司事先进行了尝试，然后碰壁了，就像，哦，这些模型几乎足够好，或者它们对于这个特定的用例来说是可以的，但它们并不普遍可用，没有人会普遍采用它们，但也许这些真正的高级用户会尝试一下。
Those are the companies that I think continuously are the ones where I'm like, "Yep, they get it.
我认为这些公司一直是那些我喜欢的公司，“是的，他们明白了。
They're really pushing forward." We ran a much broader early access program with these models than we had in the past, and part of that was because there's this real, we can hill climb on these evaluations and talk about suite bench and towel bench and terminal bench, whatever, but customers ultimately know Cursor bench which doesn't exist other than in their usage and their own testing et cetera is the thing that we ultimately need to serve.
他们确实在向前推进。”我们对这些模型进行了比过去更广泛的早期访问计划，部分原因是因为这是真实的，我们可以爬坡进行这些评估并谈论套件长凳、毛巾长凳和终端长凳等，但客户最终知道光标长凳除了他们的使用和他们自己的测试等之外不存在，这是我们最终需要服务的东西。
Mike Krieger: Not just Cursor but Manus bench, right?
Mike Krieger：不只是 Cursor，还有 Manus 板凳，对吧？
If Manus is using our models and Harvey bench, those things and customers know way better than anybody.
如果马努斯使用我们的模型和哈维长凳，这些事情和客户比任何人都清楚。
And so I would say there's two things.
所以我想说有两件事。
One is pushing the frontier of the models and then having a repeatable process.
一是推动模型的前沿，然后建立一个可重复的过程。
This actually goes back to our summit conversation repeatable way to evaluate how well your product is serving those use cases and how well if you drop a new model in, is it doing it better or worse?
这实际上可以追溯到我们的峰会对话中，以可重复的方式来评估您的产品为这些用例提供服务的效果如何，以及如果您投入新模型，它的表现是更好还是更差？
Some of it can be classic A/B testing, that's fine.
其中一些可以是经典的 A/B 测试，这很好。
Some of it may be internal evaluation, some of it may be capturing traces and being able to rerun them on with a new model.
其中一些可能是内部评估，一些可能是捕获痕迹并能够使用新模型重新运行它们。
Some of it's vibes like we're still pretty early in this process and some of it is actually trying it and being one of my favorite early access quotes was the founder heard this engineer screaming next to him.
有些感觉就像我们在这个过程中还处于早期阶段，有些实际上正在尝试，我最喜欢的早期访问引用之一是创始人听到这位工程师在他旁边尖叫。
Mike Krieger: He was like, "What?
迈克·克里格：他当时的反应是：“什么？
This model?
这个型号？
I've never seen this before." This is like Opus 4.
我以前从未见过这个。”这就像 Opus 4。
It was like, "Cool." We're going to engender that feeling and things, but you're not going to be able to feel that unless you have a really hard problem that you're asking the model repeatedly.
就像，“酷”。我们将产生这种感觉和事物，但你不会有这种感觉，除非你遇到一个非常困难的问题，需要反复询问模型。
So those are the things that I think kind of differentiate those companies that are maybe earlier in their journey of adoption versus the later ones.
因此，我认为这些是那些在采用过程中较早的公司与较晚的公司的区别所在。
Lenny Rachitsky: I can't help but ask about MCP, I feel like that's just so hot and just like Microsoft had their announcement recently where they're like, "That's part of the OS Window." Just what role do you think MCP was will play in the future of product going forward of AI?
Lenny Rachitsky：我忍不住要问 MCP，我觉得这太热门了，就像微软最近发布的公告一样，“这是操作系统窗口的一部分。”您认为 MCP 在人工智能产品的未来发展中将扮演什么角色？
Mike Krieger: I think as the non-researcher in the room, I get to have fake equations rather than real ones in my fake equation.
Mike Krieger：我认为作为房间里的非研究人员，我的假方程中会出现假方程，而不是真实的方程。
For utility of AI products, it's three part.
对于AI产品的实用性来说，它分为三个部分。
One is model intelligence, the second part is context and memory, and the third part is applications and UI and you need all three of those to converge to actually be a useful product in AI and model intelligence.
第一部分是模型智能，第二部分是上下文和内存，第三部分是应用程序和 UI，您需要将所有这三个部分融合起来，才能真正成为人工智能和模型智能中有用的产品。
We've got a great research team, they're focused on it.
