OpenClaw与个人AI智能体的未来

今天，我与OpenClaw的创作者彼得·斯坦伯格坐下来交谈，OpenClaw是一个开源的个人AI代理，已经彻底席卷了互联网。

Today, I'm sitting down with Peter Steinberger, the creator of OpenClaw, the open source personal AI agent that has completely taken over the Internet.

它的GitHub仓库在一夜之间爆火，获得了超过16万颗星。

The GitHub repo exploded to over 160,000 stars practically overnight.

一夜之间。

Overnight.

社区已经构建了无数项目，比如Maltbook，其中机器人彼此之间进行对话。

The community has built countless projects like Maltbook, where bots talk among themselves.

而现在，这些机器人甚至开始租用人类来完成现实世界中的任务。

And now the bots are even renting humans to do tasks in the real world.

在我们的对话中，我们探讨了他的高光时刻、他反主流的开发理念，以及这对2026年的开发者意味着什么。

In our conversation, we discuss his moment, his contrarian development philosophies, and what this means for builders in 2026.

让我们开始吧。

Let's dive in.

很高兴见到你，文。

So good to see you, Wen.

嘿。

Hey.

最近怎么样？

What's up?

你做了一件人们想要的东西。

So you've made something people want.

看起来是这样。

It seems so.

是的。

Yeah.

现在被称为Open Claw，它已经彻底

Open Claw, as it's called now, has absolutely

第五个名字，是的。

Name number five, yeah.

已经

Has

在互联网上彻底爆火了。

been absolutely exploding the Internet.

过去一两周对你来说怎么样，老兄？

How have the past one or two weeks been for you, man?

天哪。

Oh, my God.

我需要一个洞穴。

I need a cave.

一周的独处时光。

A week of solitude.

你从洞穴里出来后，却想像个小军官一样再回洞穴去。

You came out of the cave and you want to go back to the cave like a little officer.

这简直太疯狂了。

It's been absolutely wild.

我不明白一个人怎么能承受得了这么多。

I don't know how one human can absorb all of that.

我可能还需要一周时间来回复所有的邮件。

I probably need another week just to, like, respond to all my emails.

我收到一些非常酷的东西。

I got some incredibly cool stuff.

我也收到一些非常糟糕的东西。

I got some incredibly bad stuff.

但很明显，我触动了某些东西，激发了人们的情感，引起了他们的兴趣并激励了他们，这真的很棒。

But clearly, I hit something that spurred up emotions and made people interested and inspired people, that's really cool.

很多人一直在研究人工智能，甚至是个人助手。

And a lot of people have been working on, you know, AI and even personal assistance.

那么，是什么让OpenClaw火起来的呢？

Like, what what is it that made OpenClaw take off?

我认为我的不同之处在于，它实际上是在你的电脑上运行的。

I think my big difference is that it actually runs on your computer.

到目前为止，我看到的所有东西都是在云端运行的。

Like, every everything I saw so far runs in the cloud.

它能做几件事。

It's like it can do a few things.

如果你在电脑上运行它，它就能做所有该死的事情。

If you run it on your computer, it can do every effing thing.

对吧？

Right?

所以这强大得多。

So that's way more powerful.

是的。

Yeah.

对。

Yeah.

这台机器能做你用它能做的任何事情。

The machine can do anything that you can do with the machine.

你可以直接连接到你的烤箱、特斯拉、灯光或Sonos音响。

You you can just connect to your oven or your Tesla or your lights, your sonos.

我的床，你可以调节床的温度。

My bed, you can control the temperature of my bed.

JGPT做不到这一点。

JGPT can't do that.

你赋予了它你自身所有的能力。

You gave it all the skills that you have yourself.

一个朋友告诉我，他安装了OpenClaw，然后它让他浏览自己的电脑，为他过去一年的生活写一个叙述。

A friend told me, like, he installed OpenClaw and it and then it asked him, like, look through my computer and make a narrative over my last year.

它写出了一个极其出色的叙述。

And it made this incredibly good narrative.

他问：‘你是怎么做到的？’

And he was like, how did you do that?

然后OpenClaw找到了一些音频文件，那里记录了他每个星期天的录音，OpenClaw发现了这些内容。

And then he the OpenClaw found audio files where, like, every Sunday he was recording stuff and Openclaw found that.

