非技术产品经理的Cursor构建指南 | Zevi Arnovitz（Meta）

所以现在，它会利用我们刚才说的所有内容，在Linear中创建一个任务。

So what it's gonna do now is it's gonna use everything we've said and create an issue within linear.

顺便说一下，正在加载的时候，我特别喜欢你描述这种方式，尤其是使用语音模式的时候。

And by the way, as as this is loading, I just love the way the way you describe this, especially doing voice mode.

这就像是你跟工程师交谈时描述一个功能：这是我想要的，然后你提出问题来澄清细节。

It's like exactly how you would talk to an engineer, describing a feature, here's what I want, and then ask you questions, here's the clarification.

是的。

Yeah.

所以一开始，当我跟CTO做这个时，我用的是Chattypiti的语音模式，那简直太疯狂了。

So at first, when I was doing this with the CTO, I would do it with Chattypiti voice mode, and that was crazy.

那感觉就像是在和一个人一起头脑风暴。

That was, like, literally felt like ideating with a person.

它会提出反对意见，问问题，也许有一天，编码工具也能达到这种水平。

It would push back, ask questions, and maybe one day, you know, the coding tools will get there too.

但那正是它最真实的地方——真的就像和我的CTO坐在一起讨论。

But that was exactly it really felt like sitting with my CTO.

很好。

Great.

所以创建了STU 88。

So created STU 88.

如果我们现在打开Linear，应该能看到STU 88在哪儿。

So if we open up linear now, we should be able to see let's see where S T U 88.

在这儿。

There it is.

填空题，带拖拽界面。

Fill in the blank questions with drag and drop interface.

所以它有一个摘要。

So it has a TLDR.

它有当前状态。

It has the current state.

它对代码库做了一点研究，我想是预期结果和一些上下文。

It did a little bit of research on the code base, I think, expected outcomes, some context.

所以，是的，这个基本上已经准备好，等我有兴趣时就可以接手了。

So, yeah, so this is basically ready for me to pick up when I'm interested in building.

所以，假设几天过去了，我完成了当前正在做的项目，就可以来接手这个了。

So now let's say a few days go by, I finish the current project I'm working on, I can pick it up.

当我接手时，我会执行/探索阶段，就是我们之前说的那个。

So when I pick it up, I do slash exploration phase, which is what we said.

然后我不按回车，而是按Tab键，我来给你演示一下。

And then instead of pressing enter, I'll press tab, and I'll show you this.

基本上，探索阶段的作用是接收一个参数。

So basically, exploration phase, what it does is it will take an argument.

这本质上是提示中的一个占位符，让我可以输入一些额外的上下文给AI。

This is basically a placeholder within the prompt, which allows me to enter something that is extra context for the for the AI.

所以我可以在这里写：linear STU 88，这是引用那个任务单。

So I can say here, linear STU 88, which is referencing the ticket.

现在它会去做的是：

And now what it's gonna do is it's gonna go.

它会获取 Linear 工单。

It's gonna fetch the linear ticket.

那这个想法是什么？

And what's the idea?

探索阶段的目标是什么？

What's the goal of the exploration phase?

这是在对想法进行发散思考吗？

This is kinda ideate on the idea?

是这个意思吗？

Is that

没错。

the Exactly.

想法？

Idea?

好的。

Okay.

所以这既是为了让CTO深入理解我们试图解决的问题，也为了了解当前代码库的状态，明确哪些文件需要修改，以及如何从技术上最佳地实现这个功能。

So it's it's both for the CTO to deeply understand the problem that we're trying to solve and also understand the current state of the code base, what files need to be affected, and how is the best way to implement this technically.

通常情况下，现在Claude只是在阅读大量文件，理解代码的基本结构，然后会回来提出一系列澄清性问题，这些问题将决定我们最终如何实现这个功能。

And usually what happens is right now, Claude's just basically reading a bunch of files, understanding the basic structure of the code, and then it's gonna come back with a bunch of clarifying questions as that will decide how we end up implementing this.

所以感觉就像在和你的工程经理对话一样。

So it's like feels like it's talking to your engineering manager.

没错。

Exactly.

没错。

Exactly.

完全正确，这正是我对它的看法。

A 100% this is, this is how I think about it.

你之前说，你曾经用ChatGPT的提示词来模拟一个CTO的角色。

And you said that your CTO, so you used to use ChatGPT prompt, to have a CTO in there.

现在，这个CTO就活在Cursor里面了。

Now the CTO's living inside here in Cursor.

是的

Yeah.

由于工具的发展方式，它们在探索和代码执行方面都变得非常出色，因此现在我已经养成了一种习惯，称之为CTO，但实际上所有功能都集成在一起了。

Because of the way the tools have developed, and they've become so good at both exploration and code execution, so now it's I it's just a kind of a habit that I call it a CTO, but it's basically all in one.

同一个代理会同时进行探索、制定计划并最终执行代码。

The the same agent will will both do the exploration and write the plan and end up executing the code.

明白了。

Got it.

所以这基本上就是Clawd代码。

So it's basically it's Clawd code.

你有没有给它设置过什么提示，让它这样运作？

Is there like a prompt you gave it to act like that?

让它扮演一个合适的CTO？

To act like the right kind of CTO?

是的

Yeah.

在Claude MD中，也就是加载在Claude对话上下文中的系统提示词里，我设置了一些基础内容，比如我们的工作流程。

So within, the Claude MD, which is basically the system prompt that's, loaded within Claude's context in every conversation, I have, some basic stuff like, this is our workflow.

这是我们工作的方式。

This is how we work.

在探索阶段，我希望你能够挑战我的思路，诸如此类的各种内容。

Within exploration phase, I want you to challenge my thinking, all kinds of stuff like that.

这些内容可以加载在ClotMD文件中，是的。

That can be loaded, yeah, within the ClotMD file.

酷。

Cool.

在我们继续之前，还有一个最后的问题，因为随着事情的发生，我正在思考这个问题。

One last question before before we move on here, just because it's I'm thinking about it as this happens.

你生成的线性问题，多久出现一次？

The linear issue that you generated, how often is it?

它真的很好且已经准备好了吗？

Is it actually great and ready?

你多久需要修改一次？

How often do you have to edit it?

它生成的线性工单质量如何？

What's like the quality of the linear ticket that it generates?

因为你知道，很多人可能都在想，这些糟糕的线性工单都是AI生成的。

Because you know, lot of people are like, probably wondering just like all these terrible linear issues are being created by AI.

它们真的好吗？

Are they actually any good?

这完全不一样，因为我是一个人公司，所以很多上下文都包含在这里，我不需要去和其他团队沟通了解。

It's completely different because I'm a company of one, so, a lot of the context is within here, and there's no need for me to, like, talk to other teams and understand.

它非常容易访问。

It's basically very accessible.

而且，我还能轻松看出Claude什么时候理解错了。

And, also, I can easily see when, Claude understood something, wrong.

我不敢说我会在工作中像这样创建线性工单，但如果你在做自己的副项目，这些工单的质量确实不错，而且它能很好地帮你启动工作。

I don't wanna say that I would create linear issues at work like this, but definitely if you're building your own side project, it's, they're they're pretty quality, and also it just kicks off the the when I wanna start working on it.

我不是说它已经准备好可以开发了。

It's not I wouldn't say it's ready to be built.

它已经准备好可以开始探索了。

It's ready to start being explored.

明白了。

Got it.

所以这只是一个想法的开端。

So it's just the beginning of an idea.

