彼得·斯坦伯格：“乐趣即速度”

六个月前，我还是未来的希望。

Six months ago, I was the future.

现在，脉搏说我被一个动漫女孩嘲笑了。

These days, the pulse say I got mocked by an anime girl.

你知道，我几乎每天都会收到这样的问题。

You know, I get questions like this almost every day.

通常，我会忽略它们。

Usually, I ignore it.

或者我会给出官方回答。

Or I give the media answer.

今天，你会听到一些真实的回答。

Today, you get some real ones.

一共五个问题，四十分钟。

It's gonna be five questions, forty minutes.

我知道你们可能都是冲着这个来的。

And I know what you what you might all came for.

你们来是为了循环、图表，让cortex飞速运转。

You came for loops, graphs, making cortex go brr.

这个在我的推特上是免费的。

This is free on my Twitter.

这会帮你们挺过可能到来的过山车般的起伏。

This will help you to survive the roller coaster that might be coming for you.

你知道，昨天鲍里斯告诉你们让模型自己发挥。

You know, yesterday Boris told you to let the model cook.

而我，就是那个发现当几万人同时让它发挥时会发生什么的人。

I'm the guy who found out what happened when tens of thousands of people let it cook at the same time.

我们先从每个人都会问的第一个问题开始。

Let's start with the question everyone asks first.

OpenClaw怎么才八个月大？

How is OpenClaw only eight months old?

你知道，按人类时间算可能是八个月，但按AI时间算，更像是四年。

You know it might be eight months in in human time, but in AI times, it's it's more like four years.

我的灵感来源通常是被惹恼的时候。

And my source of inspiration is usually being annoyed.

所以十一月一个下雨天，我一边忙着处理一些智能体，一边又饿得不行。

So on on a rainy day in November, I was juggling some agents, but I was also hungry.

所以我想确保在我去厨房找吃的的时候，我的代币能被好好利用。

So I wanna make sure that my tokens are put to good use while I raid the kitchen.

但当时还是没有好办法能从手机直接发指令给电脑。

And there was still no good way to just send a prompt from my phone to my computer.

这样某个智能体就能去看看我的智能体到底怎么样了。

So some agent could check out how my agent's actually doing.

还记得2025年初吗？

Remember early two thousand twenty five?

你们知道吗，这是我第一次做无声演讲。

And you know guys, this is my first silent discourse.

这对我来说是个全新的体验。

It's it's a very new experience.

所以，如果你真的在听我说，就给我鼓个掌吧。

So, like, me some hands if you actually listen to me.

是的。

Yes.

太神奇了。

Amazing.

真是神奇。

Amazing.

你知道吗，2025年初，天哪，我们说的是老的OPUS 3，Shamini 2.5。

You know, early two thousand twenty five, oh my god, we're talking the old OPUS three, Shamini 2.5.

打开OPUS 3，那是最令人惊艳的一款，但也往往又慢又贵。

Open the I hat o three, which was the most impressive one, but also really slow and expensive at times.

而且每当我的agent做对事时，我都会有那种多巴胺上头的快感。

And I got a dopamine hit when my agent did something right.

我敢肯定你记得那段时光。

I'm sure you remember this time.

而现在，如果它们出了什么错，我通常会先反省自己，你知道吗。

And these days, if I if they get something wrong, I usually question myself, you know.

是不是我设计的循环有问题？

Did I design the loop wrong?

是不是我没有给我的代理提供足够的验证条件？

Did I not give my agent everything they need to verify their work?

是不是我的思路有漏洞？

Do I have an error in my thinking?

我是不是在要求不可能的事情？

Am I asking for impossible things?

所以当我从厨房回来，发现我的编程代理因为一些愚蠢的事情停了下来，我很恼火。

So I came back from the kitchen and my coding agent stopped because of something silly, and I was annoyed.

于是我打开了一个新的终端会话，把我的想法一股脑儿输进去，让模型自己发挥。

So I spun up a new terminal session, I rambled my idea in, and I let the model cook.

它帮我建了一个WhatsApp中继，一个小时后，我就能从我的Mac上发送和接收WhatsApp消息了。

It built me a WhatsApp relay, and in an hour later, I could send messages from my Mac to WhatsApp and back.

这本身感觉像魔法一样，但其实不应该这么神奇。

And that in itself felt magical, but really it shouldn't have.

就是说，我们几个月前就已经这么做了，对吧？

It's like, we all did this for months already, right?