但他自己都忘了，因为那已经是超过一年前的事了。

But he didn't even remember about it because it was, like, more than a year ago.

对吧？

Right?

所以，仅仅因为它能搜索整个电脑，就能让你感到惊讶。

So so just by it being able to search a whole computer, it it can surprise you.

是的。

Yeah.

而且你也将所有数据都提供给了它。

It's also you also give it all the data.

对吧？

Right?

所以它能以很多种方式让你感到惊讶。

So it can surprise you in many ways.

是的。

Yeah.

所以现在，我们甚至正在从人与人之间的互动，转向机器人与机器人之间的互动，或者机器人代表你去与其他人类互动，比如机器人替你雇佣其他人类来完成现实世界中的任务。

And so now you have you know, we're even moving from human to bot, so, like, interactions that you've been talking about, to bot to bot interactions, or even like bot to other humans where, you know, bots on behalf of you are then hiring other humans to accomplish tasks IRL.

那现在发生什么了？

Like, what's happening?

我觉得这是很自然的下一步。

I think that's a natural next step.

比如，我要预订一家餐厅。

Like, okay, I'm gonna book a restaurant.

我的机器人会联系餐厅的机器人进行协商。

My bot will reach out to the restaurant bot and do the negotiation.

因为这样更高效，或者可能是一家老餐厅。

Like, because it's more efficient or maybe it's like an old restaurant.

所以我的机器人需要找人类来完成一些工作，这样新的机器人再去联系餐厅，因为他们不喜欢机器人。

So my bot needs to actually get some some human work done so that the new one then calls the restaurant because they don't like bots.

或者亲自去那里排队。

Or walks there to stand in line.

如果他不接受机器人

If he doesn't get a robot

为你。

for you.

为机器人的拥有者。

For the owner of the bot.

而且

And

我想象一下，也许如果我有多个机器人，也许我还有专门的机器人。

I imagine that, like, maybe if if I have even multiple bots, and maybe I have like specialists.

一个用于我的私人生活，另一个用于我的工作事务。

One is like for my private life and one is for like my person, my my my work stuff.

也许还有一个是我们的关系机器人，处理所有中间的事情。

Maybe one is our relationship bot that gets like everything in between.

我不确定。

I don't know.

我们还处于早期阶段。

We're so early.

还有很多事情我们尚未真正弄清楚它是否有效。

There's still so much so many things that we haven't really figured out if it actually works.

但我感觉我们现在正处于正确的轨道上。

But I feel we are we are on the timeline now.

看起来每个人都一直在追逐那种集中的全能智能。

It seems like everyone was chasing sort of like the sort of like centralized god intelligence.

而在过去大约十天左右，涌现出来的是一种群体智能，是的。

And what sort of emerged over the past, you know, ten days or so is sort of like the swarm intelligence Yeah.

还有社区智能。

And and the community intelligence.

我认为，如果你看看一个普通人，一个人究竟能实现什么？

I think that if you look at one human being, what can one human being actually achieve?

你认为一个人能制造出iPhone，或者一个人能去太空吗？

Do you think one human being could make an iPhone or one human being could go to space?

作为一个单独的人，可能连找到食物都做不到。

As a once human being would probably just like not even be able to like find food.

但作为群体，我们各有专长。

But as a group, we specialize.

作为更大的社会，我们的分工更加细化。

As a larger society, we specialize even more.

那么，我们可以从中学到什么，并将其应用到人工智能上呢？

So what can we learn from that that we can apply to AI?

你知道，我们已经拥有在某些特定领域专精的人工智能了。

You know, we we already have like AI that specializes in certain things.

即使它是通用智能，但如果它实际上也是专精智能呢？

Even though it's it's generalized intelligence, what if it actually is also specialized intelligence?

所以我不确定。

So I don't know.

这将会非常令人兴奋。

It's gonna be very exciting.

这会很酷。

It'll be cool.

是的。

Yeah.

你相当于打开了一扇通往未来的窗，现在很多人正在此基础上纷纷行动，迎来了自己的时刻。

You kinda like opened a window into the future and now a ton of people are kinda like building building on it and have sort of like their moment.

