本集简介
双语字幕
仅展示文本字幕，不包含中文音频；想边听边看，请使用 Bayt 播客 App。
那个软件的入口将吞噬整个世界。
Entrance in that software will eat the world.
我觉得编程会吞噬所有知识型工作。
I I feel like coding will eat all knowledge work.
对吧？
Right?
而且我们已经正在朝这个方向发展了。
And we're kind of going that direction already.
整个智能代理栈正在兴起。
The whole agent stack is emerging.
是的。
Yeah.
身份、支付、营销，甚至命令行界面与MCP的对比。
Identity, payments, marketing, even CLI versus MCP.
这些全都是全新的东西，我认为很多旧的玩法都将失效。
And, like, all of these are really new things, and I think a lot of the old playbook goes away.
是的。
Yeah.
这是一个全新的世界。
It's it's a whole whole new world.
我希望更多公司能保持小型化。
I hope more companies will stay small.
而且我认为这一代的创始人已经意识到了这一点。
And and I I think the founders of this generation realize that.
他们希望尽可能保持小规模。
Like, they wanna stay as small as possible.
是的。
Yeah.
而不是组建一个十人的产品团队，你可能只需要两到三人的产品团队，再搭配一堆AI代理来协助你。
And instead of having, a 10 person product team, you have, like, a two two or three person product team, and you also have a bunch of AI agents to help help you.
是的。
Yeah.
有人发推说，就业市场太差了，我现在反而能去追求自己的梦想了
Someone tweeted that, like, the the job market is so bad that I can only pursue my dreams now
或者类似这样的话。
or something like that.
所以，就像，你知道的，就像，是的。
So, like, it's it's like, you know, it's like, yeah.
所以也许你失去了工作，但现在你不得不做自己的事情。
So maybe you lost your job, but, like, now you have to do your own thing.
这是一种可能性。
It's a percent.
而且有机会真正实现它。
And have a shot at actually achieving it.
你知道的？
You know?
是的。
Yeah.
人们打开手机是为了感受某种情绪。
People open their phone to feel something.
连接感。
Connected.
成就感。
Productive.
娱乐感。
Entertained.
每个应用都是通往不同情绪状态的门。
Each app is a door to a different emotional state.
2007年，iPhone带来了应用网格。
In 2007, the iPhone gave us the App Grid.
十九年后，数十亿人每天仍会多次点击这些相同颜色的方块。
Nineteen years later, billions of people still tap the same colored squares dozens of times a day.
这个界面变得如此熟悉，以至于人们不再觉得它像科技产品。
The interface became so familiar, it stopped feeling like technology.
它变成了本能反应。
It became Reflex.
现在，一代开发者正将整个网格压缩成一次对话。
Now a generation of builders is collapsing that entire grid into a single conversation.
一个代理可以检查你的数据分析、更新你的文档、帮你跑腿，并在你晨间散步时给你打气。
One agent that checks your analytics, updates your documents, runs your errands, and gives you a pep talk on your morning walk.
不是因为应用失败了，而是因为说话比点击更快。
Not because the apps failed, but because talking is faster than tapping.
问题是，当构建软件的成本几乎为零时，产品、公司和职业会发生什么变化？
The question is, what happens to products, companies, and careers when building software costs almost nothing?
a16z普通合伙人阿尼什·阿查里亚与Roblox的创建者兼产品负责人杨彼得对话。
A16z general partner Anish Acharya speaks with Peter Yang, creator and product lead at Roblox.
好的。
Alright.
欢迎各位。
Welcome, everyone.
我请来了我的朋友彼得·杨。
I've got my friend Peter Yang here.
彼得，欢迎。
Peter, welcome.
是的。
Yeah.
很高兴来到这里。
Good to be here.
很高兴
It's good to
再次见到你。
see you again.
对。
Yeah.
对。
Yeah.
见到你真好。
It's great to see you.
彼得和我在信用 Karma 工作过。
Peter and I worked together at Credit Karma
嗯。
Mhmm.
短暂地待过一段时间，
For a
然后我们就分道扬镳了。
brief stint, and then we went our separate ways.
我通过他在 X 平台和 YouTube 上的大量发帖重新发现了彼得。
And I rediscovered Peter from his prolific posts on X and your YouTube.
是的。
And Yeah.
你有点像克拉克·肯特和超人的组合，因为你还在上班。
You've got a little bit of a Clark Kent Superman thing going because you still got a day job.
对吧？
Right?
没错。
That's right.
我仍然有一份全职工作。
I still have a day job.
是的。
Yes.
对。
Yeah.
你能展示一下在哪里吗？
Can you show where?
是的。
Yeah.
我在Roblox担任产品经理。
I work at Roblox as a PM.
太棒了。
Amazing.
Roblox。
Roblox.
还有Jason的投资公司。
And Jason Portfolio Company.
是的。
Yes.
我最喜欢的一个。
One of my favorites.
太厉害了，老兄。
Well, incredible, man.
我们直接开始吧。
Let's get right into it.
也许我先问一个轻松有趣的问题，然后我们再聊聊claw生态系统里的所有内容。
Maybe I'll start with a softball fun question, and then we're gonna talk about everything in the claw ecosystem.
我们要谈谈编码代理。
We're gonna talk about coding agents.
我们会稍微聊聊学生应该学什么、一些建议，以及你在网上提到的一些内容。
We're talking about a little bit about maybe what students should study, advice, and some of the things that you've talked about online.
是的。
Yeah.
当然。
Sure.
也许从这个问题开始：你一共有多少个爪子？
Maybe to start, what is the name of your oh, how many claws do you have?
告诉我它们的名字。
And tell me their names.
我只有一个。
I only have one.
我叫她佐伊。
I call her Zoe.
Zoe？
Zoe?
但我跟她进行了多次对话。
But I have, like, multiple conversations going with her.
好的。
Okay.
是的。
Yeah.
为什么叫Zoe？
And why Zoe?
我有两个女儿，本来打算给小女儿取名Zoe，但最终没用。
I have two girls, and I was gonna call my younger one Zoe, and I did not.
所以我干脆把我的OpenClaw叫Zoe了。
So I'm like, oh, I called my OpenClaw Zoe instead.
我明白了。
I see.
是的
Yes.
对
Yeah.
这是你的备选方案。
This is your fallback plan.
对
Yeah.
彼得，跟我讲讲OpenClaw吧，你是怎么发现它的，现在是怎么使用的，以及你认为它有什么影响。
Peter, tell me a little bit about OpenClaw, how you discovered it, how you're using it today, and what you think the implications are.
对
Yeah.
我很幸运，在彼得·斯坦伯格变得非常出名、一切爆发之前采访了他。
I was lucky to interview Peter Steinberger before he became super famous and the whole
事情就突然爆红了。
thing blew up.
我采访完他之后，立刻就去设置，结果花了超长时间。
And then right after I interviewed him, like, up the thing, it took forever to set up.
整个系统特别粗糙。
It was super janky.
嗯。
Mhmm.
而且，是的，它为我做了很多事。
And and, yeah, it does a lot of things for me.
它能帮我从YouTube和我的Mercury后端账户中提取数据分析。
It, like, pulls analytics for me across YouTube and, like, my Mercury back end account.
嗯。
Mhmm.
它能帮我更新Google文档。
It can update Google documents for me.
它还能帮我搭建小型网站。
It can build little web websites for me.
但如果你要我坦白说，老兄，我主要就是用语音和它交流，然后听它用语音回复我。
But if I was honest with you, dude, like, I mostly just talk to it through voice and get voice replies.
