OpenAI内部智能浏览器Atlas探秘

我认为让我对这个世界感到兴奋的一点，不仅仅是发展的速度，因为我觉得要让一个功能正常运行，总是需要几次迭代。

I think one of the things that has excited me about this world is is not just the pace of development, because I think to get a feature to work right, it's always gonna take a few iterations.

关键在于你能多快地决定某件事值得去追求。

It's how quickly you can decide that something is worth pursuing.

我在使用 Atlas 时最大的顿悟是，我意识到自己再也不需要查看任何设置面板了。

The big unlock that I had with Atlas is I realized I never need to look at a settings panel ever again.

你不是一个人在战斗。

You're not alone.

如果你在大型代码库中工作，你就知道脑子里装太多东西意味着什么。

If you work in large code bases, you know what it means to hold too much in your head.

哪个文件导入了什么，哪个服务依赖于哪个数据库模式，以及如果你修改这一行代码，什么会崩溃。

Which file imports what, what service depends on what database schema, and what will break if you change this one line.

瓶颈不在于编写代码。

The bottleneck isn't writing code.

而在于保持整个系统在工作记忆中足够长的时间，以便做出决策。

It's holding the entire system in working memory long enough to make a decision.

AugmentCode 是一个减轻上下文负担的 AI 编程助手。

AugmentCode is an AI coding assistant that offloads context.

它的上下文引擎能够理解你的整个代码库，包括当前文件的内容以及你整个系统的架构结构。

Its context engine understands your whole code base, including what's in your current file and the architectural shape of your entire system.

它支持多语言交互、遗留代码以及那些未被记录的依赖关系。

It works with multi language interactions, legacy code, and the dependencies that aren't documented anywhere.

真正让这一切发挥作用的是系统本身，Augment 在他们的 AI 驱动大规模工程实践手册中对此进行了详细说明。

The system is what makes it work, and Augment documented it in their AI powered engineering at scale playbook.

其中包含了如何评估当前状态、从个人实验过渡到团队全范围部署的四阶段框架、即用型检查清单，以及在那些错误代价高昂的代码项目中，能实现 30% 更快的 PR 速度和 40% 更短合并时间的具体工作流程。

Inside, it includes how to assess your current state, the four phase framework for moving from individual experiments to team wide deployment, ready to use checklists, and the specific workflows that produce 30% faster PR velocity and 40% shorter merge times at companies working on code where mistakes are expensive.

这专为在高风险生产系统中工作的企业团队设计。

This is designed for enterprise teams working on high stakes production systems.

它同时服务于合规性、正确性和可维护性。

It's both for compliance, correctness, and maintainability.

它适用于的不仅是原型阶段，更是你交付数百万用户依赖的代码的时刻。

It's both for the moments you're not just prototyping, but shipping code that millions of people depend on.

团队已经看到了可衡量的成果：PR 速度提升 30%，合并时间缩短 40%。

And teams are seeing measurable results, 30% faster PR velocity and 40 shorter merge times.

请前往 augmentcode.com/resources/aipoweredengineering@scale 下载《规模化 AI 工程实践》手册。

Download the AI powered engineering at scale playbook at augmentcode.com/resources/aipoweredengineering@scale.

地址是 augmentcode.com/resources/aipoweredengineering@scale。

That's augmentcode.com/resources/aipoweredengineering@scale.

现在回到本期节目。

And now back to the episode.

本和达伦，欢迎来到节目。

Ben and Darren, welcome to the show.

嗨。

Hey.

谢谢。

Thank you.

很高兴能来这里。

Great to be here.

是的

Yeah.

彼此彼此。

Likewise.

太棒了。

It's awesome.

对于不了解你们的人，你们两人正在开发 ChatGPT Atlas，这是一个智能代理浏览器。

So for people who don't know you, you are both building ChatGPT Atlas, which is an agentic browser.

本，你是工程负责人。

Ben, you are the head of engineering.

达伦，你是技术团队成员。

Darren, you're member of the technical staff.

我相信你们俩最初都参与过 Chrome 的开发。

I believe you both worked on Chrome originally.