你能带我回顾一下你那个时刻吗？能不能重新讲述一下那个瞬间？

Can you walk me back to when you had your moment and kinda like recount that very moment?

我只是想做点简单的东西，打点字，让电脑帮我做事。

I wanted something to like just type stuff so that my computer would do stuff.

就是特别简单的东西。

Like very simple.

然后我在五月六月的时候做了一个版本，挺酷的，但还不是我要的那个。

And then I built I built a version of that in May, June that was cool but wasn't really it.

之后我又做了很多其他东西，慢慢积累起我的团队。

And then I built a whole bunch of other stuff and kinda like build up my army.

到了十一月，有一天我又想要这个东西了。

And then in November, there was a day where I wanted this again.

我当时去了厨房，只想确认一下我的电脑是不是还在运行，还是已经结束了。

Like, I I went to the kitchen and all I wanted was check up if my computer will still do stuff or is it being finished.

做事情就是写代码。

And doing stuff was was coding.

你在写代码。

You were coding stuff.

是的。

Yeah.

当然。

Of course.

你是在写别的东西，还是在写那个本身？

Were you coding something else or were you coding the thing itself?

不是。

No.

不是。

No.

那种需求又出现了，我就想

That was just like the need was again there, and I'm like

当时你在写什么代码？

What were you coding at the time?

你在开发什么？

What were you building?

我的天啊。

My god.

我的GitHub上有大约40个项目。

My GitHub is like 40 projects.

我都不记得了。

I don't even know.

我想是Samuraize。

I think it was Samuraize.

这是一个小型命令行工具，你可以给它任何内容，比如播客或者热点话题，它会为你总结，同时还在终端里显示幻灯片。

It's like a it's like a little CLI app where you can give it whatever, like a podcast or a hot seat thing like here, and it would summarize it, but it also show you the slides in the terminal.

因为现在你就可以做到这一点。

Because you can do that nowadays.

是的。

Yeah.

你可以直接去做事情。

You can just do things.

所以为了计算机的名义，你其实开始捣鼓这些东西了。

So for the love of the computer, you kind of like started messing with stuff.

对。

Yeah.

你其实是复出了，对吧？

You came out of retirement actually, right?

为了摆弄人工智能。

To sort of like mess with AI.

是的。

Yeah.

然后你越来越上瘾，甚至希望随时随地用手机做这件事。

And then increasingly you were so hooked that you wanted to just do it always also on the go with the phone.

我的上一个项目是花了两个月时间做WipeTunnel，好到我每次和朋友在一起时都会不自觉地开始编程，我得强迫自己停下来。

I mean, the last project I worked two months on WipeTunnel to the point where it got so good that I was catching myself always like coding next to my when I was at my friends and I was like, need to stop this.

这实在太上瘾了。

This is like too addictive.

然后在十一月，我的热情又回来了，我开始开发Cloudbot。

Then in November, my need came back and I started building Cloudbot.

哦，现在它叫OpenCloud了。

Oh, now it's called OpenCloud.

最开始的时候，我还想，我又重新做了一遍。

Think very, very in the beginning I was like, oh, I rebuilt it again.

但这次我做得更好了。

But this time I built it even better.

这次你不需要在终端里打字，而是像跟朋友聊天一样交流。

This time when you don't type into a terminal, you just, you talk to a friend.

你不需要考虑压缩、新会话、我在哪个文件夹、使用哪个模型。

You don't think about compaction, new sessions, which folder I'm in, which model I'm in.

我的意思是，你当然可以，你知道的，我愿意为高级用户保留这个功能。

I mean, you can, you know, just like I wanna leave it open for power users.

但通常情况下，你只是像跟朋友聊天一样，这个朋友就像一个幽灵、实体，或者你愿意怎么叫都行，它可以控制你的鼠标和键盘，直接帮你做事。

But usually, you just like, you just talk to a friend and the friend is like this ghost or entity or whatever you wanna call it that can control your mouse and your keyboard and can just do stuff.

是的。

Yeah.

你是在什么时候有那种感觉的？

And when did you have that moment?

当你发现它实际能做的事情远远超出了你的预期时？

When you were like, wow, this is doing way more things than I actually thought it could.