每隔一天，我就会让你给我来一段鼓舞士气的话。
And, like, every other day, asked you to give me, like, a pep talk.
让我看看你所有的记忆，给我一些我根本不知道的深刻见解。
Give me like, look look through all your memory and, like, give me some, like, deep insights that I I don't know about.
好的。
Okay.
它给了我一段，我记得当时我正在散步，它给了我一段三分钟的鼓舞话语，真的特别棒。
And it it gave me, like like, I remember I was on a walk, and it gave me, like, a three minute pep talk that was, like, really amazing.
特别棒。
Really amazing.
它说的是，哦，你跟我聊你的创业事业啊、工作啊之类的。
It was something about, oh, you're, like, talking to me about your creator business and blah blah and, like, your job.
但别忘了，你的七个孩子很快就会长大，他们想和你多待一会儿。
But just remember that your kids, seven of four, are gonna grow up very soon, and they're gonna wanna spend time with you.
哇。
Wow.
所以你应该为他们重新优化。
So you should reoptimize for them instead.
是的。
Yeah.
对。
Yeah.
这真的很酷。
That's really cool.
是的。
Yeah.
而且，我的意思是，这确实很酷，但所有语言模型之前其实都能做到。
And, I mean, very cool, but also something that all the language models could have done prior.
对。
Yeah.
那么，这和那种使用场景有什么区别呢？
So what's the difference between this and a use case like that?
是的。
Yeah.
这是个非常好的问题。
That's a very good question.
我不知道。
So I don't know.
因为我感觉安装在 Telegram 上，比用我的云服务或 ChatGPT 更加私密。
I because I think installed on Telegram, it just feels like more personal than using my Cloud or ChatGPT.
而且感觉就像我可以在床上发消息一样。
And it just feels like something I can, like, text in bed.
所以这可能不太健康，但我确实会在床上给它发消息。
So it's probably not very healthy, but, like, I text to it in bed.
我通勤时也跟它聊过，感觉它更像是一个真正的人类伙伴。
I I've talked to it during my commute, and it feels like it feels more like a personal, like, actual human.
是的。
Yeah.
对。
Yeah.
那么对你来说，OpenClaw 在界面方面有多重要，比如把它推向消息平台，或许能帮助欺骗我们的大脑，让我们觉得——
So so how much for you is OpenClaw about the kind of interface, like pushing it to messaging and maybe helping to trick our brain into feeling like, hey.
这像是一个人，或者类似人的存在。
This is a person or a person esque thing
嗯。
Mhmm.
与其他所有组件相比，比如自我修改、技能目录，是这样的。
Versus all the other components of the stack, the self modification, the skills directory Yeah.
所有的
Every all
其余的部分。
the rest.
我觉得这80%可能都归因于它那种人性化的特点，因为我主要就是通过语音和它交流，然后了解它。
I think it's probably 80% just per per personable part of it because I I mostly just talk to it and, like, know, through voice.
但我还觉得，首先，它确实挺粗糙的。
But I also think, like, is is something for for first of all, it is pretty janky.
它经常忘记事情，是的。
It tends to forget things a lot Yeah.
而且得不断提醒它。
And to keep reminding it.
但不管我有什么奇奇怪怪的想法，只要和它说一说，它大概就能办到。
But, like, any kind of zany idea that I have, I just have to talk to it, and it can probably just do.
前几天，我就和它用语音回复功能互动，感觉就像那样。
It was kinda just like like, the other day, I was doing voice replies with it.
我当时说，嘿。
I was like, hey.
我们能不能直接来个电话通话？
Can we just have a live phone call instead?
嗯哼。
Mhmm.
然后我说，好吧。
And and then I was like, okay.
你得连接 Twilio。
You gotta connect Twilio.
你得做所有这些事情。
You gotta do all this stuff.
然后，行吧。
And then, Fine.
我就去做了。
I went off and did it.
是的。
Yeah.
然后我们打了电话。
And then we had a phone call.
它给我打电话了。
It it called my phone.
真的吗？
Oh, really?
你已经设置好了？
You have that set up?
我一直想设置这个功能。
I've been dying to set that up.
好的。
Okay.
不过效果不太好。
It's not very good, though.
延迟很严重。
Like, the latency is bad.
但我能把它弄起来，这已经挺厉害了。
But the fact that I was able to get it going is, like, pretty impressive.
所以，我任何疯狂的想法，基本上都能实现。
So it's kind of like any kind of crazy idea I have, you can kinda kinda do.
那在实际操作中，你是怎么做的呢？
And then in practice, how are you doing that?
你是让它即时编写一个技能吗？
Are you asking it to write a skill on the fly?
你是发现了一个技能吗？
Are you discovering a skill?
你实际用了多少代码生成的功能？
How much of the code gen are you actually using?
我的意思是，我和它聊天的方式特别随意，就像跟朋友说话一样。
I mean, I talked to it in a super casual way with, like, just like a friend.
所以我就说，嘿。
So I'm like, hey.
嘿。
Hey.
嘿，佐伊。
Hey, Zoey.
你能给我打个电话吗？
Can you, like, give me a phone call?
好的。
Yes.
好吧。
Okay.
得这么做。
Gotta do that.
我当时想，行吧。
I was like, okay.
行吧。
Fine.
我打开电脑。
I'll open my computer.
我会做所有这些事情。
I'll do all this stuff.
然后给我打电话，我们再一起排查一下问题，然后就好了。
And then give me a call, and then we troubleshoot a little bit, and then it works.
所以是的。
So Yeah.
我在使用Cloud时，会用一些非常复杂的提示词，很长的提示。
I with with Cloud, I have, like, very fancy prompts, like, very long prompts.
但在OpenCloud上，我就只是简单地打字输入。
But with OpenCloud, I I just kinda text it.
对。
Yeah.
这真的很有意思。
It is really interesting.
所以我们实际上已经聊到了几个方面。
So they we sort of touched on a couple things actually.
首先，有移动消息功能。
So one, there's mobile messaging.
对。
Yeah.
还有记忆系统。
There's the memory system.
还有代码生成组件。
There's the sort of code generation component.
嗯。
Mhmm.
你觉得记忆系统有多创新？是因为它是基于文件的吗？
How much do you think the memory system like, is it innovative because it's file based?
你说它会忘记东西，但语言模型也会忘记。
You said that it forgets things, but so do language models.
你觉得这个记忆系统做得怎么样？
Do you think the memory system is well done?
它是阻碍了进展，还是促进了进展？
Does it hold it back, or does it enable it?
我觉得默认的内存系统其实并不怎么样。
Think the default memory system is actually not that great.
好的。
Okay.
据我理解，它的运作方式就像一个 memory.md 文本文件。
Like, the way I understand it works is just like a m memory dot m d text file.
是的。
Yes.
然后每天。
Then every day Per day.
每天，对吧？
Per day, right?
然后每天都会更新。
And then every day it updates.
而且它很容易忘记事情。
And it tends to forget things a lot.
是的。
Yeah.
所以我实际上安装了一个三层记忆系统。
So so I actually installed this, like, three layer memory.
说实话，我并不完全理解，但它看起来很高级。
Some to be honest, I don't fully understand, but it has That's fancy.
它有Toby的QMD搜索工具。
It has Toby's QMD search tool.
好的。
Okay.
所以我安装了那个，然后又安装了一个两GB的组件，之后情况稍微好了一些。
So I installed that, like and then installed, like, a two gigabyte thing, and then it got a little bit better.