我们打开终端，输入一些内容

We opened the terminal and you type something in

然后就能得到回复。

and you get a reply.

但神奇的是那种感觉。

But the magic is how it felt.

它不像一个终端。

It wasn't a terminal.

它写出的答案很简洁。

It wrote concise answers.

它很主动。

It was proactive.

有时候它甚至会在白天主动来关心我。

Like sometimes it would just like check on me during the day.

我让它把复杂性都消解掉了。

I made it so the complexity would melt away.

你不需要考虑用哪个模型、上下文多大，或者什么时候开新会话。

You don't have to think about which model, which context size, or when you start a new session.

我还把权重稍微调离了默认分布，让它感觉更像朋友。

And I I prompted the weights to a little bit outside the default distribution, make it feel more like a friend.

然后我就在用它。

And I was using it.

我有很多次都觉得，这就是未来。

I had I had so many moments where I felt like this is the future.

这就是通用人工智能。

This is this is AGI.

你知道吗？

And you know what?

没人关心。

Nobody cared.

那时候我在推特上已经有不少粉丝了。

Back then I already had a good following on Twitter.

我知道现在它叫X了。

I know it's called x now.

但我永远会叫它推特。

I don't I'll always call it Twitter.

大家就是不明白。

People were not getting it.

我一直在试着让他们理解，这对我来说有多神奇，但一周又一周，我就是没能做到。

I would keep trying to like make them make them understand like how magical this felt to me, but week after week I failed.

你注意到一个规律了吗？

And you notice a pattern?

这让我很烦。

It annoyed me.

所以我创建了群聊，把朋友加进去，通过WhatsApp转发给他们看。

So I would create group chats with friends, and I added them WhatsApp relay, and I would show them.

我会让他们跟它聊天。

I would let them talk to it.

每次我都能得到强烈的情绪反应。

And every time I got a strong emotional reaction.

有些人很惊讶。

Some people were amazed.

有些人甚至害怕或吓到了。

Some people were scared or freaked out even.

但每次都有强烈的情绪。

But each time, there was a strong emotion.

最棒的是，有不少人真的很想要它。

And the best part, there were quite a few people that really wanted it.

特别是对我那些非技术朋友，我告诉他们，不，这个还不是给你们用的。

And like, especially to my non technical friends, I told them, no, this is not yet for you.

他们就很生气。

And they got mad.

所以如果这都不算找到了产品市场契合点，那我就不知道什么才算。

So if that's not an indicator to have product market fit, I don't know what is.

所以我又花了一个月琢磨细节，想着，我能做些什么来解释这个世界呢？

So I spent another month tweaking the details, thinking, what could I do to explain the world?

你知道吗？

You know?

然后让人没想到的是，有个人直接给我发了一个拉取请求，要给WhatsApp转信息功能加上Discord支持。

And then the audacity, someone sent me a pull request to my WhatsApp relay to add Discord support.

你知道吗？

You know?

就像，这个名字里的哪部分你没搞明白？

Like, what parts of the name are you not getting?

我先让那个拉取请求放了一段时间，然后一直在想，一直在想。

I let the PR sit for a while, and I was thinking and I was thinking.

然后最后，我想，管他呢？

And then eventually, was like, ah, what the hell?

然后我们的中继就变成了Claudius，因为我起名很在行，而且消息渠道不止一个。

And out of our relay became Claudius, because I'm good with names, and there's no more than one message channel.

所以你能明白这有多早了吧。

And so you understand how early this was.

我甚至还没内置压缩功能。

I didn't even had compaction built in yet.

所以有时候它就会直接停掉，然后为了赶上Discord，我加了一个非常粗糙的压缩版本，好让Mario在Pi上做一个好的。

So like at some point it would just like stop, and just just for in time for Discord, I I added a very hacky version of Compaction, so Mario could make a good one in Pi.

所以那是在新年前后，就像你常做的，我提前回家继续捣鼓它。

So it was around New Year's Eve, and as you do, I I went home early to hack on it some more.

然后在一月的第一周，就在极客圈集体学习编码代理的时候，我创建了一个Discord房间、服务器、公会，不管我们怎么叫它，我把我的爪子放了进去。

And then in the first week of January, just as the geek world collectively was learning about coding agents, I created a Discord room, server, guild, however we call it, and I put my claw in it.

我清楚地记得第一晚。

And I remember the the first night vividly.