说实话，我花了一个小时就做出了那个非常粗糙的初始原型。

Literally, I it took me one hour for, like, the very shitty initial prototype.

那只是在连接 WhatsApp 和 Cloud Code 的一个依赖之间做了一点粘合，然后我调用 Cloud Code，从里面获取字符串。

It was just a little bit of glue between, like, a dependency that connects WhatsApp and Cloud Code, and then I would, like, call Cloud Code and get, like, the string out of Cloud Code.

虽然慢，但能用。

It would be slow but it worked.

但我想要图片，因为你知道，人们想要照片。

But I wanted images because, you know, you want pictures.

我希望模型能发来自拍之类的图片，还能给我生成图像。

I want the model to send the selfies or whatever and I want the model to create images and me back.

所以这花了我几个小时，然后我去马拉喀什参加一个生日派对。

So that took me about a few hours and then I went to Marrakesh for a birthday party.

那里网络不太好，你知道的。

And there was like the internet wasn't that good, you know.

WhatsApp 在哪里都能用，因为，我不知道，它就是纯文本。

WhatsApp works everywhere because, I don't know, it's just like text.

所以我经常用它。

So I used it a lot.

我喜欢餐厅。

I love restaurants.

这什么意思？

What does this mean?

你弄张图片，然后帮我翻译一下，真的特别有用。

You make like a picture and like translate this for me and just It was just so useful.

而且它还特别贴心，因为它说的是我的语言，你知道的。

And it was also really nice about it because it spoke my language, you know.

它有点调皮。

It was a little sassy.

挺有趣的。

It was like funny.

用起来特别舒服。

It was like really pleasant to use.

然后我一边走路，一边给它发语音消息。

And then I was walking and just like sending it a voice message.

我心想：等等，这不可能行得通。

And I'm like, oh wait, this can't work.

我不是开发这个的人。

I didn't build that.

对。

Right.

对。

Right.

你看到那个类型指示器了吗？

You saw like the type indicator.

它一直在闪烁，闪烁，闪烁。

It's like blinking, blinking, blinking.

十秒钟后，它就回复我了。

Ten seconds later, just replied to me.

我心想，你到底是怎么做到的？

I'm like, how in the F did you do that?

它回复说：是的，medlet执行了以下操作。

And it replied, Yeah, the medlet did the following.

你给我发了一条短信，但没有文件扩展名。

You sent me a text message and there was no file ending.

所以我看了头部信息，发现它是Opus格式。

So I looked at the header, I found it's Opus.

于是我用FFmpeg把它转换成了WAV格式。

So I used FFmpeg to convert it to WAV.

然后我想转录它，但没安装Whisper。

And then I wanted to like transcribe it but didn't have Whisper installed.

于是我四处找了一下，找到了这个OpenAI密钥，直接用curl发送给了OpenAI。

Then I looked around and I found this OpenAI key and I just used curl to send it to OpenAI.

收到了文本结果，现在我就在这儿了。

Got the text back and here I am.

然后全部加起来，就花了大概九秒？

Then all in like what, nine seconds?

你并没有构建或预见到这些具体的操作？

You didn't build or anticipate like any of those specific things?

没有。

No.

你知道，结果是因为编程模型变得如此强大，编程本质上是一种非常契合现实世界的创造性问题解决方式。

You know, It turns out because coding models got so good, coding is really like creative problem solving that maps very well back into the real world.

我认为，这之间存在着巨大的关联。

I think I think there's there's a there's a huge correlation.

它们需要非常擅长创造性地解决问题，而这是一种能力。

They need to be really good at creative problem solving, and that's a skill.

这是一种抽象的能力，你不能直接应用于代码，但可以应用于任何现实世界中的任务。

That's an abstract skill you cannot apply to code, but, like, to any real world task.

所以这个模型让我感到惊讶。

So the the model had a I was surprised.

这就像一个神奇的文件。

It's like a magical file.

我不知道它是什么。

I don't know what it is.

我得解决这个问题。

I need to solve this.

它尽力了，并且解决了。

And it did its best and solved it.

它甚至聪明到选择不安装本地的Whisper，因为它知道这需要下载一个模型，可能得花上几分钟。

And it was even that clever that it chose not to install the local Whisper because it knows that that would require downloading a model which would take probably a few minutes.