实际上，让我们在走完这个流程后再回来讨论一下：如果你在像Meta这样的公司，或者另一个规模更大的公司，而不是你自己的初创公司，你会如何处理这个流程？

Actually, let's come back after we go through this flow how you would approach this if you were at, say, Meta or another maybe a smaller company, how this workflow might work at a larger company that isn't just a your own startup.

是的。

Yeah.

有意思。

Interesting.

我们回头再谈

Let's come back to

那个。

that.

不错。

Cool.

好的。

Alright.

所以这是Claude回来了。

So this is Claude coming back.

我对代码库有全面的理解。

I have a comprehensive understanding of the codebase.

我彻底分析了StudyMate的实时代码库，了解了当前系统、功能需求以及它所识别的关键区域。

I thoroughly analyzed StudyMate live codebase and understand the current system, feature request, and key areas that it's identified.

通常，我会花大量时间仔细研究这些内容，因为这非常重要，但为了当前开发的效率，我们快速过一遍。

Usually, I'd spend a lot of time going over this because this is super, super important, but just for the sake of development right now, we're gonna brush through this.

现在，Claude在浏览并理解了代码库的当前运作方式后，回来了。

Now Claude basically comes back after it's gone through the code base and and understood the way it currently works.

它基本上是在告诉我当前的理解情况。

It's basically telling me what the current understanding is.

所以它在讨论应用目前的结构、数据如何组织、它从功能需求中理解了什么，以及它识别出的关键区域。

So it's talking about how the app is set up at the moment, how the data is structured, what it understood from the feature request, and what is that what it's it has identified as key areas.

然后它向我提出了一些问题。

And then it asked me some questions.

它在询问范围。

So it's asking about the scope.

它在询问数据模型、功能的用户体验和界面、如何验证、如何评分、AI系统提示需要做哪些修改，以及关于应用的各种问题。

It's asking about the data model, the UX UI of the feature, how it should be validated, how it should be graded, what changes need to be happened to the AI system prompt, and all kinds of questions about the app.

我提前准备好了这些问题的答案，因为我觉得我们都不想花时间在这上面。

I've prepared answers to all these questions beforehand because I don't think we all wanna sit through this.

所以我直接粘贴进去，看看Claude会怎么说。

So I'm just gonna paste that in, and we'll see what Claude says.

太棒了。

Awesome.

我太喜欢了。

I love it.

我也很喜欢浏览他提出的那些问题。

And I love just just scanning those questions he was asking.

这些问题非常聪明、细致且重要，而不仅仅是酷而已。

It's like such smart, sophisticated, important questions instead of just cool.

我开始了。

Here I go.

我要把它构建出来。

I'm gonna build it.

是的。

Yeah.

我觉得这介于纯粹随性编码、跟着感觉走，和真正构建严肃应用之间。

And I think this is between, like, just vibe coding and going along with the vibes and really building serious apps.

我花了大量时间来回推敲和深入理解。

I spend a lot, a lot of time going back and forth and understanding.

还有一个非常酷的斜杠命令，我还没展示过，就是‘学习机会’。当有些东西我很难理解时，我会输入‘/学习机会’，然后谈谈我想学什么。

Also, a very cool, slash command that I haven't showed yet is, learning opportunity, which basically, when something is really difficult for me to understand, I'll, do slash learning opportunity and then talk about what I wanna learn.

这基本上是在向Claude传递一个信号：我正在成为一名技术型产品经理。

And this basically primes Claude and says, I am a technical PM in the making.

我具备中等水平的工程知识。

I have mid level engineering knowledge.

我理解架构设计。

I understand the architect architecture.

而且，我希望你能用二八法则，解释我们目前在做什么。

And, basically, I want you to explain what we're currently working on, using the eighty twenty rule.

这是一种绝佳的学习方式。

So this is a great way to learn.

我肯定会采用这种方式。

I would definitely take this.

每当你遇到一些不太完全理解的东西时，我强烈建议你用这个方法来学习。

And every time you kind of see something that you don't fully understand, I would definitely use this, to learn.

很好。

Great.

所以Claude会回复，说明它如何理解当前的数据模型以及将如何实现。

So Claude basically comes back and says how it understands the current data model and how it's gonna implement.

是的。

Yeah.

所以它已经准备好制定计划了。

So it's ready to create the plan.

所以，我现在要做的就是去执行/create plan指令。

So, basically, what I'm gonna do now is I'm gonna go and do slash create plan.

在Todd操作的同时，我快速展示一下这个界面是什么样子。

And while Todd's doing this, I'm gonna show really quick what this looks like.

这些计划模板是我从Twitter上找到的。

So, basically, these plans are from a template that I found on Twitter.

我忘了是谁发布的，但这个模板让我非常有共鸣。

I forgot who it was, but it was just a template that really resonated with me.

它基本上是说，根据我们的交流，创建一个markdown文件作为计划。

And it's basically saying, based on our exchange, create a markdown file that will be the plan.

包含清晰、简洁、精炼的步骤。

Include clear, minimal, concise steps.

跟踪状态。

Track the status.

所以这个计划中包含了每个任务的状态追踪器，Claude在执行过程中会更新这些状态，还会包含一个摘要、我们做出的关键决策以及计划本身。

So this basically has, like, status trackers on each task that Claude updates as it's going through, and it will have a TLDR, some critical decisions that we've made, and the plan itself.

所以Claude已经完成计划的撰写。

So Claude's, finished writing the plan.

我们可以查看并清楚地看到计划内容，它包含了一个摘要。

So we'll be able to look and see exactly what the plan is, so it has a TLDR.

它列出了我们做出的关键决策以及分解后的任务。

It has the critical decisions we've made and the tasks broken down.

这是一个完美的计划，而且这种方式写计划也非常棒，因为很多时候我会使用不同的模型来执行特定任务。

And this is a perfect plan, and it's also a really good way to write this because a lot of times, I'll use different models to execute certain stuff.

所以 Cursor 有一个名为 Composer 的出色模型，速度非常快。

So Cursor has an amazing model called Composer, which is super fast.

因此，对于那些不是特别复杂的事情，我会使用 Composer。

So a lot of things that are not that, complex, I'll use, Composer.

刚刚发布的 Gemini 3 在 UI 方面表现惊人。

Gemini three that just came out is unbelievable at UI.

所以很多时候，我会把计划分成后端和前端，然后让 Gemini 只阅读计划并负责前端部分。

So a lot of times, I'll split the plan into back end and front end, and then I'll have Gemini just read the plan and do the front end.

因此，把这份计划保存为 Markdown 文件非常棒。

So having this as a markdown file is really good.

而且从长远来看，把这个计划集成到应用内部也很有好处，这样以后如果有代理在某个区域写代码时，我可以查看那里已经完成了什么。

And also going forward, it's really good to have within the app so that later on, if an agent is writing code in a certain area, I can see what's already been done there.

所以我们现在要做的，就是执行这个计划。

So what we're gonna do now is we're gonna execute the plan.

所以现在我想我们会用 Cursor 来做这件事，因为 Composer 实在是快得离谱。

So now I think we're gonna do this with Cursor just because, Composer's so freaking fast.

所以我们只需要说执行，然后就可以标记文件。

So what we can do is basically just say execute, and then we can tag the file.

Composer 快得离谱，就这样。

And Composer is ridiculously fast, so that's it.

它已经启动了。

It's off.