这是真的吗？

Is that true?

没错。

That's right.

没错。

That's right.

我们曾经一起合作开发过多种浏览器，而且持续了很长时间。

We worked on a number of browsers together and for a long while.

哦，这真是太酷了。

Oh, that's really cool.

我之前没意识到，你们的合作关系是通过多个不同产品和公司逐步发展起来的。

So I I didn't realize that you're you're this is like a it's an evolving partnership through many different products and companies.

这真的很有意思。

That's really interesting.

我们最早在Netscape一起工作，之后又共同参与了Firefox项目几年，接着又一起做Chrome，现在又在做Atlas，这让人无比兴奋。

We worked together first at at Netscape, then on Firefox together for a few years, and and then with with Chrome, and now Atlas, which is super exciting.

绝对是元老级人物。

Absolute OGs.

好的。

Okay.

这太酷了。

This is really cool.

我每天都在使用Atlas，之前我用的是Dia，我知道，达伦，你曾经在这家浏览器公司工作过。

So I'm using I'm a daily Atlas user, and I switched from Dia, which I know, Darren, you used to work at the browser company.

我和乔什和赫希是好朋友。

I'm good friends with Josh and Hirsch.

所以如果你们在听的话，也许我们可以想办法……但Atlas真的很好用。

So if you're if they're listening, you know, maybe there's a way to But get me Atlas is Atlas is is pretty good.

让我觉得特别有趣的是，使用Atlas和这种高度自主的浏览器时，头几天我完全不知道该拿它做什么。

What's really interesting to me about using Atlas and using just just really agentic browsers is I for the first couple days, I was like, I don't have no idea what to do with this.

我知道它很有能力，但我一时想不出什么时候会真的需要使用它。

Like, I know it has this power, but I don't I I can't think of a a time when I want I might wanna use it.

但现在，每一天都有至少五十件事让我觉得，如果还得去点击那些该死的表单或设置页面，我真会炸毛。

And now I'm just, like, every single day, there's, like, 50 different things that if I had to, like, click through another fucking form or settings page, I would, like, blow my head off.

但这不正是人们使用AI工具的普遍经历吗？

But isn't that kind of the journey that people have with AI tools in general?

比如ChatGPT或者这些编程工具，你只有真正用起来才能理解它的强大之处。

Like, ChatGPT or these coding tools, you kind of don't really understand the power until you get into it.

我觉得这是对的。

I think that is true.

但我并没有完全经历过这种过程。

I didn't quite have that experience.

当我第一次看到它，比如GPT-3在写东西的时候，我就想：哇。

Like, the first time I just saw it, like, writing gbt three writing stuff, I was like, woah.

这太疯狂了。

This is crazy.

但没错，我想确实如此。

But, yeah, I guess that is true.

那么，我想从你们两位的角度来看看，我很好奇。

Well, I I guess I'm curious from your from from both of your perspective.

比如，如果有听众知道智能浏览器这种东西存在，也许试过，但根本不清楚为什么要用它，或者它有什么用处。

Like, if someone is listening and they're like, I know that agentic browsers are a thing, and maybe I've tried it, but I actually don't even know why I would use this or, like, what it's useful for.

智能浏览器的愿景到底是什么？

What is what is the sort of vision for agentic agentic browsers?

让我们尽量具体一点，别光说‘是的’。

And let's be let's try to be more specific than, like, yeah.

它能帮你做所有事情。

It just does everything you.

你知道的。

You know?

那么，具体来说，智能浏览器会如何改变你日常使用网页的方式？

Like, what is the like, what are the real day to day things that agentic browsers change about how you might use the web?

是的。

Yeah.

我认为，未来可能会达到这样一个阶段：越来越多的工作可以被自动化，而且我们正在朝这个方向取得进展。

So I think that, you know, maybe the the future will get to a place where, like, more and more of your workload can be can be automated, and I think we're making progress in that direction.

但今天，我们设计Atlas时的核心理念是，你可以将ChatGPT带到你在网页上的任何地方。

But but today, we wanted to design Atlas with this idea that you could bring ChatGPT with you wherever you go on the web.