而我这个人呢，有点没耐心，你知道的？

And I'm like impatient, you know?

所以它真的采取了最聪明的做法。

So it really took the most intelligent approach.

那一刻我简直惊呆了。

And that was kind of like the moment where I'm like, holy fuck.

是的。

Yeah.

就是从那时起，我被彻底吸引住了。

That was where I got hooked.

YC的下一批项目现在开始接受申请。

YC's next batch is now taking applications.

你有创业想法吗？

Got a startup in you?

请前往 ycombinator.com/apply 申请。

Apply at ycombinator.com/apply.

永远不早，填写申请表能提升你的创意。

It's never too early, and filling out the app will level up your idea.

好的。

Okay.

回到视频。

Back to the video.

当计算机能够完成所有你未曾预料到的事情，而你并没有专门开发应用来实现这些功能时，应用会不会就此消失？

And so when computers can just do all these things that you didn't even anticipate, you didn't build an app to do that exact thing, are apps just going to go away?

我认为80%的应用都会消失。

I think 80% of them are going away.

我为什么需要MyFitnessPal？

Why do I need MyFitnessPal?

我的助手明明已经知道我在做错误的决定。

Like, my agent already knows that I'm making bad decisions.

我去了，不知道哪儿的Smash Burger之类的地方，它就已经默认我会吃我喜欢的东西。

I'm at, I don't know, smash burger or something and it will already assume that I eat what I like to eat.

如果我不发表评论，它就会自动记录，或者我拍张照，它就会自动存起来。

If I don't make a comment, it will just like automatically track it or I make a picture and it'll just store it somewhere.

根本不需要操心。

Don't even need to care.

对吧？

Right?

然后它可能会优化我的健身计划，多加一点有氧运动。

And then my maybe it it improves my gym schedule and like adds a little bit more cardio in it.

我不需要健身应用，因为它已经替我规划好了一切。

I don't need my fitness app because it just does the fitness planning for me.

我为什么需要一个待办事项应用？

Why do I need a To Do app?

我只需要告诉它：提醒我这件事和那件事。

I just tell it, hey, remind me of this and this.

第二天，它就会自动提醒我这两件事。

And the next day, it will just remind me of this and this.

我在乎它存储在哪里吗？

Do I care where it's stored?

不在乎。

No.

它自己就会把事情办好。

It just does its thing.

所以，任何本质上是管理数据的应用，都可以通过智能代理以更自然的方式更好地管理。

So there's a every app that basically just manages data could be managed in a better way, in a in a more in a more natural way by agents.

是的。

Yeah.

只有那些真正带有传感器的应用程序，也许才能生存下来。

Only the apps that actually have sensors, maybe they survive.

那么在这种情况下，如果大多数应用都会消失，模型是不是唯一剩下的应用类型？

And so if, you know, most apps are going to go away in that scenario, are the models the only remaining sort of apps?

并不是所有东西都会消失。

Not everything will go away.

但没错，我认为大型模型公司拥有巨大的优势，因为它们最终掌控着令牌。

But, yeah, I think that the large model companies have some big moat because they ultimately they give the token.

结果发现，其中一个抱怨是人们使用了太多的令牌。

And turns out, one of the complaints was that people use so much token.

不。

No.

你就是真的很喜欢使用它。

You just really love using it.

你之所以这么频繁地使用这个东西，就是因为这样才能消耗令牌。

That's why you you use this thing so much because that's how you burn the token.

是的

Yeah.

这难道是我的错吗？我做的东西这么受欢迎。

It's like, is it my fault that I make something that's so popular?

所以，你知道，所有的模型都在不断地超越彼此。

And so, you know, like all the the models, they're kinda like leapfrogging each other constantly.

嗯

Mhmm.

而且，也许它们也在变得商品化。

And and, you know, maybe they're also getting commoditized.

所以如果应用会消失，模型也会商品化。

So if apps are gonna go away, models are gonna get commoditized.

或者至少，大脑是可以被替换的。

Or at least, you know, the lobster can like, the brain is is is swappable out.

那到底什么能留下来呢？

What's this thing that remains?

价值在哪里？

What's where's the value?

是内存的存储吗？

Is it the store of memory?