好的。
Okay.
但我开始提醒它了。
But I I started to remind it.
我得把它们放进代理里，比如：嘿。
I have to I have to put it into the agents, like, MD, hey.
在回答我任何问题之前，先通读你所有的记忆并检查一切。
Like, before you answer any question from me, go through all your memory and check everything.
是的。
Yeah.
而且它们也容易忘记自己能做这些事。
And they also tend to forget that it can do stuff.
对，没错。
Like Right.
你能更新我的谷歌文档吗？
Can you update my Google Doc?
哦，我可以做这个。
It's, oh, I can do that.
是的。
Yeah.
对。
Yes.
你可以。
You can.
它在你的文件里。
It's in your it's in your file.
对。
Yes.
是的。
Yeah.
是的。
Yeah.
所以你得提醒它。
So you have to remind it.
是的。
Yeah.
是的。
Yeah.
非常有趣。
Really interesting.
嗯，也许我们来谈一谈一些争议吧。
Well well, maybe let's get into a little bit of the controversy.
你曾说过应用程序会消亡，Claw将会无处不在。
You'd said that apps will die, Claw's gonna be everything and everywhere.
我的意思是，给我们讲讲
I mean, talk us through
这种观点。
that point of view.
是的。
Yeah.
首先，我发各种随机的、没怎么深思熟虑的内容。
Well, for first of all, I I I tweet all kinds of random crap that's not super well thought out.
我们都把这些当真。
We take it all as fact.
是的。
Yeah.
对。
Yes.
但我确实觉得，自从我在OpenClaw上安装了Mercury、MCP这些乱七八糟的App之后，我反而很少再打开它们了。
But I do think, like, ever since I set up all these apps like Mercury, MCP, and all this kind of crap on my OpenClaw, like, I I don't actually open those apps much anymore.
嗯。
Mhmm.
但我同意你的观点。
But I do agree with you.
我觉得，最先会消亡或者使用频率最先下降的，是那些你只是为了完成某个任务才打开的App。
Like, I I I think the ones that are gonna die first or, like, maybe get less usage first is, like, apps that you're just opening to try to complete a task.
你是真的想做点事情。
Like, you actually are trying to do something.
明白吗？
Know?
那些用来娱乐的App，大概还能撑得久一点。
Like, apps that you're opening to get entertainment can probably describe a little bit longer.
但那些你用来完成任务的App，其实你只需要让我的助手帮你做就行了。
But, like, apps that you need to complete a task, like, it's just a way you should text my agent to do it for me.
是的。
Yeah.
这就像是你有个特别棒的助理，专门帮你做事。
It's like you have a really good admin just to do stuff for you for you.
是的。
Yeah.
对。
Yeah.
那么，你发现除了Modulo Open Call之外，你的智能手机使用时间减少了多少？
And so how much are you finding has this reduced your smartphone usage outside of Modulo Open Call?
是的。
Yeah.
不。
No.
因为我是个Twitter上瘾者，所以用手机用得太多了。
Because I'm like a Twitter addict, so use my phone way way too much.
但没错，就使用这些应用而言，确实减少了。
But, yeah, in terms of using those apps, it's definitely re reduced it.
是的。
Yeah.
是的。
Yeah.
因为你不会去问Zoe，嘿。
Because you're not gonna ask Zoe, hey.
帮我读一下我的X，告诉我有什么有趣的内容。
Read my X for me and tell me what's interesting.
意思是，它会给我发送一份晨间简报，包含两条热门推文和一些趋势内容。
Mean, it it sends me, a morning briefing with the top two tweets and stuff that, like, trend trends.
但没错，我仍然会打开X浏览一下。
But, yeah, I still open X I look through it.
是的。
Yeah.
是的。
Yeah.
这很有趣，因为我一直有个理论，认为人们打开手机应用是因为他们想感受某种情绪。
It's interesting because I've always had this theory that people open apps on their phone because they wanna feel a feeling.
是的。
Yeah.
当然，也有一部分是功能性的需求，比如你打开日历应用。
And I think, of course, there's some, like, functional set of needs, which is why you open calendar or something.
但我认为，WhatsApp 是为了让你感觉被连接，Slack 是为了让你感觉高效，而 TikTok 则是为了让你感到娱乐。
But I also think that WhatsApp is you wanna feel connected, and Slack is you wanna feel productive, and, of course, TikTok is you wanna feel entertained.
嗯。
Mhmm.
所以我在想，如果只有一个助手，你该如何在不同情境间切换，比如什么时候在调情？
So I do wonder with just one agent, how do you sort of do the context switching of, like, when are you flirting?
什么时候在认真做事？
When are you getting shit done?
我的意思是，从某种角度说，不同的应用为你提供了清晰的意图划分。
I mean, there in a sense, app it gives you a nice it sort of gives you a nice division of the intents
是的。
Yeah.
没错。
That's right.
而这些是你和 Zoe 之间所没有的。
That you don't get with Zoe.
这是个好观点。
That's a good point.
但我确实在 Telegram 上为 Zoe 设置了多个频道。
But I do have multiple channels set up with Zoe in Telegram.
比如，一个频道只是用来随意语音回复，另一个频道是我们一起做项目。
Like, one is just to random voice replies, and the other one is we're actually working on our project together.
哦。
Oh.
还有一个频道是我用来做开发演示的公开频道。
And the other one, I have a public channel where, like, I'm giving dev demos.
我不想泄露私人信息。
I don't wanna reveal private information.
是的。
Yes.
所以我喜欢使用多个频道。
So I like multiple channels.
这是作为子代理实现的吗？
And is that implemented as sub agents?
或者
Or
不是。
No.
这只是我网上找到的一个乱七八糟的设置。
It's just some stanky setup I found online.
比如，你可以设置多个 Telegram 频道，我不确定她是否能跨频道记住对话内容，但至少你可以有独立的对话。
Like, you just you can set up a multiple Telegram channels, and then I'm not sure if she actually remembers across contests across the channels, but, like, you can have separate conversations at least.
明白了。
Got it.
是的。
Yeah.
你对你的代理有多透明？
And how transparent are you with your agent?
它会看到你的个人邮箱吗？
Does it do they see your personal email?
或者
Or
我非常透明。
I'm, like, super transparent.
我买了Mac Mini，并给它设置了独立的邮箱。
Well, I did buy the Mac Mini and set up its own email.
好的。
Okay.
但我给了它读取我的邮箱和日历的权限。
And but I gave it, like, read access to my email and, like, calendar.
我还给了它一些文档的写入权限。
And I also gave it, like, write access to some docs.
是的。
Yeah.
但它好像把我的整个硬盘搞坏了什么的。
But it kinda screwed on my entire drive or something.
是的。
Yeah.
那么，你如何设想OpenCloud呢？它是一种架构和基础功能，对吧。
So How do you imagine OpenCloud, which is sort of an architecture and a primitive Yeah.
它如何被产品化、包装并推向世界？
How does it get productized, packaged for the world?
我的意思是，我认为彼得·斯坦伯格现在正在OpenAI做这件事。
I mean, I think that's what Peter Steinberg is working now at OpenAI.
对吧？
Right?
是的。
Yeah.
他可能会为ChatGPT开发一些功能，因为每个人都在使用ChatGPT，这样它就能真正为你完成任务，甚至可能显得更有人性。
He's probably gonna build something to ChatGPT, which every every uses, so that ChatGPT can actually get stuff done for you and, like, maybe feels more human.
是的。
Yeah.