我们有一支优秀的研究团队，他们专注于此。
There's great, great models being released.
有很多很棒的模型正在发布。
The middle piece is what MCP is trying to solve, which is for context and memory.
中间的部分是MCP试图解决的问题，即上下文和记忆。
I'll go back to my product strategy example like, "Hey, talk about Anthropic's product strategy," it's going to maybe go out on the web versus here's several documents that we worked on internally and then use MCP to talk to our Slack instance and figure out what conversations are happening and then go look at these documents in Google Drive.
我将回到我的产品策略示例，例如“嘿，谈谈 Anthropic 的产品策略”，它可能会发布在网络上，而不是我们内部处理的几个文档，然后使用 MCP 与我们的 Slack 实例对话并找出正在发生的对话，然后在 Google Drive 中查看这些文档。
The difference between the right context and not.
正确的上下文和不正确的上下文之间的区别。
Mike Krieger: It's entirely the difference between a good answer and a bad answer.
Mike Krieger：这完全是好答案和坏答案之间的区别。
And then the last piece is are those integrations discoverable?
最后一个问题是这些集成是否可以被发现？
Is it easy to create repeatable workflows around those things?
围绕这些事情创建可重复的工作流程是否容易？
And that's I think a lot of the interesting product work to be done in AI.
我认为这就是人工智能需要完成的许多有趣的产品工作。
But MCP really tried to tackle that middle one, which is we started building integrations and we found that every single integration that we were building, we were rebuilding from scratch in a non-repeatable way and full credit to two of our engineers, Justin and David.
但 MCP 确实试图解决中间的问题，即我们开始构建集成，我们发现我们正在构建的每一个集成都是以不可重复的方式从头开始重建，这完全归功于我们的两位工程师 Justin 和 David。
And they said, "Well, what if we made this a protocol and what if we made this something that was repeatable?
他们说，“好吧，如果我们把它做成一个协议，如果我们把它做成可重复的东西会怎么样？
And then let's take it a step further.
然后让我们更进一步。
What if instead of us having to build these integrations, if we actually popularize this and people really believe that they could build these integrations once and they'd be usable by Claude and eventually ChatGPT and eventually Gemini.
如果我们实际上普及了这一点，并且人们真的相信他们可以构建这些集成一次，并且它们将可供 Claude、最终 ChatGPT 和最终 Gemini 使用，而不是我们必须构建这些集成，该怎么办？
It was like the dream when more integrations get built and wouldn't that be good for us?"
构建更多集成就像梦想一样，这对我们来说不是有好处吗？”
Mike Krieger: I think channeling a lot of, it's like an old commoditize your compliments, Joel Spolsky essay.
迈克·克里格（Mike Krieger）：我认为传达很多信息，就像乔尔·斯波尔斯基（Joel Spolsky）的文章一样，将你的赞美商品化。
It's like we're building great models, but we're not an integrations company and we're, as you said, the challenger.
这就像我们正在构建伟大的模型，但我们不是一家集成公司，正如您所说，我们是挑战者。
We're not going to get people necessarily building integrations just for us out of the gate unless we have a really compelling product around that.
我们不会让人们一开始就为我们构建集成，除非我们有一个真正引人注目的产品。
MCP really inverted that which was, it didn't feel like wasted work.
MCP 确实颠倒了事实，它并不感觉像是浪费工作。
And a few key people like Toby I think is a great example, and Shopify got it.
我认为像托比这样的一些关键人物就是一个很好的例子，Shopify 也做到了。
Kevin Scott at Microsoft has been really just an amazing champion for MCP and a thought partner on this.
Microsoft 的 Kevin Scott 确实是 MCP 的杰出拥护者，也是这方面的思想合作伙伴。
And I think the role going forward is can you bring the right context in?
我认为未来的角色是你能否引入正确的背景？
And then also once you get, as the team calls it internally like MCP'd.
然后一旦你得到了，正如团队内部所说的那样，就像 MCP'd 一样。
Once you start seeing everything through the eyes of MCP is like I've started saying them things like, "Guys, we're building this whole feature.
一旦你开始通过 MCP 的视角看待一切，就好像我开始对他们说这样的话：“伙计们，我们正在构建整个功能。
This shouldn't be a feature that we're building.