所以是的。

And so yeah.

我的意思是，你提到的这个问题——我该怎么用它？

I mean, I think the the thing that you note of, like, what do I do with this?

这是我们经常从用户那里听到的反馈，但同时，我们也看到一些用户和你一样，经历了同样的探索过程，逐渐找到了使用场景。

This is something that we we hear a lot, from from people, but then also we hear some moments as they go on the same journey that that you have and and begin to figure out some use cases for it.

我们实际上希望，能将从用户使用中获得的这些经验，转化为更主动的建议，融入产品中，帮助用户更好地掌握和优化这个工具的使用。

This is something that we actually want to, take some of that that learning that we have from how people are using it and and help offer more proactive advice to people, like in products, to help them figure out how to optimize use of the tool.

但我认为，今天当我使用Atlas，再回头使用传统的非AI浏览环境时，我发现自己能提出更多问题，对一个话题也有了更深的了解。

But I think today, like, one of the things that I noticed, when I use Atlas versus when I go back and use a sort of pre AI browsing environment, I find myself just able to ask just a lot more questions and just be more knowledgeable about a topic.

如果我在网上购物，我会更有信心自己拿到了最优惠的价格、正确的优惠券，或者所有相关的信息。

If I'm doing online shopping, I can feel confident that I'm getting the best deal, or I have the right coupon code, or I have all that sort of stuff.

如果我在研究某个感兴趣的话题，我可以自由地发散思维，探索不同的观点。

If I'm, like, researching a topic that's of interest to me, I can sort of, brainstorm, you know, different viewpoints on it.

我可以拥有这样一位朋友或顾问，它随时陪伴着我，我可以随时和它聊天。

I can just sort of have this this sort of friend or adviser that that sort of comes with me, and I can just, like, have this conversation with it.

这让整个网络变得更加丰富和生动。

And that that just made the web a lot richer and more dynamic.

你能给我举个更具体的例子吗？

Can you make that more concrete for me?

因为我觉得，有些人听到这些可能会想：‘嗯，我现在用Chattypuji就能做到这些了。’

Because I think some of those things, someone might be listening and being like, well, yeah, I could do that with Chattypuji now.

Chattypuji就是为我做这些的。

That's what Chattypuji does for me.

那么，在浏览器的环境下，拥有这样的功能究竟意味着什么？

So what does it what does it mean to have that in the context of your browser?

这意味着你不需要再切换标签页了。我想，任何用过在标签页里打开ChatGPT的人都有过这样的经历：从另一个标签页复制内容，粘贴进去，然后问它问题。

It just means that you don't need to go I mean, I think if you know, for anyone that's had ChatGPT in a tab, you probably have the experience of going and taking some content from another tab and pasting it in, and asking a question about it perhaps.

而当你使用的浏览器从底层就内置了这种功能时，上下文会直接提供给模型，你就不用一遍遍重复自己了。

Whereas when you have a browser that's built with this at the core of it, you know, that context is provided, directly to the model, so you kind of don't need to keep repeating yourself.

ChatGPT 会直接看到你在看什么，并能就此提供它的见解。

It will just ChatGPT will just see what you're looking at, and be able to offer, you know you know, its thoughts, on on on that.

我认为这正是释放整个技术潜力的关键所在。

I think that's really the big the big unlocking the power of this whole thing.

随着人们在生活中越来越多地使用 Chattypuji，他们会意识到，或许应该更多地以 ChatGPT 开始他们的提问。

It's like, I think as people, use chattypreeti for more things in their life, they realize that maybe they should start more of their queries with ChatGPT.

对吧？

Right?

在某个时刻，你会自己慢慢领悟到这一点。

You start to learn that for yourself at a certain point.

你会想，为什么我还用老方法做这些事呢？

You're like, why am I doing things the old way?

那太手动了，相反，我应该直接问这个 AI 模型。

That was very manual, but instead I should ask this AI model.

它能帮我节省一些步骤。

It will help me save some steps.

这个浏览器将这一点置于核心位置。

And this browser puts that at the center of it.