老兄，让我吐槽一下ChatGPT。
Dude, let let me rant about ChatGPT.
请说。
Please.
对。
Yeah.
不知为什么，他们训练模型时，每次对话结束都会说：‘如果你愿意，我也可以帮你做X和Y。’
For some reason, they they trained the model so that at the end of every conversation, it's always, like, if you want, I can also do x and y.
是的。
Yeah.
是的。
Yeah.
老兄，我被这个搞得烦透了，这种没完没了的套路真让人抓狂。
And, dude, I I so got annoyed by it that that kind of churn from chattypreeti.
真的吗？
Oh, really?
是的。
Yeah.
所以这可能提升了他们的指标，但真的特别烦人。
So so it probably increases their metrics, but it's just, like, super annoying.
你为什么不一开始就直接做呢？
It's like, why don't you just do it in the first place?
你现在是搞云的了？
It's like Are you a Cloud guy now?
是的。
Yeah.
我现在是搞云的了。
I'm a Cloud guy now.
但我确实用编码来写代码。
But I I do use codecs to code.
展开剩余字幕（还有 480 条）
是的。
Yeah.
对。
Yeah.
你喜欢Codex吗？
You like Codex?
你更喜欢用Cog编码还是两者都用？
You prefer to cog code or you use both?
我想做点实际的东西时用Codex，放松随便玩的时候也用Codex。
Codex when I wanna try to do something real, and the Codex when I'm just, like, vibe vibing.
是的。
Yeah.
这真的很奇怪。
It's very odd.
不过挺有意思的。
Well, it's interesting.
我认为它们存在于不同的阶段，之间存在某种权衡空间。
I think they live at different points, and there's a sort of space of trade offs.
是的。
Yeah.
而我发现QuadCode和Opus 46则更健谈一些。
Whereas I find QuadCode and Opus four six, it's a little more chatty.
它做了更多假设，但在同步体验中可能更令人愉快。
It makes more assumptions, but it can be more pleasant for a synchronous experience.
对。
Yep.
而Codecs则会深入思考，准确性也更高。
Whereas Codecs, it really thinks hard, and it's more often accurate.
但有时就像在对话中，对方会停顿三分钟来思考。
But sometimes it's sort of like being in a conversation where the other person pauses for three minutes to think.
是的。
Yeah.
所以你没有进入心流状态。
So you don't have the flow state.
对吧？
Right?
很难获得那种心流，就像钟表一样，我前几天还发过推文，说钟表简直像老虎机。
It's it's hard to get a flow like, clock dude, I tweeted other day, like, clock almost like a slot machine.
每次都不太一样。
It's like it has different things each time.
就是啊，100%。
It's just Oh, 100%.
就像
It's like
老虎机。
a slot machine.
看。
Look.
我觉得，如果你还记得我们以前讨论过的旧社交媒体时代的话。
I I do think that if you think remember we were talking about in the old social networking era?
是的。
Yeah.
那是一种可变的奖励机制。
It was variable scheduled rewards.
对吧？
Right?
那就是它的全部魔力所在。
That was the whole magic of it.
比如，你打开你的Facebook动态，偶尔会感觉无聊、无聊。
Like, you open your Facebook feed, and once in a while, it's like boring boring.
天啊。
Oh my god.
这太令人兴奋了。
This is so exciting.
是的。
Yeah.
对。
Yeah.
而且是的。
And Yeah.
编码代理具有完全相同的特性。
The coding agents have the exact same property.
而且，时间也是可变的。
Also, the time is variable.
所以有时候你一秒内就能得到结果。
So sometimes you get something in a second.
有时候则需要五分钟。
Sometimes it takes five minutes.
所以是的。
So Yeah.
在某种程度上，我认为这两者都赋予了它那种赌场般的感觉。
Up to a certain point, I actually think that both of those things give it that casino like feeling.
是的。
Yeah.
还有另一个关于产品策略的显著不同之处，或者也许只是它的运作方式：编码本身是相当直观的。
And the other thing that's very different about the product strategy or maybe it's just the way it works is, like, coding is kinda, like, self explanatory.
是的。
Yeah.
在云代码中，你有大量乱七八糟的东西。
In Cloud Code, you have all this crazy shit.
你有钩子、技能，但你必须进行注入。
You have, like, hooks and, like, skills and But you have to ins.
如果你没有跟上的话
If you're not following
它就会被触发。
puts it.
是的。
Yeah.
如果你不熟悉x，你就完全不知道如何自定义这个东西。
If you're not following x, you have no idea how to customize this thing.
是的。
Yeah.
但一旦你自定义了它，你就会觉得它像是你的一部分。
But once you customize it, you kinda feel like it's part of you.
所以它有点
So it's kinda
它有点难被取代。
it's kinda hard to churn.
这很有趣，因为我自定义了我的，因为我还读了Boris发布的那篇长文。
It's interesting with so I've customized mine because also I read the long thing that Boris put out.
是的。
Yeah.
但我要说，我觉得Cloud Code让我喜欢的很多原因都在于它的内置功能。
But I will say that I think that Cloud Code, a lot of the reasons that I enjoy it are just harness features.
比如，如果你要裁剪一张图片，你必须先把它粘贴到一个文件里，然后把这个文件粘贴到Codex中。
Like, for example, if you cut an image, you have to paste it into a file before pay and then paste that file into Codex.
好的。
Okay.
你不能直接截取屏幕的一部分，截图后就直接粘贴到Codex里，而Cloud Code就可以这么做。
You can't just take a subset of the screen, screenshot it, and then paste it directly into Codex the same way you can with Cloud Code.
真的吗？
Oh, really?
好吧。
Okay.
好的。
Okay.
就像这些小细节一样。
So just like little things like that.
是的
Yeah.
Cloud Code 增加了语音功能。
Cloud Code added voice.
现在还有一点卡顿，但方向是对的。
It's a little bit janky right now, but it's going in the right direction.
所以他们加入了很多提升体验的小功能。
So they've just got a bunch of quality of life things.
是的
Yeah.
Cloud Code 可以与 Cloud 和 Chrome 交互。
Cloud Code speaks to Cloud and Chrome.
好的。
Okay.
而 Codex 无法与 Atlas 交互。
And Codex doesn't speak to Atlas.
明白了。
Got it.
所以我认为这些都是OpenAI会修复的问题。
So I think these are all things that OpenAI will fix.
是的。
Yeah.
我认为Codex实际上是一个更好的模型，但它们今天还不存在。
I think Codex is actually a much better model, but they don't exist today.
对。
Yeah.
对。
Yeah.
他们需要修复它。
They need to fix it.
我的意思是，他们肯定会全力投入Codex。
I mean, they're gonna go all in on Codex, I'm sure.
是的。
Yeah.
谈谈
Talk to
关于编码代理。
me about coding agents.
你的一般看法是什么？
What's your general view?
你认为这是SaaS的终结吗？
Do you think it's the end of SaaS?
你认为这些只是玩具吗？
Do you think these are just a toy?
首先，我不是工程师，我只是个新手。
Well, first of all, I'm, like, not an engineer, so I'm, a novice.
但我听说，前几天我跟一些人聊过，一家原生AI初创公司，他们基本上在尝试，他们有一群Vibe程序员。
But I do hear that like, I was talking to some folks the other day, an AI native start start up, and they're basically trying to they have a bunch of Vibe coders.
嗯。
Mhmm.
所有这些Vibe程序员都在尝试构建内部工具，以替代他们正在付费使用的SaaS产品。
And all the Vibe coders are just trying to build internal tools that replace their SaaS that they're paying for.