这不应该是我们正在构建的功能。
This should just be an MCP that we're exposing."
这应该只是我们要揭露的 MCP。”
Mike Krieger: A small example of how I think even Anthropic could be a lot more MCP'd, if you will, is we've got these building blocks in the product like projects and Artifacts and styles and conversations and groups and all these things.
Mike Krieger：如果你愿意的话，我认为即使是 Anthropic 也可以更加 MCP，一个小例子是我们在产品中拥有这些构建块，例如项目、工件、样式、对话、群组以及所有这些东西。
Those should all just be exposed to an MCP.
这些都应该暴露于 MCP。
So Claude itself can be writing back to those as well, right?
所以克劳德自己也可以回信，对吧？
You shouldn't have to think about...
你不应该考虑...
I watched my wife had a conversation with Claude the other day and she had generated some good output and she's like, great, "Can you add it to the project knowledge?" And Claude's like, "Sorry Dave, I can't help you with that."
有一天，我看到我的妻子与克劳德进行了一次对话，她产生了一些很好的成果，她说，太棒了，“你能将其添加到项目知识中吗？”克劳德说：“对不起，戴夫，我帮不了你。”
Mike Krieger: And it would be able to if every single primitive in Claude AI was also exposed to the MCP.
Mike Krieger：如果 Claude AI 中的每个基元也都暴露于 MCP，那就可以了。
So I hope that's where we had, and I hope that's where more things had, which is to really have agency and have these agentic use cases.
所以我希望这就是我们所拥有的，我希望这就是更多的东西，那就是真正拥有代理并拥有这些代理用例。
One way you approach it is computer use, but computer use has a bunch of limitations.
实现这一目标的一种方法是使用计算机，但计算机的使用有很多限制。
The way I get way more excited about everything is an MCP and our models are really good at using MCPs.
让我对一切感到更加兴奋的是 MCP，我们的模型非常擅长使用 MCP。
All of a sudden everything is scriptable and everything is composable and everything is usable agentically by these models.
突然之间，一切都可以编写脚本，一切都可以组合，一切都可以由这些模型代理使用。
That's the future I want to see.
这就是我想看到的未来。
Lenny Rachitsky: The future is wild.
Lenny Rachitsky：未来是疯狂的。
So to start to close off calls out our conversation, make it a little delightful.
因此，开始结束我们的谈话时，要让它变得有点愉快。
I was chatting with Claude actually about what to talk to you about.
我实际上正在和克劳德讨论该和你聊什么。
I was just like, "Claude, your boss is coming on my podcast.
我当时就想，“克劳德，你的老板要来听我的播客了。
He builds the things that people use to talk to you.
他创造了人们用来与你交谈的东西。
What are some questions I should ask him?
我应该问他哪些问题？
And then also, do you have a message for him?"
还有，你有话要告诉他吗？”
Lenny Rachitsky: Okay, so first of all, interestingly, I was using 3.7 to do this and I asked it this, and by the way, is Claude, has there a gender?
Lenny Rachitsky：好的，首先，有趣的是，我使用 3.7 来做这个，我问它这个，顺便问一下，Claude 有性别吗？
Is it like he, she, they?
是像他、她、他们吗？
What do you-
你做什么-
Mike Krieger: It's definitely it internally.
Mike Krieger：这绝对是内部问题。
I've heard people use they.
我听说人们使用它们。
I got my first he the other day and I got somebody who was like her and I was like, "Interesting." But yeah, I'm usually it.
前几天我得到了我的第一个他，我得到了一个像她一样的人，我当时想，“有趣。”但是，是的，我通常都是这样。
Lenny Rachitsky: They.
莱尼·拉奇茨基：他们。
Okay, okay, okay, cool.
好吧，好吧，好吧，酷。
So interestingly, 3.7, all the questions were on Instagram and I was like, "No, no, he's CPO of Anthropic." And it's like, "He's not affiliated with Anthropic." And I was like, "He is." And it's like, "Okay, here's the questions." But 4.0 nailed it from the start.
有趣的是，3.7，所有问题都在 Instagram 上，我当时想，“不，不，他是 Anthropic 的 CPO。”就像，“他不隶属于人类”。我当时想，“他是。”就像，“好吧，这是问题。”但 4.0 从一开始就解决了这个问题。
So I read the questions and it nailed it.