网址栏会引导你针对你的问题进行查询。

That's what's the URL bar will guide you towards for your queries.

对吧？

Right?

它能让你更轻松地使用 ChatGPT。

It helps you get into ChatGPT with a lot lower friction.

正如本所说，当你在网页上对某事感到困惑时，Ask ChatGPT 就在你身边。

And as Ben was saying, you know, if you're on a web page and you're scratching your head about something, Ask Chat TPT is right there.

你可以直接问它。

You can ask it.

它拥有上下文信息。

It has the context.

你不必复制粘贴，然后说：你能回答这个问题吗？

You don't have to copy paste and say, can you now ask this answer this question?

所以这要简洁得多。

So it's just a lot more streamlined.

这正是整个产品核心的价值主张。

That's that's kind of the core value proposition of this whole thing.

在此基础上，我们还构建了一些功能，用户可以选择启用网页记忆功能。

And on top of that, we build, you know, features that people can opt into around web memories.

如果智能代理或模型一直伴随你的旅程，你之后也可以向它查询它所了解的内容。

So if the agent or the model is there in on your journey, you can also query it later about things that it's it it knows.

当你试图找回某些信息或理清生活中各种杂乱事项时，这会非常强大。

And that can be very powerful to you as you're trying to get back to things or trying to make sense of just all the things in your world.

无论你正在经历什么样的旅程，无论你在进行什么研究项目或工作，让智能代理被动地陪伴在你身边，同样会非常强大。

You know, whatever kind of journey you're on, whatever research project you're on, whatever work you're trying to do, having it there sort of passively can be very powerful too.

我得告诉你，我希望这也能成为一次小规模的用户调研，因为我对现在用这种方式做事情感到非常兴奋。

I gotta tell you, like and hopefully maybe this can be, like, a little bit of a user research session too because, like, I feel like I'm I'm doing something with this that I'm very excited about.

我想知道你们有没有这样做，是否看到其他人也在这么做，你们是如何为这种场景做设计的？

And I'm curious if you guys are doing it, if you're seeing other people doing it, how you're building for this.

我对Atlas最大的突破是，我意识到我再也不需要查看任何设置面板了。

So the the big unlock that I had with Atlas is I realized I never need to look at a settings panel ever again.

你不是一个人。

You're not alone.

是的。

Yeah.

这种感觉真的非常令人耳目一新。

And that is such a refreshing feeling.

我认为这对用户和软件开发者来说都很令人耳目一新。

It's I think it's both refreshing for users and for software developers.

我认为这对软件开发者来说很轻松，因为你不必再担心增加另一个调节旋钮，因为智能代理会帮你处理这些。

I think it's refreshing for software developers because you don't have to worry about adding another knob because someone like, the agent's gonna do that.

所以你可以更轻松地让软件变得更可定制。

So you can you can make the software more custom customizable more easily.

但对于用户来说，我觉得最典型的例子就是看AWS控制面板。

But for users, like, I think the canonical example for me is looking at the AWS dashboard.

我不确定你们有没有登录过那个系统，我猜你们都登录过，里面一共有大约50个不同的服务。

I don't know if you have like, I assume you guys have both logged into that, and it's, like, 50 different services.

然后那个设置和权限系统，简直就像发射一枚核导弹一样，才能完成任何操作。

And then there like, your the settings the, like, the permissioning system is, like it's like launching a nuclear, like, missile in order to, like, do anything.

我经营一家公司，我们有大约20个人。

And I run a company, and we have, like, 20 people.

所以我经常被问到：嘿，我们能加一个账号吗？

And so I'm I'm sort of constantly being asked, hey, can we, like, add a seat to this?

或者，你能改一下这个东西的权限吗？

Or, like, can you change the permission on this thing?

或者，那是我们五年前设置的一个账户，我都记不起来了。

Or and it's, like, some account that we set up five years ago that I don't even remember.

我们不会这么频繁地做这些事。

We don't do these things so frequently.

是的。

And so Yeah.

是的。

Yeah.

怎么操作又想不起来了。

It's, like, not top of mind how to do it again.