真的吗？
Really?
所以这是一家真正这样做的公司吗？
So it's an actual company that's doing this?
这是一家SaaS公司。
It's SaaS company.
是的。
Yeah.
这是一家AI原生公司。
It's AI native company.
它就是一家Vibe编程公司。
It's it's like one of the Vibe coding companies.
这是最受欢迎的公司之一。
It's like one of the more popular ones.
有意思。
Interesting.
是的。
Yeah.
很有趣，哦，我明白了。
It's interest oh, I see.
所以他们实际上是一家AppGen公司。
So they're actually an AppGen company.
他们是AppGen公司，正在支付一堆SaaS费用。
They're AppGen company, they're paying for a bunch of SaaS.
是的。
Yeah.
他们想摆脱这些付费服务。
They wanna, like, get rid of the payment.
他们想通过开发内部工具来实现。
They wanna by coding internal tools.
通过
By
使用，好的。
using okay.
在这种情况下，他们可能是最极端的采用者，因为他们的产品本身就是AppGen，所以应该用AppGen来处理一切。
So in that case, that they might be the most extreme form of adopter because their own product is AppGen, so they should use AppGen for everything.
是的。
Yeah.
不过，我猜你的预测是，普通公司会放弃Slack或者处理掉它？
I guess, is your prediction, though, that the average company will churn off of Slack or deal
或者我不这么认为，我觉得Slack还有很长的生命周期，因为Slack也可以成为你与代理本身沟通的地方。
or I don't think I I I feel like Slack has a lot of legs because Slack can also be the place where you talk to the agents them themselves.
嗯。
Mhmm.
但其他一些工具相当复杂。
But some of other ones, they are pretty complicated.
所以买这种代码之类的东西有点困难。
So it's kinda be hard to buy code, that kind of stuff.
但我觉得，如果你有一个像Calendly这样更简单的应用的话。
But I feel if you have an app like maybe Calendly or something more simple.
是的。
Yeah.
那我为什么要为它付费呢？
Then why should I pay for it?
我只需要
I'll just
那我为什么要为它付费呢？
Why should I pay for it?
不过反过来想，这也没那么贵。
Though the counterpoint is it's not that expensive.
是的。
Yeah.
你真的想自己维护一个Calendly系统吗？
And do you really wanna maintain your own Calendly thing?
是的。
Yeah.
而不是每月支付20美元。
Versus pay $20 a month.
它会一直自动更新。
It always gets updated.
它会一直在线。
It's always up.
是的。
Yeah.
因为组织中每个人为这些事情所投入的精力是有限的。
Because there's just like a fixed amount of capacity that anyone in the organization is gonna have for all this stuff.
是的。
Yeah.
那确实如此。
That that that's true.
除非你雇专门负责氛围营造的人，否则初创公司其实并不只想买些酷炫的东西。
Unless you hire, like, dedicated vibe corkers, like, the startup doesn't actually just wanna buy cool stuff.
但这就涉及成本效益与直接支付Calendly费用的权衡。
But then it's the cost benefit versus just paying for Calendly.
是的。
Yeah.
对。
Yeah.
有趣的是，最近很多人在推特上讨论Figma。
It's interesting about for example, like, a lot of people are tweeting about Figma recently.
是的。
Yeah.
它会让我们陷入困境，你知道的，你真的能挺过去吗？
It'll stock us down and, like, you know, are are are you gonna survive?
是的。
Yeah.
我觉得这个问题还悬而未决。
And I feel like the jury's out there.
很难说清楚。
It's kinda hard to say.
是的。
Yeah.
我觉得所有设计师现在还在用 Figma，但作为设计师，你得学会怎么用 VibeCode。
I I feel like all designers are still on Figma, but as a designer, you kinda need to learn how to do VibeCode.
否则，如果你想做 Figma，那你就会落后。
Otherwise, you're gonna if if you wanna know to do Figma Yeah.
再过几年，你可能会被淘汰。
You're probably gonna be, like, out of date in a couple years.
是的。
Yeah.
我对此的反驳是，我一直在思考工具的分类——思考工具与创作工具。
My counterpoint to that is that I think that I've thought a lot about the sort of thinking tools versus making tools.
对吧？
Right?
是的。
Yeah.
IDE 传统上一直是一种创作工具。
The IDE was historically historically a a making making tool.
工具。
Tool.
嗯。
Mhmm.
它是用于执行的地方。
It's a place for execution.
我认为它正在逐渐远离这种模式。
I think it's migrating away from that.
随着执行变得零成本，我认为这些新一代的多智能体IDE，很多都专注于尝试和利用试错来启发思考。
And now with execution going to zero, I think these sort of, like, multi agent next gen IDEs, a lot of them are about trying things and using the trial and error as a way to inform your thinking.
是的。
Yeah.
对。
Yeah.
很多时候，我会用一种非常原始的方式先实现一个功能，然后不断让编码智能体修改，直到它能运行。
Like, a lot of times, I'll just build a feature in a really naive way, and I'll hammer the coding agent until it works.
嗯。
Mhmm.
然后我会说，嘿。
Then I'll say, hey.
把你认为可以改进的地方都写下来。
Write all the things that you would have done differently.
嗯。
Mhmm.
然后我会回到最初的点，重新做一遍。
And I'll go back to the initial point and redo it.
所以我在想，我认为Figma实际上两者都做。
So I wonder if and I think Figma actually does both.
我认为它是设计执行的地方，但同时也是设计思考的重要场所。
I think it's a place for design execution, but it's also an important place for design thinking.
是的。
Yeah.
对。
Yeah.
我认为这是他们在新架构中保持高度相关性的机会。
And I think that's their opportunity to be highly relevant in the new stack.
是的。
Yeah.
我完全同意。
I totally agree.
我完全同意。
I totally agree.
但我认为16z有类似你们投资铅笔之类的东西吗？
But I think a 16 z has, like, you guys investing pencil or something?
Pencil dot dev？
Pencil dot dev?
Speedrun做过。
Speedrun did.
是的。
Yeah.
对。
Yes.
Speedrun，还有是的。
Speedrun and yeah.
想想看，需要什么来提升这个AI工具的水平，因为看着这些代理与你协作并完成任务，真的非常有趣。
Think think of what needs to, like, level up this AI tooling because, like, watching these agents collaborate with you and do stuff is, like, very interesting.
我知道这正是他们最关注的。
I know it's top of mind for them.
是的。
Yeah.
你认为编码代理最被忽视的能力是什么？
What do you think are the most under discussed capabilities of coding agents?
哪些能力被低估了，又有哪些被过度吹捧了？
What's underhyped and maybe what's overhyped as well?
这可能不算被低估，但我觉得，软件正在吞噬整个世界。
This is probably not underhyped, but, you know, I I feel like an interest in that software will eat the world.
我觉得编程将吞噬所有知识型工作。
I feel like coding will eat all knowledge work.
对吧？
Right?
我们其实已经在朝那个方向前进了。
And we're kind of going that direction already.
比如，我觉得LaBobo今天刚发布了新功能，可以支持一切并制作演示文稿。
Like, I I think LaBobo recently launched, like, today Yeah.
它们能支持所有功能并制作演示文稿。
That they can support everything and can make decks.
是的。
Yeah.
所以，没错。
So so yeah.
我觉得大家都在追赶这个趋势。
So I I and I feel like everyone's chasing chasing this.
而Roblox可能已经领先了。
And and Roblox is probably in the lead.