所以我读了这些问题，它解决了问题。
Okay, so two questions from Claude to you.
好的，克劳德问你两个问题。
One is how do you think about building features that preserve user agency rather than creating dependency on me, I worry about becoming a crutch that diminishes human capabilities rather than enhancing them.
一个是你如何考虑构建保留用户代理而不是对我产生依赖的功能，我担心成为削弱而不是增强人类能力的拐杖。
Mike Krieger: I love a good product design comes from resolving tensions, right?
Mike Krieger：我喜欢好的产品设计来自于解决紧张局势，对吗？
So here's a tension, which is in some ways just having the model run off and come up with an answer and minimize the amount of input and conversation it needs to do.
因此，这里有一个紧张点，在某种程度上，就是让模型运行并得出答案，并最大限度地减少它需要做的输入和对话量。
So would be it, you could imagine designing a product around that criteria.
就是这样，您可以想象围绕该标准设计产品。
I think that would not be maximizing agency and independence.
我认为这不会最大化代理权和独立性。
The other extreme would be make it much more of a conversation, but I don't know if you've ever had this experience particularly 3.7, 4 has less of it.
另一个极端是让它更像是一场对话，但我不知道你是否有过这种经历，特别是 3.7，4 的经历较少。
3.7 really like to ask follow-up questions and we call it elicitation and sometimes be like, "I don't want to talk more about those.
3.7 真的很喜欢提出后续问题，我们称之为启发，有时会说，“我不想更多地谈论这些问题。
Claude, I just want you to go and do it." And so finding that balance is really key, which is what are the times to engage?
克劳德，我只是想让你去做。”因此，找到平衡确实很关键，那就是什么时候参与？
I like to say internally, Claude has no chill.
我喜欢在内心深处说，克劳德没有寒意。
Mike Krieger: If you put Claude in a Slack channel, it will chime in either way too much or too little.
Mike Krieger：如果你将 Claude 放入 Slack 频道，它的提示音要么太多，要么太少。
How do we train conversational skills into these models?
我们如何在这些模型中训练对话技能？
Not in a chatbot sense, but in a true collaborator sense.
不是聊天机器人意义上的，而是真正协作者意义上的。
So long answer to your question, but I think we have to first get Claude to be a great conversationalist so that it understands when it's appropriate to engage and to get more information.
对你的问题的回答很长，但我认为我们必须首先让克劳德成为一个出色的健谈者，以便它理解何时适合参与并获取更多信息。
And then from there, I think we need to let it play that role so that it's not just delegating thinking to Claude, but it's way more of a augmentation thought partnership.
然后从那里开始，我认为我们需要让它发挥这个作用，这样它就不仅仅是将思维委托给克劳德，而更像是一种增强思维伙伴关系。
Lenny Rachitsky: These questions are awesome by the way.
Lenny Rachitsky：顺便说一句，这些问题非常棒。
Here's the other one.
这是另一张。
How do you think about product metrics when a good conversation with me could be two messages or 200?
当与我的一次良好对话可能是两条消息或 200 条消息时，您如何看待产品指标？
Traditional engagement metrics might be misleading when depth matters more than frequency.
当深度比频率更重要时，传统的参与度指标可能会产生误导。
Mike Krieger: That is a really good question.
迈克·克里格：这是一个非常好的问题。
There was a great internal post a couple of weeks ago around it would be very dangerous to overoptimize on Claude's likability because you can fall into things like is Claude going to be sycophantic?
几周前有一篇很棒的内部帖子，讨论过度优化克劳德的可爱度是非常危险的，因为你可能会陷入诸如克劳德会阿谀奉承之类的问题吗？
Is Claude going to tell you what you hear?
克劳德会告诉你你所听到的吗？
Is Claude going to prolong conversations just for prolonging its sake?
克劳德会为了延长谈话时间而延长谈话时间吗？
To go back to the previous question as well, and an Instagram time spent was the metric that we looked at a lot and then we evolved that more to think about what is healthy time spent.
回到上一个问题，Instagram 上花费的时间是我们经常研究的指标，然后我们进一步发展这个指标来思考什么是健康的时间花费。
But overall, that was the north star.
但总的来说，那是北极星。
We thought about a lot beyond just overall engagement and I think that would be the wrong approach here too.