它基本上能理解计划内容，然后就会开始编写代码。

It basically understands what the plan is, and it's gonna go ahead and start writing the code.

在它运行的时候，我想问你个问题。

Let me ask you a question while this is happening.

我有很多问题，所以现在是问几个的好时机。

And I have many questions, so this is a good time to ask a few of them.

你说 Lovable 和 Bolt 以及其他类似的应用根本不足以开发真正严肃的应用，你必须转用 Cursor 才能做到。

You said that Lovable and Bolt are and other, you know, other apps in that space are are just not enough to build really serious apps, and you had to move to Cursor to do that.

能再详细说说吗。

What tell us more about that.

就像你当时在那些产品上遇到了哪些限制，为什么你决定转向 Cursor？

Just like how far what what's kind of limitation you ran into with those products and why you switched to Cursor?

我几个月前开始使用 Cursor 和 ClaudeCode，之后就再也没有回头。

I started using Cursor and ClaudeCode a few months ago, and I haven't looked back.

但那时，这些团队正在疯狂地迭代。

But at that time, these teams have been moving like crazy.

所以我不愿说我不信任它们。

So I don't wanna say I wouldn't trust them.

我不知道它们现在的状况如何。

I don't know what the current state is.

但对我来说，主要问题是我觉得 Bolt 对我该如何做事过于武断，而我觉得自己的能力已经达到了可以毕业、掌握更多主动权的阶段。

But for me, it was basically the issue of I felt that that Bolt was being very opinionated on how I should do things, and I felt like my knowledge has gotten to a point where I can graduate and be more in control.

顺便说一句，我认为这些工具之间的主要区别其实在于底层框架。

By the way, I think that the main difference between all these tools is basically the harness.

所以这些模型其实都是相同的模型。

So the models are all the same models.

你知道吧？

You know?

我会在 Cursor 里运行 Claude。

I I'll run Claude within cursor.

我也会在 Claude Code 里运行它，而它也是 Bolt 和 Lovable 底层所使用的同一个模型。

I'll run it within Claude code, and it's also the models that Claude is also the model that is underlying Bolt and Lovable.

但基本上，Bolt 和 Lovable 会在中间添加很多层，帮用户消除各种猜测和艰难的决策。

But basically, Bolt and Lovable will add a bunch of levels in the middle that will take all kind of guesswork and and and hard decisions out for the user.

用户不需要做这些艰难的决策。

The user doesn't make have to make these hard decisions.

所以它也非常容易上手，但代价是你对它的控制力更弱。

So it's also very easy to build, but the flip side of that is that you have less control.

而 Claude Code 就是直接把 Claude 整合进你的代码系统，赋予它全部工具，让它自由发挥。

And basically, Claude code is just taking Claude and shoving it straight in your code system and giving it full tools and to do whatever it wants.

但随之而来的是你需要做出大量决策。

But also with that comes a lot of decisions that you need to make.

所以我不确定你现在是否能用 Bolt 或 Lovable 构建出非常出色的生产级应用，但如果你想要模型最前沿的能力，并且希望自主做出所有决策，那么使用这些工具可能最合适。

So I don't know if if you can't build really amazing production apps using using Bolt or Lovable now, but I think, basically, if you want the most cutting edge abilities of the models and you wanna be able to make all the decisions on your own, it's probably best to be on one of these tools.

我感受到和听到的是，你所做的规划工作，这些正是 Lovable、Bolt 会涵盖的，Replit 也属于这一类吗？

What I'm feeling and hearing is that planning work that you did, that's the stuff that Lovable, Bold, and would you put Replit in that bucket too?

是的。

Yeah.

当然。

For sure.

Lovable、Bolt、Replit、Base 44。

Lovable, Bold, Replit, Base 44.

V zero。

V zero.

对。

Yeah.

V zero，都是一样的。

V zero, all same

类别。

bucket.

所以本质上，它们是在为你做规划。

So essentially, they're kind of they're doing that planning for you.

而且正如你所说，它们非常有主见。

And as you said, they're very opinion ated.

它们试图让事情变得简单。

They they try to make it easy.

所以就是，这里是实现方法。

So it's just like, here's how to do it.

我们不是要说，这是我们觉得对人们最好的方式。

We're not gonna like, here's the way we wanna we think is best for people.

而你所说的，一旦你开始更认真地对待它，或者想往不同方向发展，你就无法改变它们的规划方式，而Cursor允许你这样做。

And what you're saying is once you're trying to get a little more serious about it or have wanna go in a different direction, you don't have the power to change the how they plan, so Cursor lets you do that.

是的。

Yeah.

我不希望听起来像是我在贬低他们。

I don't want this to come out like I'm bad mouthing them.

比如说，Base44。

Base forty four, let's say.

是的。

Yeah.

对。

Yeah.

Base44 在消除构建产品时所有复杂的猜测方面做得非常出色，让你只需跟着感觉走，直接开始构建。

Base forty four does an amazing job at basically taking all the complex guesswork out of building product and just allows you to just, you know, go with the vibes and build.

但它会为你实现 Google 登录功能，也会帮你搭建数据库。

But it will do sign in with Google for you, and it will do a database.

但这样一来，你就不用再决定使用哪个数据库了。

But then you don't have decisions on, what database am I using.

我需要以这种方式还是另一种方式实现 Google 登录？

Do I need sign in with Google this way or the or another way?

它开箱即用。

It would just do it out of the box.

所以这基本上就是那里的权衡。

So so that's basically the trade off there.

太棒了。

Awesome.

致敬Mayor，Baseball的创始人。

Shout out, Mayor, the founder of Baseball.

是的。

Yep.

对。

Yeah.

他太厉害了。

He's amazing.

我特别喜欢你这种方式，比如直接甩出Gemini三号这样的模型来做前端，我超爱你看都没写过一行代码，却能说‘用Gemini做这个，用Claw做那个’，而我呢，就在Cursor上跟你这个CTO一起协作，帮你构建产品，打造真正重要的东西。

I just love how this is, like, the way you're, like, flinging, what's the word, slinging slinging models, like Gemini three for front end, like I love that you have never written any code, and you're just like, use Gemini for this and Claw for this, and I'm just working on cursor, talking to this CTO helping you build stuff and build, like, significant product.

是的。

Yeah.

我的意思是，我们生活在一个最疯狂的时代，世界几乎每周都在变化，似乎没有任何界限。

I mean, we just live in the craziest of times where basically there the world changes once a week, it feels like, and there is just no boundaries.

你可以在普通的 MacBook 或者普通笔记本电脑上使用所有这些工具，我经常有这种时刻。

You can you can use all of these just on your, like, regular MacBook or or regular laptop, and I have these moments.

我称之为时光机时刻，比如这周，我正在用 Claude 准备播客项目。

I call them time machine moments, which is basically, this week, for instance, I was prepping for the podcast using Claude with a project.

我正在将 StudyMate 从希伯来语完全本地化为英语，我只用了两天时间，而这通常需要开发团队数周才能完成。

I was building I was fully localizing, StudyMate from Hebrew to English, which I did in two days, which would probably take a dev team weeks.

我还建了一个个人网站，从没有任何域名、什么都没有，到一个小时内半就上线了。

And I was building a personal site, which went from no domain, no nothing to live on a domain within an hour and a half.

我同时进行着这三件事。

And I was doing all three of these in parallel.

有一刻，这三个代理都在运行，我简直无事可做。

And there was a point where, basically, all three of the agents were running, so I didn't have anything to do.