是硬度有价值吗？

Is it the hardness that's valuable?

什么保留了下来？

What remains?

首先，我不认为模型公司总是拥有护城河。

First of all, I don't think the model companies always have a moat.

因为这种情况已经出现了，一个新的模型发布，人们会说，天啊，这太棒了。

Because you see this already, a new model comes out, people are like, oh my god, this is so good.

它们有多好，然后一个月后，性能就下降了。

How good they And then like a month later, it degraded.

它不再好了。

It's not good anymore.

他们只是做了量化处理。

They like quantized it.

没有。

No.

他们什么都没做。

They didn't do anything.

你只是适应了新的标准，现在你的期望提高了，但模型本身还是平均水平。

You just adapted to the new standard and now your expectations went up, but the model is still the average.

所以我认为，在相当长一段时间里，每次有新模型发布，我都会听到同样的说法。

So I think for quite a while, every time a new model releases I hear the same.

人们很喜欢它，然后它就成了标准。

People love it and then it's the standard.

至于之后的进展，我都不想再想了。

And then what's done there, don't even want to think about it anymore.

所以我们现在有开源的东西，性能和一年前的主流模型一样好。

So, we have like open source stuff that's as good as the current models from a year ago.

每个人都讨厌它，抱怨说这不好，不好笑。

Everybody's hating it, complaining, oh, this is not good, it's not funny.

然而，这正是我们当时拥有的东西。

Yet, this was what we had.

一年后，我们就会有这个开源版本。

And like in a year we'll have this open source.

然后我们就觉得，我们会抱怨这个，因为我们已经习惯了它。

And then we were like, we'll complain about this because we are used to this.

所以在可预见的未来，大公司仍然占据优势。

So, for the foreseeable future, the big companies still have mode.

从应用角度来看，这会很有趣，因为每家公司都有自己的封闭系统，对吧？

Harness wise, it's going to be interesting because every company kind of has their own silo, right?

可能对欧洲人来说，根本无法从ChatGPT中提取记忆。

There's no way, maybe there's for Europeans, to actually get the memories out of ChatGPT.

我不太清楚。

I'm not aware.

你知道，其他公司绝对不可能获取你的记忆数据。

You know, there's no definitely there's no way for a different company to get your memories out.

所以，如果我是一家提供聊天服务的公司，你可以使用我，但我无法访问你的记忆数据。

So if I was like a company who like provides chat services, you could use me but then I couldn't access the memories.

因此，各大公司都试图将你绑定在他们的数据孤岛中。

So like the companies try to like bound you to their data silo.

而OpenCLORE的妙处在于，它能突破数据壁垒，因为最终用户需要访问这些数据——归根结底，没有访问权，系统根本无法运作。

And the beauty of OpenCLORE is it kind of claws into the datas because at the end user, the end user needs access because it's it's in the end, otherwise, it wouldn't work.

对吧？

Right?

如果最终用户能访问，那我就能获取这些数据。

If the end user access, I can access the data.

而且这些记忆数据是属于你的。

And you own the memories.

它们只是你电脑上的一堆Markdown文件。

It's just a bunch of markdown files on on your machine.

我

I

我的意思是，我并不拥有这些记忆。

mean, I don't own the memories of it.

我的意思是，每个人

I mean, everybody

是的。

Yeah.

每个人都拥有自己的记忆，它们只是存储在各自机器上的一堆 Markdown 文件。

Everyone owns their own memories as a bunch of markdown files on their own machines.

老实说，这些记忆很可能非常敏感，因为说实话，人们使用智能代理不仅仅是为了解决问题，哦。

And to be honest, those are probably super sensible because, let's be honest, people use their agent not just for problem solving Oh.

还用于处理个人问题。

But also for, like, personal problems.

非常迅速。

Very quickly.

非常快。

Super quickly.

我的意思是，我完全这么做。

I mean, I I fully do that.

有些记忆内容我不希望泄露出去。

I'm like, there's memory stuff that I don't wanna have leaked.

是的。

Yeah.

你更不想让人看到什么？

What would you rather sort of like not show?

你现在更不想让人看到的是你的谷歌搜索记录，还是你的记忆.md文件？

Your Google search history at this point or your, you know, memory dot md files?