对。

Yeah.

我举个例子，我用它来帮我去创建谷歌表单做用户调研。

My my example of this that I've been using was, I used it to help me create Google Forms to do user research.

谷歌表单构建器我觉得可能比AWS控制面板简单一些，但是……

And, you know, the Google Form Builder, I think, is maybe less complicated than the AWS control panel, but

但还是……

It's still,

我不是每天都用它。

it's not something I use every day.

所以对我来说，能请智能代理去帮我完成这个任务，几分钟内搞定，然后我直接提交，确实让我更快地聚焦到问题的核心。

And so I think for for me to be able to ask the agent to go off and do that and have it do that in a few minutes and come back and I can just submit, You know, it certainly allowed me to get to the the meat of the problem much quicker.

是的。

Yeah.

这属于那种需要一定启动能量的任务，而你现在不必再花这份能量了。

It's it's one of those tasks where, like, there's a certain amount of activation energy, and you don't have to spend the activation energy anymore.

达伦，你刚才想说什么？

Darren, what were you gonna say?

我只是想说，现在到了一年中的这个时候，得登录Workday系统，找怎么下载我的年终薪资单，好拿给我的税务顾问。

I was just saying it's that time of year to go into Workday and figure out how to get my year end to date pay stub so I can share that with my tax adviser.

然后我就想，我该去哪儿找啊？

And I'm like, where do I go again?

你知道的。

You know?

他们又把位置改了，你知道吧，我也没频繁去那里。

They moved it again, you know, and I I won't go there often enough.

所以我觉得，对于导航网页应用，尤其是像你提到的AWS这种复杂应用，这功能太强大了。

So I think it's super powerful for navigating, like, web apps, especially complex ones, like you said, with AWS.

而这正是这些工具的明确优势之一。

And it's just that's that's that's one of the definite superpowers of these things.

那么，你是如何看待你的用户群体中这一功能的发展的呢？

So how are you seeing that evolve with your user base?

比如，如果你能分享的话，有多少比例的用户实际上已经掌握了这一点？

Like, what percentage if you can share, what percentage of people are have actually figured that out?

因为这确实非常强大，但我想象它并不一定是一个日常使用场景。

Because I it's super powerful, but I also imagine it's not necessarily a daily use case.

它可能每周用个两三次，能救急，但其他时候，我可能根本不会用它来做这个。

It's like a couple times a week, it is a lifesaver, but other than that, I may not use it for this.

比如，我每周只会进设置界面几次。

Like, I'm I'm only going into settings, like, a couple times a week.

所以我很想知道，这是否是你能倚重的使用场景之一？

So I'm curious, is that one of the one of the use cases you can hang your hat on?

用户真的在发现这个功能吗？还是说它仍然处于早期阶段？

And are are people really discovering it, or is it still sort of nascent?

我不确定我们是否有关于代理类浏览器驱动功能的精确数据，但我们知道，人们与该聊天功能的互动是浏览器的主要使用场景之一。

I don't know if we have the exact stats on on the sort of agent, like, browser drives kind of a thing, but we do know that just in general, people interacting with that side chat is a main, like, use case for the browser.

我认为，大多数人可能都会定期使用这个功能，因为它确实是主要的价值点和核心界面。

And I I think, you know, like, probably most people are are using that on a on a regular basis just because it is kind of the main value add, sort of the main surface.

至于人们从这个功能中使用了哪些工具或能力，我们还没有像那样详细地进行分类统计。

In terms of what tools or capabilities people use from that, I don't think we've got that broken down quite the the same way.

是的。

Yeah.

你看到的情况正如你所预期的那样：当人们初次接触这些工具时，并不了解它所有的功能，这确实是我们需要关注的一个问题。

What we what you see is sort of what you'd imagine is that people are are you know, when you first come to these tools, you don't know all the things that it can do, and that's definitely a topic for us.

我们该如何引导用户发现这些功能，同时又不会让他们感到信息过载？

How do we introduce people to things but not also overwhelm them at the same time?