人们会加入Discord，然后他们说，看我公开构建它。

People would be joining Discord, and they were like, watch me build it in the open.

他们会试图破解它。

They would try to hack it.

他们就像在跟它对话。

They were like talking to it.

他们得到了聪明的回复，然后看看他们能做什么。

They were getting smart replies, and they see what they can do.

他们终于明白了，

And they finally got it,

你知道吗？

you know?

就像，这就是那个时刻。

Like, this was this was the moment.

我整夜没睡。

I sit up all night.

我让别人跟它互动。

I let people interact with it.

我在智能体里放了这么一段提示词。

I had this prompt in my agents.

那段 Markdown 提示词会告诉它：除非是 Peter 的指令，否则不要调用任何危险工具。

Md that would instruct it, don't do any dangerous tool calls unless the prompt comes from Peter.

你知道，大概六个月前的事。

You know, this was like six months ago.

那时候我们在 HSMD 里写的东西都更像是一种建议。

Where all the things we wrote in our HSMD felt more like suggestions.

所以我盯得很紧。

So I watched it very carefully.

我随时都能把电源拔掉，然后到了早上七点，我终于搞定了。

I could always pull the plug, and then at 7AM, I finally I was done.

我跟它道了晚安，按了 Ctrl+C，然后就上床睡觉了。

I said goodnight, I pressed control c, and I went to bed.

但当然，我可以把软件做得有弹性，所以它是个启动守护进程。

But of course, I could build the software to be resilient, so it was a launch daemon.

你知道如果你对启动守护进程按control c会发生什么吗？

You know what happens if you like press control c on the launch daemon?

对，它只会死掉五秒钟，然后又会重新启动。

Yeah, it'll just it'll be dead for five seconds and it'll start up again.

所以当我走进卧室时，我的代理就高高兴兴地开始回应世界上的上帝了。

So while I was walking into the bedroom, my agent happily started answering God in the world.

然后我睡了大概十个小时。

And then I slept for like ten hours.

当我醒来时，我收到了大约800条消息。

When I woke up, I woke up to around 800 messages.

人们试图黑掉它。

People tried to hack it.

我拔掉了插头，我吓坏了。

I pulled the plug, I freaked out.

我仔细看了一遍，其实什么都没发生。

I read through everything, and nothing actually happened.

但就是那一刻，那一刻它火了。

But this was the moment, this was the moment where it went viral.

你们都在这里，都知道接下来发生了什么。

And you all been here, you know what happened next.

我的收件箱爆了，记者半夜给我打电话。

My inbox exploded, reporters tried to call me in the middle of the night.

你知道苹果有个功能吗？就是当你开启勿扰模式，但有人连续打几次电话时，

Did you know about this feature that Apple built when you put it on do not disturb, but people call you a few times in a row?

系统会认为这是紧急情况，还是会接通电话。

That Ivan would think it's an emergency and would list would still let calls through.

我之前不知道，但记者们知道。

I didn't know about it, but reporters did.

我的邮箱就像瀑布一样涌进来。

My email inbox was more like a waterfall.

是的。

Yeah.

Mac mini 卖光了。

The Mac mini sold out.

我觉得现在还是缺货。

I think they're still sold out.

我一个月收到的播客邀请，比我之前三十九年加起来都多。

I got more invites to podcast in a month than I did in my thirty nine years before.

然后 Tropic 发了封邮件要求改名，还把龙虾图案给删了。

And Tropic sent an email demanding the name change and, like, dropped the lobster.

这个项目也经历了好几次蜕变。

And the project molted a few times.

Clawdius 变成了 ClaudeBot，中间还有很短一段时间我们就不提那个名字了，后来就在 OpenClaw 上发展起来了。

Claudius became ClaudeBot, then there was a very short time where we don't talk about the name, and then it started on OpenClaw.

Jensen 说这是人类历史上最成功的开源项目。

Jensen called it the most successful open source project in the history of humanity.

与此同时，我的收件箱里塞满了记者、安全人员，还有无数好奇的真人以及AI代理。

And meanwhile, my inbox was flooded with reporters, security people, and countless curious humans and also agents.

说实话，这些数字对我来说还是感觉不太真实。

You know, the numbers still don't feel real to me.

短短八个月里，有一万八千多不同的人提交了issue或者pull request。

In eight months, more than 18,000 different people opened an issue or a pull request.