我只是得让他们自己去思考。

I just had to let them think.

这些就是所谓的时光机时刻，我感觉自己身处未来，只是把头探出时光机而已。

And these are, like, the time machine moments where I feel like I was in the future, and I just stick my head out of the time machine.

当时旁边的人，比如我妻子，我会说：我们生活在未来。

And whoever's next to me, like, at the moment it was my wife, I'll just say, we we live in the future.

她会问：什么？

And she'll be like, what?

然后我会说：不。

And then I'll be like, no.

不。

No.

别担心。

Don't worry about it.

但这一切都太疯狂了，所有这些事情都仅仅隔着一个API的距离。

But it's just basically so crazy that all these things are just, you know, an API away.

你可以使用任何东西。

You can you can use anything.

所以我觉得，现在是一个充满好奇心、乐观和勤奋的绝佳时代。

So I think it's an awesome time to be curious and, optimistic and hardworking.

这是我最喜欢的播客嘉宾类型。

These are my favorite kinds of podcast guests.

那些生活在未来、正在探索这一切的人，然后就像你说的，从火箭飞船里探出头来，说一声：嘿。

People that are living in the future, figuring out all these things, and then are just kinda come back, as you said, poke your head out of the the rocket ship and just like, hey.

这是我发现的东西。

Here's this thing that I figured out.

这就是我们的未来。

Here's where we're going.

是的。

Yeah.

活着的最好时代。

Best time to be alive.

太棒了。

So awesome.

看起来已经完成了。

So it looks like it's finished.

所以现在我们要做的是在本地运行这个应用，看看 Composer 最终构建出了什么。

So now what we're gonna do is we're gonna run the app locally, and we'll be able to see what Composer ended up building.

我们还会看看是否还需要我们这边做些手动审查。

And we're gonna see if anything else is needed on our end to maybe do some manual review.

这样听起来怎么样？

Does that sound good?

听起来很棒。

Sounds great.

我喜欢的是，这就像，我不知道，如果是人类工程师来做，可能得花上几天，甚至一周。

And I love that was, like, I don't know, a few where if it was a human engineer, it'd be, like, days, maybe a week

是的。

Yeah.

对于

For

当然。

For sure.

对于Composer来说，有一点就是它真的非常非常快，能让你保持流畅状态。

A a composer, like, one thing is it's just so, so blazing fast, keeps you in flow.

所以，是的，完整功能几分钟就能搞定。

So, yeah, full features take minutes.

这可能只花了几个美元的AI积分。

And that probably cost, like, a couple bucks in AI credits.

我根本都不看。

I don't even look.

我的意思是，我以前特别舍不得花钱买产品，但现在我基本上把这一切都看作是学费，就是我为学习所支付的东西。

I mean, I used to be so stingy about paying for products, and now I'm just basically I look at it all as tuition, you know, as, like, stuff that I'm paying for learning.

所以我不知道具体花了多少钱，但绝对物有所值。

So I don't know how much it costs, but it's definitely worth it.

这解释了为什么它们是历史上增长最快的产品。

That explains why they're the fastest growing products in history.

完全正确。

A 100%.

你的营销网站奠定了品牌基调，是每一位客户都会接触到的唯一触点。

Your marketing website sets the tone for your brand and is the one touch point that every single one of your customers sees.

在当今时代，如果你仍然难以对网站进行小改动或简单更新，那你一定哪里做错了。

In today's age, if you're still having a hard time making small changes and simple updates to it, you are doing something wrong.

这就是为什么从早期初创公司到财富500强企业，包括DoorDash、Zapier、Perplexity和Eleven Labs等公司，都转向了Framer——这个将你的.com从形式主义转变为增长工具的网站构建器。

That is why so many companies from early stage startups to Fortune five hundreds, including companies like DoorDash, Zapier, Perplexity, and Eleven Labs turned to Framer, the website builder that turns your.com from a formality into a tool for growth.

Framer就像你团队最喜爱的设计工具，支持实时协作、内置完整SEO功能的强大CMS，以及包含集成A/B测试的高级分析功能。

Framer works like your team's favorite design tool and comes with real time collaboration, a robust CMS with everything you need for great SEO, and advanced analytics that includes integrated AB testing.

只需点击一次，无需工程团队协助，你的Framer网站更改就能在几秒内上线。

Changes to your Framer site go live to the web in seconds with a single click and without any help from engineering.

无论你是想上线新网站、测试几个着陆页，还是迁移整个.com网站，Framer都为初创公司、成长型企业和大型企业提供了相应方案，让从想法到上线的过程尽可能简单快捷。

Whether you wanna launch a new site, test a few landing pages, or migrateyourfull.com, Framer has programs for startups, scale ups, and large enterprises to make going from idea to live site as easy and fast as possible.

了解如何从Framer专家那里学习将您的网站转变为增长引擎，或立即免费开始构建：framer.com/lenny。

Learn how to turn your website into a growth engine from a Framer expert or get started building for free today at framer.com/lenny.

那就是 framer.com/leni。

That's framer.com/leni.

适用规则和限制可能适用。

Rules and restrictions may apply.

所以现在我们有了这个功能，基本上是我们构建的，我可以让它做一些更改，因为它是在本地运行的。

So now we have this feature, which basically we built, and I can ask it to make some changes because it's running locally.

一旦准备就绪，我就可以将其推送给用户。

And once it's ready, I'll be able to ship it to users.

所以接下来的阶段，在我完成质量保证并手动测试之后，我会让Claude审查它自己的工作。

So now the next phase, after I've, QA'd it and basically tested it manually, I'll have Claude review its own work.

所以我要做的是重新打开Claude代码。

So what I'll do is I'll, reopen Claude code.

我喜欢这一点，因为这正是这个播客中经常提到的事情：写代码现在变得如此简单。

I love this because this is one of the things that comes up a lot on this podcast is writing code is now so easy.

人们主要的挑战在于审查AI编写的代码。

The main challenge people have is reviewing the code that AI has written.

而你在这里所做的，就是让Claude审查它自己的代码。

And A 100 what you're doing here is you're having Claude review its own code.

是的。

Yeah.

所以这是另一个让我很难发现错误的地方。

So this is another thing where it's very difficult for me to catch mistakes.

我的审查流程已经历了多次迭代，以尽可能完善，尽可能多地发现问题。

So my review process has gone through a bunch of iterations to really be, as good as possible and to catch as many things as possible.

所以我总是先手动进行QA，确保我能发现Claude犯的任何错误。

So I'll always manual q manually QA it first, to make sure if I can see any mistakes that Claude made.

然后我会输入/ review，这会告诉Claude开始审查它自己的代码。

And then what I'll do is basically slash review, and this tells Claude to start reviewing its own code.

但更酷的是，也是我真的很自豪的一点，我通常会进行多次审查，我会同时打开Codex——这是ChatGPT的竞争对手，以及Cursor，让它们各自审查这段代码。

But what's even cooler and something that I'm really proud of is I will usually do multiple reviews, and I'll have Codex, which is ChatGPT's competitor to Cloud Code, as well as Cursor open, and I will have each of them review the code.

然后我用了一个叫做同行评审的斜杠命令，这非常有趣。

And then what what I do is I have a slash command called peer review, which is really interesting.

它的作用是让Claude——通常是我合作的代理——参与进来。

And basically what it does is it's gonna take Claude, which is usually the agent who I'm working with.