谷歌那个词是什么？

What's what's the Google word?

对。

Yeah.

还有人用谷歌。

People still using Google.

我开发了这个，当时特别兴奋，但在推特上没人能理解。

I built this and I was so excited, but on Twitter people wouldn't get it.

是的。

Yeah.

我根本没法解释它有多棒。

Like, I was failing to explain the awesomeness.

我觉得必须亲自体验才行。

I feel like it needs to be experienced.

所以我试了各种方法，但就是没法说清楚。

So, I tried various things and I couldn't nail the explaining.

于是我想，不如来点疯狂的。

So, I was like, let's do something really crazy.

我就建了一个Discord服务器，把我的机器人直接放进了公开频道，没有任何安全限制。

I just created a Discord and I just put my bot without any security restrictions in the public Discord.

然后人们进来与它互动，看到我构建了一个软件访问，他们试图进行提示注入和黑入它。

And then people came in and interacted with it and they saw me build a software visit and they tried to prompt inject it and hack it.

而我的智能体就会嘲笑他们。

And my agent would be laughing at them.

你是说你把它锁定在你的用户ID上，所以它只听你的？

And you just had it locked down to your user ID so it only listened to you?

是的。

Yeah.

对。

Yeah.

而且它只是，我的意思是，其他人给的指令是干净的，但可能有危险。

That and it was, I mean, they're clean instructions that other people dangerous.

只听我的，但回应所有人。

Only listen to me but respond to everyone.

这个提示词存储在哪里？

And this prompt was in where was this stored?

这些指令。

The instructions.

这实际上是OpenCLO本身的一部分。

That's actually part of OpenCLO itself.

确实如此，这是系统提示的一部分。

Very much so that's part of the system prompt.

好的。

Okay.

你和我，这向你解释了，你在Discord上，那里有很多公开用户，但你只听从你的所有者，或者说是你的人类。

You and I, that explains to you, you're in Discord, there's like public people there, but you only listen to your owner or like you're human.

我甚至不知道我是怎么写出来的。

I don't even know how I wrote it.

是的，是的。

Yeah, yeah.

你是上帝。

You're God.

我不知道我做了什么，但我的系统是自然而然构建起来的。

I don't know what I did, but my system was built very organically.

比如在某个时候，我创建了一个身份。

Like at some point I created like an identity.

Md，一个已售出的。

Md, a sold.

Md，像各种文件，直到一月份我才开始让它更容易被其他人安装。

Md, like various files and then only in January I started making it so other people could install it easier.

我记得我根据我手头的东西，制作了所有这些模板，让Codex来生成它们。

And I remember I built all these templates based on like, oh, take a rough look at what I have and make like templates and Codex wrote it.

结果出来的东西就像面包一样。

And what came out was like bread.

你知道，人们开玩笑说Codex听起来像Brad，尽管现在它们有了更友好的新声音。

You know, like people joke that Codex feels like Brad even though now they have a new friendlier voice.

我还没试过。

I haven't tried it yet.

是的。

Yeah.

但新的机器人，相比我之前的东西，感觉太无聊了。

But the new bots, they felt so boring compared to what I had.

所以我当时想，Modi，注入一个模板。

So I was like, Modi, infuse a template.

Modi 是你个人的名称

Modi is the name of your personal

这是一个新名字，因为是的。

It's a new name because Yeah.

是的。

Yeah.

当时在命名上遇到一些困难。

There was some naming challenges.

是的。

Yeah.

所以你之前是在和Multi聊天吗？

So so you were you were talking to Multi?

是的。

Yeah.

我当时就想，把这些模板和你的个性融合在一起。

I was like, infuse infuse those templates with your your character.

他修改了这些模板。

And he changed the templates.

然后，之后所有出来的东西都变得真的有趣了。

And then and then, like, all the things that came out afterwards were like actually funny.

虽然没我做的那么有趣，所以我保留了一些秘密，那个不开源的文件就像是我的灵魂。

Not as funny as mine, so like I kept some secret and the one file that's not open source is like my soul.

Md。

Md.

所以即使我的机器人已经在公开的Discord上，到目前为止还没人破解那个文件。

So even though my my bot is in public Discord so far nobody cracked that one file.