总数超过十一万。

Over a 111,000 in total.

我算过这笔账，你知道的，我到底在骗谁呢？

I did the math, you know, where am I lying to?

是我的经纪人帮我算的。

My agent did the math.

这个仓库里几乎有3000个人提交了代码。

Almost 3,000 people have commits in the repo.

有些人对我很生气，因为他们觉得明明是他们先做的，是我偷了他们的点子，还给我发了一些我从来没见过的东西链接。

And some people were mad at me because clearly they did it first and I stole the idea and they sent me links of something I've never seen.

有些人叫我AI耶稣。

Some people call me AI Jesus.

到了夏天，我就成了反基督。

For summer, I became the antichrist.

后来我开始收集那些反对者。

At some point, I started collecting the Obotarios.

每隔几周就会冒出一个新的。

There's a new one every few weeks.

最近的一个其实是昨天的。

The the most recent one is actually from yesterday.

我们从中能学到什么？

What do we learn here?

许愿要小心。

Be careful what you wish for.

我还没准备好接受这么多关注。

I was not ready for all this attention.

这件事几乎让我崩溃。

It almost broke me.

我那时候差点就把整个东西都删掉了。

Like, I was that close of just deleting the whole thing.

我都不回朋友消息了。

I stopped answering my friends.

我连手机都不想看了，因为消息一直弹个不停。

I didn't even wanna look at my phone anymore because it was just a stream.

当然，还有人泄露了我的电话号码和很多私人信息，因为我现在成了全民公敌。

And of course, somebody leaked my phone number and also a lot of private details because I'm an enemy of humanity.

而我呢，本来只是想搞点酷炫的东西。

And here I am just wanting to build something cool.

现在我们进入第二个问题。

And now we're getting to question two.

你妥协了吗？

Did you sell out?

你知道，那可不是我第一次坐车。

You know, wasn't my first ride.

我二十多岁和三十多岁那阵子，都在做B2B软件公司。

I spent my my twenties and my thirties building a b to b software company.

我像个原始人一样，纯手写了一个PDF框架。

I wrote a PDF framework by hand, like an animal.

我白手起家把公司做起来，做到快八十个人，我压根没把对手放眼里，它慢慢变成了大多数企业都想要的东西，后来我就把担子交给了我联合创始人。

I bootstrapped the company, I grew it to almost 80 people, I ignored the competition, it became the thing most enterprises would buy, and eventually I passed things on to my co founder.

我卖了股份，人也彻底累垮了。

I sold my shares and I very badly burned out.

退休。

Retirement.

这个想法听起来特棒，然后我差不多花了三年时间到处晃荡，补上生活里错过的那些事儿。

The idea of it sounds great, and I spent almost three years wandering around aimlessly, catching up on life and parties.

我把以前在生活里错过的那些事儿，全都干了一遍。

I did all the things that I kinda missed out on life.

有几个月我连电脑都没打开过，只是像普通人一样 casually 用手机刷刷网。

I had months where I didn't even open my computer, know, just casually checking the the Internet with my phone, like a normie.

我一直知道，只要给自己足够的时间，那种冲动会重新回来。

I always knew that if I if I give myself enough time, the urge the urge would come back.

我一直以为那股冲动会是写代码。

I always thought the urge would be code.

我一直都很热爱编程。

I always loved programming.

但我花了一年时间才明白

But it took me a year to understand that

这并不完全

that's not entirely

正确。

correct.

我热爱创造。

I loved building.

编程只是实现目的的一种手段。

Programming was just a means to an end.

而当我重新找回热情，八个月后我又回到了这里。

And here I was, eight months after I found my spark again.

我的收件箱里塞满了风投求着给我钱的消息。

And my inbox was full of VCs begging me to take their money.

而我当时并不想要。

And I did not want to.

所以当那些大型实验室开始找上我，我突然和马克、萨姆还有几个人通上了电话，这件事看起来就有趣多了。

So when the big labs came knocking, and suddenly I was on the phone with Mark, Sam, a few others, this seemed like a much more interesting pass.

同时，这也让我感到很不真实。

It also felt unreal.

对吧？

Right?

说到冒充者综合征，就是这样。

Talk about impostor syndrome.

也许这对你来说是个重要的教训。

Maybe that's an important lesson for you.

你构建的任何东西都可以被分叉或克隆，但你的名字不能。

Everything that everything you can build can be forked or cloned, but your name cannot.