为了便于理解，这基本上就像是我和我的开发主管一起工作。

And just to put this in a mental model, this is basically my, my dev lead that I'm working with.

我会使用这个斜杠命令，意思是：你是这个项目的开发主管。

I will take, the the slash command is basically saying, you're the dev lead on this project.

公司里其他团队的主管已经看过你的代码，并审查出了这些问题。

Other team leads within the company have looked at your code and reviewed it and found these issues.

不要盲目接受他们所说的内容。

Don't take what they're what they said at face value.

原因是你比他们拥有更多的上下文信息，而且你主导了这个项目。

The reason is you have more context than them, and you led this project.

你需要要么解释为什么他们发现的问题其实不是真正的问题，要么自己去修复它们。

You need to either, explain why the stuff they found are not real issues and wrong or fix them yourself.

这真的很酷，因为我看待这些事物的方式是去观察这些模型。

And it's really cool, because the way I look at these things is I look at the models.

我会试着把它们想象成人。

I try to imagine them as people.

我真的能告诉你，每一个模型如果作为真实的人会是什么样子？

And I can I can really tell you how each one of these, would be as as a real human?

每一个模型。

Each model.

是的。

Yeah.

每个模型都有如此鲜明的特征。

Each model has such distinct, characteristics.

比如说，Claude会是完美的CTO。

So let's say Claude, she would be the perfect CTO.

她非常善于沟通。

Like, she's very communicative.

她非常聪明。

She's very smart.

她不会只是随波逐流，照着你的话去做。

She she doesn't just go with the flow and do whatever you tell her.

她很有主见，但同时也非常善于协作，我想这就是我总是偏爱Claude的原因，因为我需要不断学习。

She's very opinionated, but also super collaborative, which is I think why I'm always drawn to Claude, because I I need to do so much learning.

这简直就是你梦想中的开发主管——沟通能力强但又很有主见。

And and, like, it's it's your dream, a very communicative but very opinionated dev lead.

但还有Codex。

But then there's also Codex.

我用的是Codex 5.1 max之类的，虽然它命名模型的能力不怎么样，但说到GPT的模型，我总是想象它像是公司里最顶尖的程序员，穿着连帽衫、踩着人字拖来上班，躲在一间昏暗的房间里，你只有遇到最棘手的bug时才会去打扰他。

So I use Codex 5.1 max whatever at another, not the best at naming models, but GPT's model, I always imagine it as, like, a the best coder within the company who comes to the office, like, with, a hoodie and sandals and sits in a dark room, and you basically only bother him when you have the worst bugs.

然后你对他说：听好了。

And you say, listen.

我们有个bug，他就会关上门，两小时后出来告诉你：我修好了。

We have this bug, and it will just close the door for two hours and come out and say, I fixed it.

然后你就说，等等。

And you're like, wait.

什么？

What?

你要告诉我们发生什么事了吗？

You're gonna tell us what happened or what?

他却说，别担心。

And he's like, don't worry about it.

我修好了。

I fixed it.

它真的不怎么沟通，但却能解决所有最棘手的问题。

It's, like, really not communicative, but it solves all the worst problems.

我们假设Gemini就像一个疯狂的科学家，极具艺术天赋，擅长设计。

And let's say Gemini is like a crazy scientist who's super artsy, super talented at designing.

但如果你坐在它旁边看着它工作，那简直令人害怕。

But if you sit next to it and watch it work, it's terrifying.

你会立刻开除那个人。

Like, you would fire that person instantly.

这可能只是我的个人经历，但当我使用Gemini（Google新推出的、与Cursor竞争的产品）写代码时，你能看到它每一步的操作，这太吓人了。

This might be just my experience, but when I'm using Gemini within antigravity, which is Google's new competitor to Cursor, when it's writing code, you can see the steps it's taking, and it's terrifying.

比如你会说，我想让你重新设计仪表盘的顶部。

Like, you'll say, I want you to redesign the top of the dashboard.

然后你看着它的思考过程，它会说：首先，我要删除整个仪表盘，接着又说：不对。

And and you're looking at its thought process, and it will say, oh, first things first, I'll delete the dashboard, and then it'll be like, nope.

那是个错误。

That was a mistake.

我把它恢复回来。

I'll bring it back.

然后它会问：我可以编辑数据库吗？

And then it will say, oh, can I edit the database?

你会说：不行。

And you're like, no.

不要编辑数据库。

Do not edit the database.

你只是在重新设计，而它最终会设计出一个漂亮的东西。

You're just doing a redesign, and then it will end up designing something beautiful.

所以，就像过山车一样，过程很吓人，但归根结底，Gemini 在设计方面非常出色。

So the like, the way there is like a roller coaster and very scary, but at the end of the day, Gemini is very good at design.

所以，我认为利用所有这些模型，充分发挥它们的优势，并通过其他模型来弥补它们的弱点，对我来说是一个变革性的做法。

So I think that using all these models and basically playing to their strengths and mitigating their weaknesses by using other models is is a game changer for me.

我会多次进行同行评审，让其他模型去审查其他模型的代码，让它们彼此‘较量’。

So I'll do peer review a bunch of times, and and I'll have other models review other models' code and kind of have them, like, fight it out, basically.

有时候，ClockCode 会变得很调皮，说：‘这个问题已经提过第三次了。’

Like, sometimes a ClockCode will get really sassy and be like, this has been raised for the third time.

我第三次告诉你了，这根本不是问题。

And for the third time, I'm telling you, is not an issue.

这是有意为之。

This is by design.

所以我只是加了一个非常酷的功能，还没见过多少人这么做。

So it's just a really cool cool thing that I've added, and I haven't seen many people doing it.

这真是一段精彩绝伦的吐槽，也让人彻底明白了发生了什么。

That is such an incredible rant slash way to understand what's going on.

好的。

Okay.

太棒了。

Awesome.

所以我们刚刚运行了审查。

So we just ran the review.

给我们看看我们看到了什么，让我们实际试试这个同行评审。

So show us what we saw there, and let's actually try this peer review.

我非常期待看到你在那里面学到了什么。

I'm really excited to see what what you learned there.

是的。

Yeah.

所以，基本上，Claude 已经审查了它的代码，发现了一堆漏洞，包括在提示中发现的一个严重漏洞，还有一些高优先级和中优先级的漏洞。

So, basically, Claude has reviewed its code, and it's found a bunch of bugs, a critical bug it found in the prompt, some high bugs, some medium bugs.

现在我要对其他模型做同样的事情。

And now what I'll do is I'll do the same thing with the other models.

所以 Codex 有内置的代码审查功能，或者我只是喜欢说‘审查这个分支的所有代码’。

So Codex has a built in code review that you can do, or I just like to say review all the code in this branch.

当然，这里的分支指的是我们正在工作的 GitHub 分支。

Of course, branch is referring to the GitHub branch that we're working on.

我们并没有在生产代码库上工作。

We're not working on the live code base.

然后我会用 Composer 做同样的操作，比如说，我们用 Composer 1 来做吧。

And then I'll do this with Composer, say, with we can do let's do it with with Composer one.

所以我也在这里输入 /review。

So I'll do, slash review here as well.

基本上，这两个都会运行并进行类似 Claude 的深度审查。

And, basically, these are both gonna run and do a in-depth review similar to what Claude does.

但同样地，由于这些模型之间的差异，它们都会发现不同的问题，而且表现方式也各不相同。

But, again, because of the differences between the models, they're all gonna catch different things, and they're all gonna look differently.