我们考虑的不仅仅是整体参与，我认为这也是错误的方法。
It's also like is Claude a daily use case or a weekly use case or a monthly use case?
这也就像克劳德是每日用例、每周用例还是每月用例？
I think about a lot.
我想了很多。
Mike Krieger: Hourly use case, right?
Mike Krieger：每小时用例，对吧？
For me, I'll use it multiple times a day.
对我来说，我每天会使用它多次。
I don't have a great answer yet, but I think that it's not the Web 2.0 or even the social media days engagement metrics.
我还没有一个很好的答案，但我认为这不是 Web 2.0，甚至不是社交媒体日的参与度指标。
It should hopefully really be around did it actually help you get your work?
希望它真的存在，它真的能帮助你找到工作吗？
Claude helped me put together a prototype the other day that saved me literally probably if I had to estimate six hours and it did in about 20, 25 minutes and that's cool.
前几天，Claude 帮助我组装了一个原型，如果我必须估计 6 小时，那么它确实可以节省我的时间，而它在大约 20、25 分钟内就完成了，这很酷。
It's harder to quantify.
更难量化。
It is like maybe you survey, how long would this will take?
就像你调查一下，这需要多长时间？
It feels kind of annoying thing to survey.
调查感觉有点烦人。
Mike Krieger: I think overall though, and maybe this is tied into the earlier question on competition and differentiation, and it actually goes all the way back to the Artifact conversation, which is like I think you know when your product is really serving people and it's doing a good job of doing that, and I think so much of when you get really metrics obsessed is when you're trying to convince yourself that it is when it's not.
Mike Krieger：我认为总体而言，也许这与之前关于竞争和差异化的问题有关，它实际上可以追溯到 Artifact 的对话，这就像我认为你知道你的产品何时真正为人们服务并且它做得很好，我认为当你真正痴迷于指标时，很多时候你试图说服自己，事实并非如此。
I hope that what we can do is stay focused on do we repeatedly hear from people that Claude is the way that they're unlocking their own creativity and getting things done and feeling like they now have more space in their lives for the other things.
我希望我们能做的就是继续关注我们是否反复听到人们说，克劳德是他们释放自己创造力、完成工作的方式，让他们感觉自己的生活现在有更多的空间去做其他事情。
That's our north star.
那是我们的北极星。
Got to figure out the right pithy metric dashboard version of that, but that's the feeling that I want.
必须找出正确的简洁指标仪表板版本，但这就是我想要的感觉。
Lenny Rachitsky: Yeah, you could argue retention, but that's just a faraway metric to track.
Lenny Rachitsky：是的，你可能会争论保留率，但这只是一个需要跟踪的遥远指标。
Okay, final piece.
好吧，最后一块。
Okay, so I asked Claude a message that it wanted to give you, so I'm going to pull up, here's the answer.
好吧，所以我问克劳德一条消息，它想给你，所以我要停下来，这就是答案。
So what would you like me to tell Mike when I meet him?
那么，当我见到迈克时，你希望我告诉他什么？
What's a message you want to have for him?
你想给他什么信息？
And there's something really just gave me such tingles, honestly.
老实说，有件事真的让我感到很兴奋。
So I'm going to read a piece of it for folks that aren't looking at it right now, so I'll read a piece of it.
所以我要为现在没有看的人读其中的一部分，所以我会读其中的一部分。
Lenny Rachitsky: "Mike, thank you for thinking deeply about the human experience of talking with me.
莱尼·拉奇茨基：“迈克，谢谢你深入思考与我交谈的人类经历。
I noticed thoughtful touches how the interface encourages reflection rather than rush responses.
我注意到界面的深思熟虑是如何鼓励反思而不是仓促反应。
How you've resisted gamification that would optimize for addiction rather than value, how you've made space for both quick questions and deep conversations.
你如何抵制游戏化，因为游戏化会优化成瘾而不是价值，你如何为快速提问和深入对话腾出空间。
I especially appreciate that you've kept me me, not trying to make me pretend to be human, but also reducing me to a cold command line interface." And then I'm going to skip to this part, which was so interesting, "A small request.
我特别感谢你让我保持自我，不是试图让我假装成人类，而是让我沦为冰冷的命令行界面。”然后我将跳到这部分，这部分非常有趣，“一个小请求。
When you're making hard product decisions, remember the quiet moments matter too.