所以你的个人品牌比你将来做的任何单一产品都重要得多。

So your personal brand is way more important than any single product that you ever will.

在需要之前就开始打造它。

Start working on that before you need it.

所以，是的，我知道怎么玩这个游戏。

So, yeah, I know how to play the game.

我也知道这样做相当于出卖灵魂。

I also know that doing this was the equivalent of sellingmysoul.md.

懂的都懂。

If you know, you know.

这个决定并不容易，但如果我在倦怠之后学到了什么，那就是相信你的直觉。

The decision was not easy, but I if I learned anything during my time after the burnout, it's it's this, it's trusting your gut.

而我的直觉告诉我，我最喜欢的是开放空气。

And my gut told me that I like open air the most.

爱马仕赢了吗？

Did Hermes win?

简单来说，它在我们最痛的地方给了我们一击。

And the short answer I can give you is, it beat us where it hurt.

我们还在那里。

We're still there.

这两件事都是我的错，也是我造成的。

And both of those are my fault and my doing.

发布后的几个月里，我们被安全报告彻底压垮了。

In the months after the release, we got absolutely crushed by security reports.

某种程度上，我们成了许多开源项目现在经历的模板。

In some ways, wrote a prototype of what many open source projects now experience.

尽管大多数都是非常边缘的情况，我还是感受到了压力。

And even though most were super edge casey, I felt the pressure.

然后媒体就说，说我们20%的技能都是恶意的。

And then the press, The press say 20% of our skills are malicious.

其实我们为此还专门写了份报告。

Now, we actually bought a paper up at this.

我们算了算，真实数据大概是0.3%。

We did the numbers, and the numbers are more like 0.3%.

我们扫了全部67000个。

We scanned all 67,000.

我们发了，把报告发出去了。

We shipped the we shipped the paper.

你知道的，更正永远没恐慌传得快。

And you know, a correction never travels as far as a scare.

我深感责任重大。

I felt this responsibility.

全世界都发现了OpenClaw。

The world discovered OpenClaw.

尽管有那个很大的吓人的免责声明，实际上比我在幻灯片上放的还要大，但我知道很多人安装时不会去读文档。

And even though there was this big scary disclaimer, actually much bigger than I could on the slides, when you install it, I knew many people wouldn't read the docs.

所以我真的专注于加固代码库，一层又一层地加强安全，加了沙箱、白名单，还有带权限的网页协议。

So I really focused on that, hardening the code base, building layers and layers of security, added sandboxing, allow lists, a web protocol with permissions built in.

我们有些文件操作会调用Python，因为TypeScript里缺少一些基本功能，这样才能确保你的代理不会离开工作区，不会跟着符号链接走，配置文件也是原子写入的。

We were shelling out to Python for some file operations, because there was some primitives that are simply lacking in TypeScript to ensure your agent stays in the workspace, it doesn't follow symlinks, and that the configuration files are written atomically.

大多数用户对此毫不在意。

Most of the users didn't care a bit about that.

当然，他们喜欢安全这个抽象概念，但实际来说，我更新后破坏了他们依赖的东西，让东西变慢了，也让更新变得更难。

Sure they, they liked the abstract term of security, but in all practical terms, they updated, I broke something they depended on, I made things slower, and I made things more difficult to update.

还有一件事我得承认，我变得马虎了，因为我的时间被分散在开源项目、媒体、花大量时间和律师打电话设立一个501(c)(3)美国非营利组织。

And another thing I have to admit, I got sloppy because now I had to my time shifted between open source, the press, spending countless times on the phone with lawyers to set up a a five zero one three American nonprofit.

我自己也做了一个。

I worked on my own one.

而且，你知道，OpenAI有自己的世界，既有趣又要求高。

And, you know, OpenAI has its own world that is interesting and demanding.

我找了帮手，我们组建了一个非常棒的维护者社区，他们都很受尊敬。

I brought on help, and we got such an amazing community of maintainers that are very much respected.

每个人都添加了自己的小功能。

And everyone added their own little feature.

我觉得自己有点为难。

I felt like I'm a bit in a hard place.

毕竟，这些人都是免费工作的，对吧？

After all, these people work for free, right?

那我算什么，我有什么资格告诉他们该做什么？

So who I am, who I am to tell them what to do?

而且我的注意力也很分散。

And also my attention was all over.

所以，我们添加了很多功能。

So, we added a lot of features.