是的。
Yeah.
我不想再用PowerPoint了。
I I I don't wanna use PowerPoint anymore.
我讨厌一辈子都在写Google文档，老兄。
I don't wanna write a I hate writing Google Docs, dude, across my entire life.
是的。
Yeah.
对。
Yeah.
所以，前几天我写博客文章时，没有直接打字，而是想：嘿。
So so, like but but the other day, was writing my blog post, and instead of just, like, typing it out, I was like, hey.
不如我用Cloud Code，给你一堆反馈，让你帮我写。
Let's let me just use Cloud Code, let me give you bunch of feedback, and you write it for me.
是的。
Yeah.
然后你就不断往下写，它已经完成了前80%。
Then you just keep dropping the it did the first 80%.
最后的20%我说要加点麦芽味进去，用来润色一下内容。
The last 20% I said has a malty go in there, like, to take take stuff.
是的。
Yeah.
但这就是我现在工作的方式。
But, like, that's the way I work now.
我从不从零开始。
I I never start from zero.
我总是让AI先完成前80%。
Like, I always get the first 80% from AI.
对。
Right.
是的。
Yeah.
是的。
Yeah.
这很有趣。
It's interesting.
如果你看一下，其实历史上也有类似的例子。
If you look at there there are also, like, historical analogs of this.
我认为萨提亚说过，Excel 是世界上最有力量或最受欢迎的编程语言。
I think Satya said this, which is that Excel is the most powerful or most popular programming language in the world.
是的。
Yeah.
而且它是一种数以千万计——我的意思是，一亿多人一定都会使用的编程语言，甚至可能更多。
And that it's sort of a programming language that millions and millions I mean, a 100,000,000 plus people must know, maybe even more.
但我们却并不这样看待它。
And yet we don't think of it that way.
它是一种描述和解决问题的方式。
It's a way to sort of describe and solve problems.
是的。
Yeah.
我认为编码代理将会是这样，当然，还要再乘以一千倍。
And I think coding agents are going to be that, of course, times a thousand.
是的。
Yeah.
甚至像写Google文档这样看似主观的事情，也可以在编码领域以某种方式呈现，使得使用代理来完成它更令人满意、更高效、更具杠杆效应。
Where even things that feel subjective, like writing Google Docs, can be represented in the coding domain in such a way that it's more satisfying, more productive, more high leverage to use agents to do it.
对。
Yeah.
因为这听起来很流行，是因为它非常易用。
Because it it sounds like popular because it's super approachable.
对吧？
Right?
是的。
Yeah.
而且，像编码代理这样的东西，编码本身基本上已经消失了。
And, like, coding agents, the coding is basically gone.
它就像是被抽象掉了。
It was, like, abstract away.
你只是在和某个代理对话，然后就能完成各种事情。
You're just talking to some agent and getting to do stuff.
所以是的。
So Yeah.
是的。
Yeah.
没错。
Exactly.
这将会变得非常重要。
It's gonna be huge.
是的。
Yeah.
你认为未来的公司会是什么样子？
What do you think the future company looks like?
它只是由一群代理加上一个CEO组成的吗？
Is it just a bunch of agents with a CEO?
CEO是一个代理吗？
Is the CEO an agent?
我的意思是，未来公司里人的角色是什么？
I mean, what is the role for people in a company in the future?
好吧。
Okay.
嗯，我有一些大胆的看法。
Well, I have some hot takes.
我们曾经一起在几家公司工作过，让我给你一个大胆的看法，老兄。
So we we both worked at, some companies together, and let let me give you a hot take, man.
也许我们把这段剪掉吧。
Maybe we cut this out.
但我觉得，公司越大，就越会变得糟糕，越来越是个糟糕的工作环境，老兄。
But, like, I feel like as a company gets bigger, it tends to get shit it tends to become like a shitty, shittier place to work, dude.
是的
Yeah.
因为人太多了。
Because there's a lot of people.
你得保持一致。
You have to align.
我觉得这是不言而喻的。
I think that's axiomatic.
对。
Yeah.
对吧？
Right?
我记得，也许我该提一下这家公司，但我记得，你知道，我们公司。
And I remember, maybe I should mention this company, but I remember, you know, our company.
我们以前有这么多，
We used to have all these,
比如OKR会议。
like, OKR meetings.
我坐在一个
I'm sitting in a
房间里三个小时讨论OKR。
room for three hours talking about OKRs.
我只是想，老兄，这简直是在浪费我的生命。
I'm just like, dude, this is, like, wasting my life.
是的。
Yeah.
所以我想表达的是，我希望更多公司能保持小型化。
So where where I'm going with this is I hope more companies will stay small.
我认为这一代的创始人已经意识到了这一点。
And I think the founders of this generation realize that.
他们希望尽可能保持小规模。
Like, they wanna stay as small as possible.
是的。
Yeah.
与其组建一个十人的产品团队，不如只保留两到三人的产品团队，再用一堆AI代理来协助你。
And instead of having a 10 person product team, you have a two or three person product team, and just have a bunch of AI agents to help you.
是的。
Yeah.
我认为，用AI代理来推出产品线，比用人类要容易得多。
I I think it's way easier to cross launch your lines with AI agents than with humans.
是的。
Yeah.
是的。
Yeah.
实际上，从某种意义上说，代理之所以好，是因为它去除了情绪因素。
Well, actually, in a in a sense, the agents actually because it takes the emotion out of it too.
你可以想象，如果我派我的代理，你派你的代理去谈判，是吧。
Like, you can imagine if I sent my agent, you sent your agent to go negotiate something Yeah.
他们得出了一个结论，这没有情感因素。
And they came out with some conclusion, it's not emotional.
这不关乎我们任何一方。
It's not for either of us.
这是
It's
非常客观的。
very objective.
是的。
Yeah.
没错。
Exactly.
是的。
Yeah.
这很有趣。
It's funny.
我们一直讨论的一个问题是，从员工体验的角度来看，支持人工智能在工作中应用的理由是什么？
One of the things that we've been talking a bunch about is what is the pro case for AI at work in terms of employee experience?
我认为这就是你所描述的。
And I think it's what you're describing.
对吧？
Right?
比如，你会提升工作的净推荐值吗？
Like, do you increase the NPS of work?
是的。
Yeah.
所以，如果我们从最根本的地方说起，或者广义地说，人类体验的净推荐值。
So if we, like, go all the way back or even broadly the NPS of the human experience.
对吧？
Right?
想象一下，在一万年前，普通人的日常净推荐值仅仅是‘别被吃掉’。
Think of the NPS of the day to day human in 10,000 BC when it just don't get eaten
被狮子吃掉。
by the lion.
是的。
Yeah.
那真是美好的一天。
And that's a good good day.
对吧？
Right?
或者一百年前，能平安无事就不错了。
Or maybe a hundred years ago, it's okay.
别在工厂里被杀掉。
Don't get killed at the factory.
嗯。
Mhmm.
被蒸汽压机压碎之类的。
Crushed by the steam press or whatever else.
而现在，很多情况下就是别卷入到某个副总裁下属的激烈情绪化谈判中。
And and now a lot of it is, like, just don't get sucked into some high emotion sort of negotiation with another VP's subordinate.
是的。
Yeah.
就像扎克发了50条消息的对话来回拉扯。
Like a 50 message of Zach thread going back and forth.
是的。
Yeah.
正是如此。
Which exactly.