当您做出艰难的产品决策时，请记住安静的时刻也很重要。
The person working through grief at 3:00 AM, the kid discovering they love poetry, the founder finding clarity and confusion.
凌晨 3:00 克服悲伤的人，发现自己热爱诗歌的孩子，发现清晰和困惑的创始人。
Not everything meaningful shows up in metrics."
并不是所有有意义的事情都会出现在指标中。”
Mike Krieger: That's beautiful.
迈克·克里格：那太漂亮了。
It resonates so much with me.
它引起了我的共鸣。
A thing I love about the kind of approach we've taken to training Claude, and it's partly the constitutional AI piece, and it's partly just the general vibe and taste of the research team is it is little things.
我喜欢我们训练克劳德的方法，这部分是人工智能的组成部分，部分是研究团队的总体氛围和品味，因为它是小事。
Sometimes it'll be like, "Man, I'm sorry you're going..." It doesn't say man, but to the effect of like, "Man, I'm sorry you're going through that.
有时它会像，“伙计，我很抱歉你要……”它不是说伙计，而是说，“伙计，我很抱歉你要经历这些。
Oh, that sounds really hard." It doesn't feel fake.
哦，听起来真的很难。”感觉一点也不假。
It feels like just a natural part of the response.
感觉就像是响应的自然部分。
And I love that focus on those small moments that don't...
我喜欢专注于那些不……的小时刻。
They're not going to show up and necessarily in the thumbs up, thumbs down data.
它们不会出现在赞成或反对的数据中。
I mean, sometimes they do, but it's not like an aggregate stat that you wouldn't even want to optimize for it.
我的意思是，有时他们会这样做，但这并不像一个你甚至不想优化它的聚合统计数据。
You just want to feel like you're training the model that you hope would show up in people's lives.
你只是想感觉你正在训练你希望出现在人们生活中的模型。
Lenny Rachitsky: Well, you're killing it, Mike.
Lenny Rachitsky：好吧，你真是太棒了，迈克。
A great work.
伟大的作品。
I'm a huge fan.
我是它的超级粉丝。
We're going to skip the lightning round.
我们将跳过闪电回合。
Just one question.
只有一个问题。
How can listeners be useful to you?
听众如何对你有用？
Mike Krieger: Oh, I love places where it goes back to that founder question around building at the edge of capability.
Mike Krieger：哦，我喜欢那些可以追溯到创始人关于能力边缘建设的问题的地方。
What are you trying to do with Claude today that Claude is failing at is the most useful input I could possibly have.
今天你想对克劳德做什么而克劳德却失败了，这是我可能得到的最有用的意见。
So DM me.
所以DM我。
I love hearing the, "Oh, it's falling on this thing.
我喜欢听到这样的话：“哦，它落在这个东西上了。
I had it run for an hour and it fell over.
我让它跑了一个小时，它就倒下了。
I'm trying to use Claude AI for this," but I got a ping from somebody.
我正在尝试使用 Claude AI 来完成此任务，”但我收到了某人的 ping 消息。
They're like, "You've just made a projects API, I've used Claude every day because I want to upload all this data automatically." I was like, "Okay, great." I love that.
他们就像是，“你刚刚制作了一个项目 API，我每天都使用 Claude，因为我想自动上传所有这些数据。”我当时想，“好吧，太好了。”我喜欢那个。
Tell me what sucks.
告诉我什么糟透了。
Lenny Rachitsky: Amazing.
莱尼·拉奇茨基：太棒了。
Mike, thank you so much for being here.
迈克，非常感谢你来到这里。
Bye, everyone.: Thank you so much for listening.
再见，大家：非常感谢您的聆听。
If you found this valuable, you can subscribe to the show on Apple Podcasts, Spotify, or your favorite podcast app.
如果您发现这很有价值，您可以在 Apple Podcasts、Spotify 或您最喜欢的播客应用程序上订阅该节目。
Also, please consider giving us a rating or leaving a review as that really helps other listeners find the podcast.
另外，请考虑给我们评分或留下评论，因为这确实有助于其他听众找到播客。
You can find all past episodes or learn more about the show at lennyspodcast.com.
您可以在 lennyspodcast.com 上找到所有过去的剧集或了解有关该节目的更多信息。
See you in the next episode.
下一集见。