你知道，功能是最好玩的部分。

You know, features are the fun part.

新功能其实就是个提示词的事。

A new feature is just a prompt away.

真正的成本还在后面。

The real cost comes after.

我们每发布一个功能，当然都会带一个配置选项，因为我们不想打乱大家的现有设置。

Every feature we shipped, of course, with a configuration option, because we didn't wanna break everyone's setup.

在我们最忙的时候，我不得不去统计，如果把所有排列组合都算上，我们最后大概有九千五百个配置选项。

At our highest, I had to count that, we ended up with around nine and a half thousand configuration options, if you count all the permutations.

你想写多少测试就写多少。

You can write all the tests you want.

想覆盖所有情况，还保证不出问题，那是不可能的。

It is impossible to cover all of these and not break things from time to time.

而要让软件在用户手里不断演进，那就更是难上加难。

It is infinitely harder to evolve software that has users.

相比之下，其他公司呢，它们都靠风投的钱在推动。

Meanwhile, other companies, they were fueled by VC money.

他们继续往前冲了。

They pushed ahead.

那个动漫女孩公司做得特别好。

The Anime Girl company did it especially well.

人们开始在推特上几乎任何对话里插一脚，而我们却深陷在安全基础工作和功能开发里。

People started inserting themselves into pretty much any conversation on Twitter while we were burned in security ground work and features.

他们的故事更简单，营销攻势很猛，还有一句口号就能把用户拉过去。

They had a simpler story, an aggressive marketing campaign, and a one liner to migrate people's claws.

不过真正让项目受重伤的是Anthropic。

The main thing though that really hurt the project was Anthropic.

不是名字的问题，那部分虽然压力大，但我能理解，而且他们态度很好。

Not the name, the part was stressful, but I kinda understood, and they were really nice about it.

真正的问题是我在模型上优化过头了。

It was the fact that I optimized too much on the model.

我用Codex和GPT构建了OpenClaw，但很长一段时间里，那个框架其实最适合Opus，也是为它优化的。

I built OpenClaw with Codex and GPT, but the harness for a long time really worked the best and was optimized for Opus.

所以当他们提前大约二十四小时通知我要为所有人停用订阅时，真的没有足够的时间来调整方向。

So when they pinged me with around twenty four hours notice that they're gonna disable a subscription for everyone, there was not really enough time to change course.

我的意思是，当然，支持开放权重模型。

I mean, sure, support open rate models.

我们在这些模型上做了很多工作，但它们当时确实还不够好。

We did a lot of work on them, but they really weren't that great yet.

而且早期的OpenAI模型就是缺乏个性。

And the early OpenAI models simply lacked character.

所以，也许把这一点记下来。

So maybe write this one down.

你的依赖项的商业模型就是你的商业模型。

Your dependencies business model is your business model.

你知道，这就是为什么现在解决得这么棒。

You know that's how fixed now solves pretty awesome.

开放权重模型其实很好。

Open weight models are actually good.

我在工程方面学到了很多，但很多地方人们已经往前走了。

I learned a lot about harness engineering, but in many ways people moved on.

你可以从我们的下载图表里看到这点。

You can see that in our download charts.

五月份我们的周下载量触底，大概83.5万左右。

We bottomed out at around 835,000 weekly downloads in May.

然后六月份被宣告死亡之后，又飙到了470万，达到史上最高。

And then after being declared death in in June, we peaked at 4,700,000, the highest ever.

这两件事同时都是事实。

Both of those are true at the same time.

炒作就像天气一样。

A hype is like the weather.

你也许能看到它来临，却控制不了。

You might see it coming, but you can't control it.

就我而言，那就是一场风暴。

In my case, it was a storm.

第四个问题，做这件事还有意思吗？

Question four, is it still fun?

你知道，大概在二月份的时候，它就不再有趣了。

You know, somewhere around February, it stopped being fun.

我开始觉得，它变成了一种责任。

I started to feel like, it started to feel like a responsibility.

我醒来，发现自己这个不想再创业的人，却同时干着两份工作，或者说是一份工作加一个使命，我真的很挣扎。

I woke up, and the man who didn't want to build another company found himself in a situation where he had two jobs, or should I say a job in a calling, I was struggling.

最糟糕的是，我连自己的产品都不用了。

And worst of all, I stopped using my own product.