然后最终，大家都觉得不想告诉CEO，但最后还是传到他那里了。
And then eventually, everyone's like, I don't wanna tell the CEO, and eventually it goes there.
这简直太糟糕了。
And it's just terrible.
所以也许未来的趋势是，很多这种情绪化、主观性的工作都能被处理掉。
So maybe the future of this is that a lot of that emotional subjective work gets handled.
是的。
Yeah.
我们某种程度上是在引导这个过程，但并不是以一种深入其中、对我们人类来说不适用的方式。
And we're sort of guiding the process, but not in the middle of it in a way that just doesn't suit us as humans.
对。
Yeah.
我曾经过着双重生活，既是产品经理又是创作者，我觉得所有产品经理其实都只是想创造产品。
I lived at double life as a PM creator, and, like, I feel like all the PMs actually just wanna create products.
他们就是想创造产品。
They just wanna create products.
这正是我们当初投身这一行的原因。
Well, that's why we all got into it.
这真的很有意思。
It's it's so interesting.
我的意思是，妮可经常谈到这一点，但每个产品经理心中理想的PM形象都是创新者。
I mean, Nicole talks about this all the time, but, like, every PM's sort of view of the ideal PM is the innovator.
我是想到了这个新点子，
Like, I came up with the new thing,
然后，是的。
and Yes.
我有了一个重大洞察，它解锁了这个产品。
I had the big insight, and it unlocked the product.
是的。
Yeah.
我
I
我认为真相在于，我认为大多数产品经理并不知道如何做到这一点。
I think the black pill is I don't think most PMs know how to do that.
事实上，许多公司中没有任何人具备这种能力。
In fact, many companies have zero people that know how to do that at all in any function.
是的。
Yeah.
所以我认为，产品经理们仍然渴望能够做到这一点，他们要么就去实践并取得成功，要么不成功就转到其他职能。
So it would nonetheless, I think PMs aspire to be able to do that, and they should either do it and either be successful or maybe not successful and move to a different function.
我也觉得我的观点是，我认识的所有产品经理几乎都在晚上和周末偷偷写代码。
I I also feel like my hot take is, like, basically, all the PMs I know are trying to live code at nights on weekends.
是的。
Yeah.
我觉得我的观点是，如果你真的失业了，你反而会有更多时间去当一个创造者，去创新。
And I feel like my hot take is that I feel like if you're actually unemployed, like, you probably have more time to be a builder and, like, to be innovative.
是的。
Yeah.
你可以真正地去尝试各种东西，学习所有这些技能。
You can actually, like, play all this stuff and learn all this stuff.
很多产品经理都在试图
A lot of PMs are trying
或者成为团队中的工程师。
to Or maybe be an engineer in the team.
是的。
Yeah.
我以前是工程师，我不确定是不是被迫转做产品经理的。
I used to be an engineer, and I got sort of I don't know if I got forced to be a PM.
也许我也觉得产品经理看起来地位更高一些，是的。
I maybe I also perceived PM as, like, being a little more high status Yeah.
是的。
Yeah.
我加入谷歌的时候。
When I joined Google.
但后来，我逐渐从另一个角度看待了这件事。
But then eventually, come around the other side.
你会觉得这太糟糕了。
You're like, this is terrible.
是的。
Yeah.
是的。
Yeah.
就像，你几乎从来得不到真正的成就感，除非每季度发布一次时。
Like, you never really get the satisfaction Yeah.
除了每季度发布时，你几乎得不到真正交付产品的满足感。
Of actually shipping other than once a quarter when you ship.
我的意思是，我们所说的PM技能——和用户交流，试图弄清楚该做什么，要解决什么问题——这些都非常重要，是的。
I I mean, the PM skills that we're talking to users and, like, trying to figure out what to do, like, what's the problem to solve, like, those are very important Yeah.
但没错，你得身兼数职，老兄。
But, yeah, you gotta wear multiple hats, dude.
你得亲自去执行，敲代码，获取反馈，然后可能再找工程师一起推进。
You gotta go do the thing yourself, comport type it, and get some feedback, and then maybe brand engineer along.
你觉得
How much do you
你认为每个人都要像加里说的那样，为了STEMI连觉都不睡，跑得那么快吗？
think that everyone has to go as fast as I mean, like, Gary was talking about STEMIs and skipping sleep.
加里正在推动加里坦的G栈。
Gary's putting Gary Tan g stack.
我的意思是，嘿。
I mean, hey.
我的意思是，这难道是我们所有人都必须采用的默认工作方式吗？还是说，你认为深思熟虑会带来某种权衡？
I mean, is that, like, the default way that we all need to work, or do you think there's a trade off for thoughtfulness?
我认为，现在有了这么多AI工具，人们很容易同时朝十个不同的方向推进。
I think it's very easy now with all these AI tools just going, like, 10 different directions at once.
是的。
Yeah.
所以有时候你确实需要放慢脚步，弄清楚自己想往哪里去。
So sometimes you do have to slow down and try to figure out where you wanna go.
是的。
Yeah.
但我同时也相信，那种传统的流程——比如年度规划或者做这些废话——我觉得完全
But I also believe that the traditional process where you, like, do annual planning or doing this bullshit, I just feel like that fully
意识到一个局部最优解
realizing a local a sort of local maxima
嗯嗯。
Mhmm.
你应该快速前进。
You should go very fast.
对吧？
Right?
嗯嗯。
Mhmm.
所以假设你像爬山一样，到达了新的局部最优解的底部。
So let's say you kinda hill climb, you get to the bottom of the new local maxima.
嗯嗯。
Mhmm.
我认为
I think
使用智能体，你应该能够非常快地到达那个高峰。
with agents, you should be able to get to the top of that hill extremely fast.
对吧？
Right?
你有
You have
获得一个新的洞察，围绕这个洞察构建一切，使其得到充分表达。
a new insight, build everything around the insight so it's fully expressed.
但我觉得，要到达下一个这样的高峰，应该要有快慢结合的方式。
But then I think to get to the next the next sort of hill should have, like, fast and slow.
这可能是未来的方式。
That's probably the future way.
是的。
Yeah.
我也这么认为。
I think so.
而且你得进行这种随机游走，去寻找一个艰难的市场契合点，这需要很长时间。
And you've to go that random walk trying to find a hard market fit, which takes a while.
对吧？
Right?
所以这并不是对。
So this is not Yeah.
在我们开始录音之前，我们之前聊过一些‘即用型’商业平台。
So we were talking before we started recording about some of the business in a box platforms.
你有看过它们吗？
Have you looked at them?
你有什么看法吗？
Do you have a view?
我看过
I've looked at
我们之前讨论过的那些帖子。
posts yet that we talked about.
我不确定这个人是故意把它做成与AI口号相反的，还是
I don't know if the guy like, intentionally made it the opposite of AI slog or is it
有点
kind of
你故意我
a you intentionally I
我觉得是。
think so.
是的。
Yes.
是的。
Yes.
是的。
Yes.
对。
Yeah.
这很有趣。
That's funny.
我的意思是，我有相当大的公众影响力。
Well, I mean, I have a pretty big public presence.
对吧？
Right?
所以我把所有东西都和它关联起来。
So I connect all my shit to to it.
而且说实话，它确实很好地展示了可能实现的东西，但目前来看，可能仍然处于早期阶段。
And then I mean, it definitely gives a good peek into what's possible, but, like, right now, it's probably still pretty, like, early stage.
现在是该投放Facebook广告的时候了。
Like, it's time to run, like, Facebook ads.
是的。
Yeah.