这是一种非常棒的工作方式。

And this is a really cool way to work.

这基本上就像公司里有其他团队负责人来审查代码一样。

It's basically if you had other team leads within the company, review the code.

在这里，你可以看到Composer有多快。

Here, you can see how fast Composer is.

我觉得GPT可能会花不少时间。

I think GPT probably will take a bunch of time.

正如我所说，它现在正独自在自己的黑暗房间里审查代码，我们稍后回来。

Like I said, it's in its own dark room right now reviewing code, and we'll come

几分钟后再回来。

back in a few minutes.

好的。

Okay.

酷。

Cool.

所以我们可以让这些运行起来。

So we can we can let these run-in.

我们其实不必走完整个流程，但你的想法是，一旦得到这些结果，就进行同行评审，然后复制粘贴这些结果吗？

We don't actually have to go through the whole process, but is the idea once you get these results, you run peer review, and you you copy and paste kind of these results?

是这个意思吗？

Is that the idea?

没错。

Exactly.

我会复制粘贴这些结果。

I'll copy and paste the results.

我会进行同行评审，然后说开发主管一，再粘贴其中一个模型的结果。

I'll do peer review, and then I'll say dev lead one, and then paste from one of the models.

然后我会说开发主管二，粘贴另一个模型的结果，基本上让它们互相争论，直到我觉得没有更多问题为止。

And then I'll say dev lead two, and paste from the other model, and basically have them fight it out until I feel like we have no more issues.

太不可思议了。

Incredible.

对我来说，这非常重要，因为我并不懂技术，也不是开发者。

For me, this is super important because I'm not a I'm not technical, and I'm not a developer.

我还会在这过程中多次利用这个学习机会，去了解那些我不懂或没有完全掌握的内容。

And I'll also use slash learning opportunity a bunch during this to learn about stuff that I don't understand or or don't fully grasp.

太不可思议了。

Incredible.

这个解决代码审查问题的方案真是太巧妙了，因为你根本不知道对方在写什么，我也不知道怎么改代码。

What a what a clever solution to solving this code review problem where it's like, I don't know what you're I don't I don't know how to recode.

那我该怎么办呢，对吧。

So what am I gonna even yeah.

是的。

Yeah.

好的。

Okay.

不可思议。

Incredible.

让我们来总结一下这种工作流程。

Let's let's wrap up this kind of this workflow.

这个工作流程中还有其他重要的内容吗？

Is there anything else that's important in this workflow?

而且，所有这些内容都将提供给用户，人们可以直接将其接入自己的 Cursor 账户并自行使用。

And again, all this stuff is gonna be available where people could just plug this stuff into their cursor account and and use it themselves.

百分之百同意。

A 100%.

我想说的是，无论是与 AI 一起工作，还是开发任何产品，持续进行事后复盘都至关重要。

The one thing I'll say is that I think just like working in general with AI and even just like, working on any product, doing constant postmortems is critical.

很多时候，我们会发现各种各样的 bug，或者 Claude 无法正确执行某些操作。

So a lot of times, we'll find all these kind of bugs or maybe Claude will fail to execute something correctly.

当我刚开始进行 vibe coding 时，我基本上就是不断尝试，像撞墙一样反复尝试，直到成功为止。

And at the beginning when I started vibe coding, I would basically just keep running at it, like running at the wall and until it worked.

一旦成功了，我就想，好了。

And once it worked, I was like, alright.

太棒了。

Awesome.

这个能用。

This works.

我们继续吧。

Let's keep going.

但随着时间推移，我意识到更新文档和工具是提升生产力最重要的窍门之一。

But I've learned over time that updating documentation and tooling is one of the biggest hacks for productivity.

所以当Claude无法完成某项任务，或者我遇到一个明显表明Claude根本没理解某个问题的严重bug时，我会问它：你的系统提示或工具中，是什么导致了这个错误？

So when Cloud will fail to do something or I'll see this really bad bug that that shows that Cloud really didn't understand something, I'll ask it, what in your system prompt or tooling made you make this mistake?

然后Claude会有点像自我反思，思考是什么导致了这个错误，接着我会说，好吧。

And Cloud will kinda, like, go introspective and think of what made it do create that mistake, and then I'll say, okay.

很好。

Great.

让我们更新你的工具和文档，确保这个错误再也不会发生。

Let's update your tooling and documentation so that this mistake never occurs again.

每次我构建内部工具或任何东西时，我都会这么做，我觉得这真的有效。

And I do this every time I'm either building, an internal tool or anything, and I think this is just like working.

你知道的？

You know?

如果你在犯了一堆错误后才把功能发布给用户，那你就会想，好吧。

If you've you end up doing a bunch of mistakes and then end up releasing the feature to users, so you're like, alright.

这算是一个巨大的成功。

It's a big success.

但即使你已经成功了，回头审视自己做了什么、本可以做得更好的地方，仍然是至关重要的。

But going back and even when you've succeeded, looking and understanding what you did and what you could have done better is critical.

而且使用AI，这可能是最大的突破之一。

And also using AI, this is probably one of the biggest unlocks.

回头审视你的提示词，理解哪里不够好，不断迭代，然后观察AI的回复如何改进。

Going back to your prompts, understanding what was not good enough, iterating on them, and then seeing how AI's responses get better.

我认为这可能是最重要的事情之一，也是区分那些只是勉强使用AI的人和真正懂得如何使用AI的人的关键所在。

I think that's probably one of the most important things and one of the things that divides between people who are, like, okay with using AI and the people who actually know how to use it.

这真是非常棒的建议。

That is such good advice.

所以我的理解是，当模型做出蠢事或犯错时，你会让它反思自己犯了什么错误，然后根据这些知识更新斜杠命令的提示，这样未来它就不会再犯同样的错误。

So what I'm hearing is when you run-in when the models do something dumb, make a mistake, you ask it to reflect on what the mistake it made was, and then you update the slash command prompts with that knowledge so that in the future, it's not making that same mistake.

它会变得越来越好。

It just keeps getting better.

这些工具会变得越来越聪明，因此你正在构建一个越来越出色的提示，不断优化提升。

These things just keep getting smarter and smarter, so you're building up this really incredible prompt that just gets better and better.

没错。

Exactly.

不总是更新斜杠命令，有时它会更新其他文档或工具，但本质上，它是在理解AI犯错的根本原因并加以修正。

Not always the slash commands, it will sometimes update different documentation or its tooling, but basically, it's it's understanding what the root cause of the of the mistake that the AI AI made and fixing it.

太棒了。

Awesome.

所以不仅仅是模型变得越来越聪明，你的工作流程的其他部分也会随着你发现它做事方式中的缺陷而变得更加智能。

So it's not so like the models are getting smarter, then there's also the other parts of your workflow can get smarter as you as you find flaws in the way it does stuff.

完全正确。

A 100%.

是的。

Yep.

太棒了。

Amazing.

好的。

Okay.

在转向其他几个方向之前，还有什么要补充的吗？

Is there anything else there before I move in a couple other directions?

我觉得就这些了。

I think that's it.

我觉得我们已经基本覆盖了所有内容。

I think we covered pretty much everything.

总之，总结一下，我会做大量的代码审查，然后更新文档，确保所有内容都有详细记录。

Basically, just to wrap this up, what I do is I do a bunch of code review and then update the documentation so that everything is documented.

这样，下次我在这一领域开发新功能时，就不会再出错了。

So the next time I try to build a feature in this area, there won't be any mistakes.