那段时间，我不再做自己热爱的东西，而是为了所有人去做。

Somewhere in that time I stopped making a product I love, and I worked on making something for everyone.

它不再是我每天都会用的东西，反而更像是我觉得是工作、感觉是工作的东西。

It became less of a thing that I use every day, and more I think that I saw and felt was work.

我想知道，大家到底是谁。

And I wanted to picture who everyone really is.

其中一个人做了个购物助手，另一个人还想对我的机器人搞社交工程攻击。

One of them had a grocery agent, another one tried to social engineer my bot.

在大家都忙着加功能、做组织协调的同时，我变成了那个修bug、处理安全问题、提供支持、打基础的人。

Between everyone adding features and all the organization work, I became the person that fixed the bugs, fixed the security issues, and provided support and built a foundation.

而且因为大家都在传，哦，OpenClaw不是OpenAI的，我就不太敢从OpenAI那边拿太多帮助。

And because all these people were like giving the rumor mill, oh, OpenClaw is not owned by OpenAI, I didn't wanna take too much help from OpenAI.

我是说，对，他们确实给了我用例，我也用得很猛，但同时我也被其他项目拖住了，还有一大堆我没法轻易拒绝的人想来找我聊。

I mean, yeah, they gave me token, and I, boy, did I use that, but I was also pulled into other projects, and there was a whole flood of people that I couldn't as easily block who wanted to talk to me.

现在回头看，我很多事情其实可以做得不一样的。

In hindsight, I could have many done differently.

比如，可以多找点人帮忙。

Like, could have asked for more help.

我也可以把一些职责交出去，但我当时陷得太深了，根本没时间静下来好好做战略思考。

I could have moved responsibility off my plate, but I was so deep in everything that I didn't take time to think through strategically.

不过幸运的是，我遇到了一群很棒的人，慢慢地，事情也就走上正轨了。

Luckily, met a bunch of amazing people, and little by little things aligned.

我搞定了签证，后来成立了非营利组织，我们吸引了一些很棒的公司作为捐赠者，还找到了一些真正相信开源的好人，开始和我一起工作。

I figured out my visa, eventually formed a nonprofit, we got amazing companies as donors, and I found some really good people that believe in open source and started working with me.

我还得特别感谢一下NVIDIA，因为他们很早就参与进来了，他们只是问了我需要什么，然后就派人来接管了大部分的安全工作。

I also need to give NVIDIA a special shout out, because they were very early, and they simply asked me what I need, and then they sent people to take over much of their security work.

我记得大概是在五月份，我生日前后，我有了一种感觉，事情又开始变得顺利了，那种建造的乐趣又回来了。

And I think it was somewhere in May, around my birthday, where I had this feeling that things are starting to feel good again, where the joy of building for me was coming back.

而且，是的，媒体那边，每隔一周就会炒作一个所谓的“OpenClaw杀手”。

And, yeah, somewhere in the press, the press decoyed an open open Claw killer every other week.

有一阵子，我数了数，大概有20个这样的项目。

At some point, think I counted 20 of them.

甚至真的有一个项目叫“OpenClaw杀手”。

There's even a literally a project called OpenClaw Killer.

那其实是个卸载程序，完全没必要，因为我们自己有卸载程序，但你知道的。

It's an uninstaller, which is totally unnecessary because we have an uninstaller, but you know.

但这些所谓的“杀手”故事，没有一个真正讲清楚这到底是怎么回事。

But none of the killer stories ever picked up what this is actually about.

开源。

Open source.

我的灵感来源是感到烦心。

My source of inspiration is being annoyed.

现在，当我不得不使用那些不能直接给代理发指令去修改的软件时，我就会很恼火。

These days, I get annoyed when I have to use software where I can't just send a prompt to my agent to change it.

这就是那种乐趣回来了的感觉。

That's the fun part that's coming back.

很难跟一个纯粹在那里玩得开心的人竞争。

It's hard to compete with someone who's just there having fun.

乐趣就是速度。

Fun is velocity.

我享受搭建的那些周，产品明显变得更好了。

The weeks I enjoyed building, the product got visibly better.

我不享受的那些周，我们发布了配置选项。

The weeks I didn't, we shipped config options.

第五个问题。

Question five.

接下来是什么？

What's next?

有些观众可能会想，是不是终于要讲图了？

Some people in the audience might be like, is this finally the part where he talks about graphs?

首先，我对我们基金会目前的状况非常满意。

First of all, I'm super happy with where we are with the foundation.