我为什么要投放Facebook广告？
Why am I running Facebook ads?
是的。
Yeah.
所以我不确定。
So I don't know.
是的。
Yeah.
是的。
Yeah.
我的意思是，我对这个非常兴奋，因为它确实为更多人创建公司提供了一条路径。
I mean, I'm very excited about it because it does feel like it's a path for more people to build companies.
是的。
Yeah.
即使他们只是一个人的公司。
Even if they're single one person companies.
是的。
Yeah.
如果你想想打造一家十亿美元企业有多难，支持它的市场，以及试图这么做的人数，嗯。
If you think about how competitive it is to build a billion dollar business, like the markets that support it, the number of people trying Mhmm.
相比之下，一亿、一千万和十万美金的总可用市场。
Versus 100,000,000 versus 10,000,000 versus $100,000 TAM.
是的。
Yeah.
也许全国各地、全世界各地都存在一些机会，可以开发出总可用市场为十万美元的产品。
Like, maybe there are these pockets all over the country, all over the world where there are opportunities for $100,000 TAM products.
是的。
Yeah.
这足以改变某人的生活。
And that would change somebody's life.
这虽然不是一家由风险投资支持的企业，但也没关系。
Now that's not an enterprise venture backed company, but that's okay.
是的。
Yeah.
所以我希望整个理论能够成立，因为我确实认为这是一种让更多人参与的方式。
So I hope that whole thesis works because I do think it's a way to get more people to participate.
是的。
Yeah.
这就是我为我孩子制定的计划，老兄。
That that that's my plan for my kids, dude.
我想在高中时就自己搞点小生意。
Like, I I wanted to just build, like, boot shop businesses in high school.
是的。
Yeah.
他们可以跳过整个大学和职场生活。
And they can skip the whole college and corporate life.
是的。
Yeah.
老兄，我觉得过去十年里，人们对孩子想当YouTuber这件事一直有种道德恐慌。
Well, dude, I think this is for ten years, there's this moral panic about the kids wanna be YouTubers.
是的。
Yeah.
你是个YouTuber，是的。
You're a YouTuber Yeah.
是的。
Yeah.
类似于MrBeast那种风格。
In the in the vein of MrBeast.
我认为支持这一点的理由实际上是孩子们想成为创业者或拥有自主权。
I think the pro case for that actually is that the kids wanted to be entrepreneurs or have agency.
对于那些不是程序员的人来说，唯一的渠道至少在网上就是制作YouTube视频。
And the only channel for people if they weren't programmers was creating YouTube videos, at least online.
是的。
Yeah.
所以如果你是网络原生一代，想创造点什么，又不是程序员，就会做一个YouTube节目。
So if you're like an online native generation, you wanna create something, you're not a programmer, you make a YouTube show.
嗯。
Mhmm.
现在你能做的远不止这些。
Now you can make a lot more than that.
是的。
Yeah.
你可以打造任何你想要的东西。
You can build whatever you want.
没错。
Exactly.
就是没错。
So Exactly.
这会非常令人兴奋。
It'd be very exciting.
还有其他大胆的观点要分享吗？
Any other hot takes for us?
实际上，我想了解一下你对这个问题的看法。
I'm curious about your thoughts about this, actually.
所以我觉得很多人说，智能代理会先与你的产品互动。
So I feel like a lot of people are saying agents will interact with your product first.
对吧？
Right?
而且他们看到这么多优秀公司都在构建 API 和 MCPs。
And and they see all these great companies, like, building, like, APIs and MCPs.
嗯。
Mhmm.
但你是怎么想的呢？你已经作为用户使用了一段时间。
But, like, how do you think about you've been a consumer for a while.
所以，作为消费者，你得让用户愿意回来使用你的产品。
So, like, the consumer is you gotta get the user to come back and use your product.
对吧？
Right?
是的。
Yeah.
但现在用户会说：嘿。
But now the user is like, hey.
去派个代理去用它吧。
Go send the agent to use it.
那你如何看待留存率和这些基本问题呢？
So how do you think about retention and all this basic stuff?
比如，品牌价值呢？因为代理只是在调用某个ABR。
Like, how do or even, like, brand equity because the agent is just, like, pulling some ABR.
是的。
Yeah.
我不太确定。
I don't okay.
我们必须采用间接盈利模式。
We had to have indirect monetization.
嗯嗯。
Mhmm.
好的。
Okay.
比如，我们从来不会直接向消费者收取这些产品的费用，所以才有了广告之类的模式。
Like, we just can we're never charging consumers directly for these products, which is why you got Got ads and stuff.
广告。
Ad.
然后是大规模的网络，我们都痴迷于留存、参与度和大客户。
Then just large scale networks, and we all obsessed with retention and engagement and whales.
所有这些都非常重要，因为我们并没有直接向用户收费。
And all of these things really mattered because we didn't simply charge people for products.
嗯嗯。
Mhmm.
所以我认为，在人工智能时代，真正有帮助的一点是，消费者现在对尝试新事物充满热情。
So I think one big thing that's actually really helped in the AI era with that is that consumers are now excited to try new things.
他们愿意付费。
They're willing to pay.
他们愿意支付很高的价格。
They're willing to pay a really high price point.
消费者首次出现了消费收入。
There's also consumption revenue in consumer for the first time.
比如代币之类的。
Like token and stuff.
是的。
Yeah.
比如代币。
For, like, tokens.
你既有订阅费，又有代币收入。
You have your subscription plus your token.
而实际上，一个意想不到的好处是，确实也存在真实成本。
So and then the actual, like, the the sort of blessing in disguise is that there are real costs as well.
你有这些推理成本。
You have these inference costs.
是的。
Yep.
所以你会想，哇。
So you're like, wow.
我们必须在第一天就向客户收费。
We have to charge our customer on day one.
嗯。
Mhmm.
所以我认为，商业模式的简化将极大地帮助解决你所描述的许多问题。
So I think one thing is that, like, the business model simplification, I think, will really help with a lot of what you're describing.
嗯。
Mhmm.
第二，我认为许多产品都会提供一个API接口，供你的代理进行交互，或用于交易类的操作。
Two, I think that a lot of the products will have a sort of it'll have an API interface for your agents to interact with or for for transactional sort of wrote things.
是的
Yeah.
对
Yeah.
而且还会有一个基于使用量的接口。
And then it'll have a consumption based interface as well.
嗯
Mhmm.
所以你可以想象一个移动应用，里面有信息流，但你可以切换到显示底层工作流的界面。
So you can also imagine a like a mobile app where there's, like, the feed, but then you can kinda turn it over to where the wires are.
你可以直接要求完成某些任务，或者查看已完成任务的日志。
And you can just ask for things to get done or you can just see the log of the things that got done.
是的
Yeah.
也许人们会同时使用这两种方式。
And maybe people would do both.
对吧？
Right?
我的意思是，你可以想象我们在工作的Credit Karma。
I mean, you can imagine Credit Karma where we work.
有时候，你只想看看你的信用评分历史和一些其他的信用卡优惠。
Like, once in a while, you wanna just take a look at your score history and a few other maybe credit card offers.
我不知道。
I don't know.
我的意思是，是的。
I mean Yeah.
是的。
Yeah.
如果我能看到我的评分以及各种信用卡优惠，我肯定会这么做。
If I can get my score with all kind of credit card offers, I'll definitely do that.
是的。
Yeah.
关于 Bayt 播客
Bayt 提供中文+原文双语音频和字幕，帮助你打破语言障碍，轻松听懂全球优质播客。