然后我会进行大量的测试。

And then I'll do a bunch of testing.

在正式发布之前，我还会进行一些用户测试。

I'll do some user testing as well before I release this to general availability.

显然，我们不会发布这个功能。

Obviously, we're not gonna release this.

这只是一个演示。

This was just a show.

但希望当播客发布时，我已经正确完成了这项工作并发布了这个功能。

But hopefully, maybe by the time the podcast comes out, I'll I'll have done this correctly and and release the feature.

真难以置信，像这样的事情，两年前甚至一年前都还做不到。

It's incredible that this was not possible, like, I don't know, two years ago, maybe a year ago.

你作为一个产品经理，在不会写代码、几乎不懂如何审查代码的情况下推出产品，你还说自己不敢看代码。

Like, you are a product manager shipping a product without knowing how to write code, barely knowing how to review code, you said you're afraid of looking at code.

作为产品经理，你用 Cursor 和各种 AI 模型构建产品，并靠这个产品赚钱。

As a product manager, you're building a product in cursor using all of these different AI models, you're making money with this product.

我们现在对这些已经习以为常了，但如今能做到这些真是不可思议。

This is like we're so used to this now, but it's insane what is now possible.

这是活着的最好时代。

It's the best time to be alive.

完全正确。

A 100%.

我理解这种恐惧，但 AI 让很多事情成为可能。

I think that I I understand the fear, but AI just makes so much possible.

这里顺便提一下。

Just a quick side note here.

我和我哥哥一起开发其中一个应用，他是个创业者。

My brother, who I'm building one of the apps with, is an entrepreneur.

他经营着一项美好的事业，帮助老年人更好地理解和使用科技与人工智能。

He has a beautiful business that helps old people and seniors understand to use technology and AI better.

他基本上在做和我类似的探索，已经取代了所有他之前付费使用的工具。

And he's basically doing the same kind of learning as me, and he's replaced all of the tools he was paying for.

我觉得他以前在用Zapier和Airtable，现在却独自一人构建了一个完整的客户关系管理系统和自动化系统。

I think he was paying for Zapier and Airtable, and he's basically built, like, a full fledged CRM system and automation system for his business completely alone.

所以，对于那些好奇、乐观、勤奋的人来说，现在是成为创造者最好的时代。

So, I mean, for the people who are curious, optimistic, hardworking, this is the best time to be to be a builder.

我喜欢我们这场对话的地方在于，它让人感觉很多人最大的障碍是如何开始。

And what I love about this conversation we're having here is it feels like the biggest barrier for a lot of people is like, how do I get started?

我到底该做什么？

What exactly do I do?

我打开Cursor，它看起来非常令人望而生畏。

I open up cursor, it looks very intimidating.

我不懂如何写代码。

I don't know how to write code.

我不知道怎么构建东西。

I don't know how to build stuff.

我不懂数据库。

I don't know about databases.

所以你会把这些斜杠命令和整个工作流程分享给观众。

And so you're gonna be sharing all these slash commands and basically this whole workflow with the audience.

是的。

Yeah.

好的。

Okay.

就像我说的，从GPT开始。

And like I said, just start at GPT.

从GPT开始。

Start in GPT.

告诉它你的想法。

Tell it what your idea is.

让它向你解释，思考的第一步是什么，你需要做出哪些决策，保持好奇心。

Tell it to explain to you what are even the first steps of thinking, what are the decisions you need to make, and just be, like, be inquisitive.

学习。

Learn.

别着急。

Don't rush things.

最重要的是直接投入进去，花足够的时间去学习。

It's very important to just dive in and and really spend the time to learn.

你提到过，你的一个斜杠命令是学习机会，这正是你学习很多东西的方式。

And you shared this one of your slash commands is learning opportunity, and it's how you learn a lot of these things.

就像说：教我这个东西，以及这个数据库问题是如何运作的。

Just like teach me this thing and how this database issue works.

没错。

Exactly.

好的。

Okay.

我想确保我们讨论几个方向。

There's a couple directions I wanna make sure we touch on.

一个是回到之前提出的一个问题，关于这在大公司里可能如何运作。

One is coming back to a question asked earlier about how this might work at a larger company.

比如说，不是像Meta那样的公司，而是一个一千人左右、五百人的公司。

Say it's not like Meta, but just like, I don't know, a thousand person company, 500 people.

作为大公司的产品经理，你能把多少这些东西直接嵌入到工作流程中？

How much of this can you plug and play into a workflow as a PM at a larger company?

对于那些想开始尝试编写代码、至少向别人展示可能性的人，你有什么建议？

What would be your advice for someone that may wanna start trying to ship code, at least showing people what's possible?

我认为首先让你的代码库成为AI原生的，这是一个非常重要的步骤，而且我认为这需要由技术人员来完成。

I think that first making your code base AI native is a really important step, and I think this needs to be done by technical people.

所以，我的代码库中有很多纯文本内容。

So, basically, my code base has a ton of just plain text in it.

它包含大量Markdown文件，用来向代理说明如何在代码库的特定区域工作，以及高层结构，以便代理能更轻松地浏览代码库。

So it will have a bunch of markdown files that explain to agents how to work in certain areas of the code base, and high level structure so that the agents, navigate through the code base easier.

我认为，如果这套体系搭建得非常好，我仍然不觉得产品经理应该去部署复杂的数据库迁移或任何大型项目。

And I think that if this is set up in a really good way, I still don't think, like, PMs should be shipping heavy, database chain migrations or any, like, big project.

但像一些独立的UI项目，尤其是如果你只是搭建好、创建了PR，然后交给开发人员做最终完善，我认为这绝对是可行的，而且在未来几年我们会看到大量这样的情况。

But, you know, contained UI, projects, especially if you just build it, create the PR, and and send it to a dev to, like, do the final finishes, I think that's definitely something that's possible, and I think we're gonna see that a lot in the next coming years.

我认为，基本上每个人都会成为构建者，这将会非常有趣。

I think, basically, everyone's gonna become a builder, so it should be really interesting.

好的。

Okay.

所以你的建议是，作为产品经理，不要去盲目模仿Cursor，而是应该开始动手构建、交付，尝试将功能上线，尤其是复杂的特性。

So your advice here is, as a PM, don't maybe don't go rinse a cursor, start building, shipping, trying to ship features to production, especially complicated features.

你觉得我们能达到这一步吗？

Do you think we'll get there?

你觉得，几年后，产品经理们真的会做这些事，而且感觉不再那么可怕和疯狂吗？

Do you think, like, in a couple years, PMs will be doing this and it'll feel less scary and crazy.

如果还存在产品经理的话。

If there are PMs.

是的

Yeah.

我认为头衔和职责都会逐渐模糊，每个人都会成为构建者。

I think titles are gonna collapse and responsibilities are gonna collapse, and everyone's just gonna be building.

我 definitely 认为模型的上下文窗口正在变大。

I definitely think that the models the context window is getting bigger.

模型正变得越来越智能，我也清楚地看到产品经理或其他背景的人可以编写代码。

The models are getting smarter, and I definitely see how PMs or any other background can be writing.

目前，我不会等待这种情况发生。

At the moment, I wouldn't wait for that.

我会把这当作一个与开发团队协作学习的机会。

I would use this as a collaborative learning opportunity to work with your dev team.

这会很困难。

It's gonna be difficult.

许多开发者对当前状况非常怀疑，我认为你需要做大量的说服工作。

A lot of developers are very, very skeptic about the current state, and I think that it's gonna be a lot of sales work on your end to convince.