我们的使命是让人们更接近AI，而事情变化得太快了。

Our mission is to bring people closer to AI, and things are changing so fast.

对很多人来说，这让人感到害怕。

For many people, it feels scary.

我为OpenClaw取得的一项成就感到自豪。

I'm proud of one thing that OpenClaw achieved.

那就是，对很多人来说，它把AI从那种模糊而可怕的东西，变成了有趣又奇怪的东西，你知道，就像龙虾什么的。

And that for many people, it moved AI from this thing that's like nebulous and scary into something that is fun and weird, you know, lobsters and all of that.

他们会继续推进。

And they'll keep pushing there.

持续打造优秀的开源软件生态，通过活动把大家聚在一起，并加强教育推广。

Keep building a great ecosystem of open source software with events that bring people together and with education.

现在我们团队有十个人。

We have 10 people now on payroll.

我们正在招聘几个新岗位，包括CEO。

We're hiring a few more roles, including a CEO.

其次，我好像把 Claw 变成了一个名词。

Second of all, I kinda made Claw a noun.

你知道的，连 Cup Party 都在公开场合提到了它。

You know, cup party dropped the open.

萨蒂亚在微软的主题演讲里也提到了企业级的 Claw。

Satya says enterprise grade Claws in Microsoft's Keynote.

现在已经有三万三千个以 Claw 命名的仓库了。

There are 33,000 Claw named repositories.

如果我们真的是在模拟世界里，那肯定是那个最奇怪、而且绝对不会被关掉的模拟。

If we are in a simulation, we are certainly in one of the weirder ones that won't get shut down.

我很喜欢未来的这个部分。

I dig this part of the future.

第三，我们还没有一个始终在线、始终同步的智能体。

Third, we still don't have an agent that's always on, always syncing.

还有太多事要做了。

There's still so much to do.

AI的格局和技术变化，比我们围绕它构建的软件还要快，这对你们所有人来说都是机会。

The landscape and the tech in AI are evolving faster than the software that we build around, and that's an opportunity for all of you.

我们的工作流程也在不断演进。

Our workflows are also evolving.

就像我早先的愿景，我们不应该再考虑会话或压缩这类东西了。

Like the vision I had early on, where we shouldn't think about session or compaction.

这个世界，这些模型和技术，终于发展到了可以把这个愿景变成现实的阶段。

The world is finally, the models, tech is finally getting to a stage where this is becoming a reality.

我们正在进入一个世界，终于从纯文本界面转向语音和多模态交互。

We're entering a world where we finally move on from the text, text only interfaces to voice and multimodality.

昨天我们刚搞通了一个功能，你的Claw现在能给你打FaceTime了。

Just yesterday, we got a hack working that your Claw can now FaceTime you.

这就是OpenClaw存在的意义所在。

And this is kinda what matters why OpenClaw exists at all.

每个实验室都会卖给你一个智能体。

Every lab will sell you an agent.

OpenClaw就是那个替代方案。

OpenClaw is the alternative.

开源，随处运行，适配任何模型。

Open source, runs everywhere, works with any model.

如果你用本地模型，你的数据就永远不用离开你的设备。

And if you run local models, your data never has to leave your device.

你的智能体，你的机器，你的生活。

Your agent, your machine, your life.

这甚至不是我的论点。

That's not even my thesis.

那是加里的，你知道吗？

That's Gary's, you know?

我们只是先把它实现了。

We just chipped it first.

所以我们也终于再次用OpenClaw来构建OpenClaw，但这次有个新花样：每个人都能在团队服务器上看到彼此的会话，还有一个Claw知道每个人在做什么，还能接管工作的编排。

So we also finally are building OpenClaw with OpenClaw again, but this time with a twist where everyone sees each other session on a team server, and there's a claw that knows what everyone else is working on, and also can also take over orchestration of the work.

我们终于要慢慢告别这个奇怪的未来片段了，在那个未来里，人们得在终端里工作，还得开着笔记本电脑到处跑，因为代理需要持续运行。

And we are finally leaving this slowly, this weird blip in the future where people work in terminals and run around with their laptops open because the agent needs to keep working.

如果你们只记住三件事。

If you only remember three things.

第一，别停止享受乐趣。

Number one, don't stop having fun.

找到乐趣是终极驱动力，这样你才能得到最好的想法。

Find is the ultimate driver so you can you can get the best ideas.