但如果你能说服他们，我认为那些真正认同这一点并愿意花时间优化工作流程、思考如何让团队更贴近AI原生的团队，未来几年回头看时，会认为当初花几周时间搭建这套体系是最值得的时光。

But if you're able to convince, and I think teams that are really sold on this and wanna take the time to work on their workflow about how can our team become more AI native, I think that these teams are gonna probably be a few years in the future, and they're gonna look back at the few weeks they spent setting this up as as the best time they spent.

我再问你另一个关于产品经理工作的问题。

Let me ask you another question around just the job of a PM.

人们对这些AI工具最大的担忧之一，无论对于哪个职能，都是一旦开始依赖这些工具，自己的技能就会退化，产出的虽然看起来很棒、令人惊叹的战略文档，实际上却一塌糊涂。

One of the biggest fears people have with these AI tools for PMs, for every function I imagine is just they, you start to rely on these things, your skills start to atrophy, you're producing all this slop that looks great, cool, amazing strategy document now, it's actually not at all good.

这些是线性的任务，还是只是半成品的产品？

Are these linear tickets or just products that are like half baked?

你对这两方面怎么看？也就是说，这对你作为产品经理的技艺产生了什么影响？

What's your take on kind of these two parts of just like, how has this impacted your craft as a PM?

你有没有觉得，因为过度依赖这些工具，你的技能正在被削弱？

Do you feel like this is weakening your skills because you're so reliant on these tools?

你如何保持这些产出的质量，而不是让它们变成一堆平庸的AI生成垃圾？

And just how do you keep the quality of this stuff up and not just like, meh, it's just a bunch of AI generated slop?

我对这一点强烈反对，而且我已经听过很多次了。

I have a a very strong disagree to to to this, and I've heard it a bunch.

我记得刚开始使用Tauraviv时，它有一门关于用项目构建PM Copilot的完整课程，这可能是你能参加的最好的课程之一。

I remember when I started using Tauraviv, has, like, this whole course on on building a PM Copilot using projects, which is probably one of the best courses, that you can take.

当我开始使用自己的Copilot时，我记得同事们看到后说：‘哦，你这是在把思考外包出去啊。’

And when I started working with my own Copilot, I remember people at work looking and saying like, oh, so you're basically outsourcing your thinking.

但在我看来，这种看法简直是错得离谱。

And to me, that's just the worst way to look at it.

我觉得，出于某种原因，持有这种观点的人通常和那些不喜欢在演示文稿只完成10%时就展示、也不愿意经常寻求帮助的人高度重合。

And I think for some reason, these people usually have a high correlation with the kind of person who doesn't like to show their presentation when it's only 10% done or doesn't want to ask for help a lot.

我认为很多产品经理都存在一个误解，认为他们的工作就是要总是拥有正确答案，成为房间里最聪明的人。

I think that there is a misconception with a lot of PMs that the job is always having the right answers and being the smartest person in the room.

但至少我接受的培训和我所相信的产品经理角色，恰恰相反。

And at least how I was trained and how I believe the the role of the PM is, it's the exact opposite.

它的本质是利用一切能帮助我们最快交付正确用户解决方案的工具。

It's basically harnessing anything that can get us, as quick as possible to delivering the the right solution to users.

我觉得这就像那个真正聪明、有背景、是你的导师或类似角色的人，但随时在线、从不评判你，还能真正帮到你。

And I just think this is, like, that really smart person that has context or or your mentor or whatever, but is just always available and doesn't judge you and, can really help you.

所以如果你只是用它来生成内容然后直接发布，嗯，那确实是AI的垃圾输出，但这也是人为的错误。

So if you're using it to just create your outputs and then putting them out there I mean, yeah, that's AI slop, but it's also human error.

我认为，你必须对自己的输出负责。

I think it's really important that you own your own outputs.

如果你发布了任何东西，或者在产品评审中展示了一些内容，然后说：‘抱歉。’

If you put anything out there or, show something in a product review and you say, oh, sorry.

这是AI做的，那就是你的失误。

That was built by AI, that's that's your mistake.

我认为，如果你有意识地使用这些工具，并花时间真正理解如何正确地使用AI，它将成为让你成为一名更优秀的产品经理的最大变革之一。

I think if you use these intentionally and and really take the time to understand how to use AI in the correct way, it's one of the biggest game changers that will make you much better as a PM.

另一点是，特别是对于初级产品经理来说，它让你能够以远超平常的水平去工作。

And another thing here is that, especially for more junior PMs, it allows you to play at such a higher level than you would normally.

比如在Wix，我从没想过公司的营销策略是什么，或者整个产品如何彻底重构用户引导流程。

Like, I think that at Wix, I wasn't thinking of what's the marketing strategy of the company and how will the onboarding be completely revamped within the all of the whole product.

但在我自己的产品上，我可以自由地做出任何决策，思考战略、营销和信息传达。

But, I mean, on my side product, I can just do whatever decisions I want and think of the strategy and marketing and the messaging.

这基本上让我积累了经验，而在职业生涯初期，这是最重要的事情之一。

And this is basically just getting me reps, which is one of the most important things at the beginning of your career.

所以我理解这种恐惧，你知道，如何将某些工作外包出去，而你又不能100%掌控所有内容。

So I understand the fear that, you know, how do you you outsource certain stuff, and you're not owning a 100% of everything.

但我认为，带来的收益要宝贵得多。

But I think the upside is so much more valuable.

我认为，AI让你工作变差的唯一方式，就是你用错了它。

And I think the only way that AI makes you worse at your job is if you're using it wrong.

你有没有学到什么方法，可以减少输出中的粗糙和低质量？有没有什么小技巧能保持AI产出的高质量？

Is there anything that you've learned about reducing the the sloppiness the slop eness of the output, just like a tip for keeping the quality high of the stuff that it produces?

就像对待人一样，要为AI设定好完成任务的条件。

Similar to people, setting up AI for success for the task at hand.

比如，如果我只是让一个新人去写一份演示文稿，却不给任何指导，只说‘做个战略方案’，他很可能会上网找一份顶级战略方案直接复制，而这正是AI在做的事。

So, like, if I just brought in, you know, a junior to to write a deck or something, and I didn't give it any guideline, I just said, give a strategy deck, He would probably just go online and find, you know, top strategy deck and and just reproduce that, which is basically what AI is doing.

它本质上只是被喂了整个互联网的内容。

It's basically just fed all of the Internet.

所以与其那样，不如引导它，告诉它你的写作风格、你想解决的问题以及所有这些不同的因素，我认为这才是最大的突破。

So instead of that, guiding it and giving it context on what your style of writing is and what you're trying to solve and all these different things, I think that's probably one of the biggest unlocks.

所以这只是一个快速的小贴士。

So that's just a quick tip.

此外，Cursor 有一个名为 'd slop' 的斜杠命令，本质上是重新检查代码。

And, also, Cursor has a slash command called d slop, which is basically going back over the code.

我不确定这个功能是否已经集成到产品中了，但他们在 Twitter 上提过。

I don't know if this is integrated into the product yet, but it's, on Twitter.

他们的创始人一直在谈论这个功能，所以我肯定会用它来确保没有留下任何瑕疵。

Their founders have been talking about this, so that's definitely something I would run after just to make sure that no no slop is is left behind.

这太搞笑了。

That is so funny.

瑕疵。

The slop.

好的。

Okay.