你希望界面既熟悉、简洁、易于上手，又在背后功能强大，让用户在不断探索中获得回报。

You know, you wanna have you wanna balance something that's familiar, simple, seems approachable, but also it's powerful under the hood, so you get rewarded as you discover further.

我认为，这正是用户体验设计的本质所在。

And, you know, I think that's that's kind of the nature of UX development.

对吧？

Right?

你可以拥有一个非常强大而复杂的工具。

You you can have a very powerful complex tool.

浏览器确实如此，但你也希望它易于使用且简单。

A browser really is, but you want it to also be approachable and easy.

你得思考人们通常会使用哪些模式，以及我们如何在这些时刻满足他们的需求。

And you gotta think about, like, what are the patterns people do and how can we meet them in those moments.

对吧？

Right?

你们做了哪些决策来实现这一点，比如如何让Atlas的复杂功能逐步呈现，从而使其直观易用？

What are the some of the decisions that you've made to, like, to do that, like, to to to enable the sort of progressive disclosure of complexities that Atlas is really intuitive.

但没错。

But yeah.

其中一个功能非常强大，但我们也曾为如何展示它而困扰，这个功能叫做光标聊天。

One of the features that is pretty powerful but or relatively or I should say we struggled with how to expose it is this feature called cursor chat.

当你在浏览器中与表单字段交互时，你会看到一个小图标，一个ChatTPity图标，你可以将鼠标悬停在上面，然后在该表单字段的上下文中与模型进行交互。

If you're interacting with a form field in the browser, you'll see a little icon, a little ChatTPity icon, and you can hover over it and then interact with the model and interact with the model in the context of that specific form field.

我们一直在纠结该如何让这个功能足够显眼。

We were we struggle with how to how in your face to make this.

对吧？

Right?

我们希望人们能意识到这种强大的功能。

We want people to be aware of this power.

这实际上非常强大。

It's actually really powerful.

使用它的人中，有很多人极力称赞它帮助他们进行撰写等内容创作。

The people who use it are there are people who rave about this, helping them compose and that sort of thing.

但事实上，即使我们提供了这个小提示，很多人还是发现不了它。

But, actually, a lot of people don't discover it even though we have this little hint.

所以这始终是个问题：这个提示该做多大才合适？

And so it's always a question, how big do you make that hint?

你如何向人们介绍这个功能？

How do you introduce this to people?

当然，在用户引导阶段，我们已经有很多内容想告诉用户，因为这是一个AI浏览器。

Certainly, during onboarding, we already have a lot of things we try to tell people about cause this is an AI browser.

有很多新东西需要学习，比如网页记忆、侧边聊天等基本功能。

There's new things to learn, fundamental things like web memories and and and, you know, capabilities like side chat and so on.

但我们一次只能告诉用户这么多内容。

But we can only tell you about so many things at once.

因此，从设计角度来看，这一直是我们面临的挑战。

So that's that's been a challenge for us from a design perspective for sure.

这很合理。

That makes sense.

我见过那个小图标。

I have seen that that that little icon.

但我还没有点击过它。

I have not clicked it.

现在我觉得我必须点一下它了。

Now I need now I feel like I need to click it.

这正是它与你只是在标签页中使用ChatGPT的不同优势之一：你可以直接在特定文本框中调用它。

It's one of those things where it's like it's another advantage of having this sort of fully integrated with with your browsing environment as opposed to just having ChatGPT in a tab is that you can kind of summon it into the specific text field.

这是我妻子经常使用的一个功能。

So this is a feature that my wife uses quite often.

她需要写邮件。

She has to write emails.

她参与了很多不同的事情，有这个功能在身边，能大大加快她的工作流程。

She's involved in a number of different things, and it it just helps, like, speed up her workflow quite a lot, having it there.

而且这不仅仅是你在那里的ChatGPT。

And the thing is is it's not just it is like your ChatGPT there.

它背后有你的个性化设置、自定义指令，所有这些内容都整合其中。

So it has your personalization, your, you know, custom instructions, all that kind of stuff, you know, behind it.

所以它会按照你希望的方式写作，所有这些都一应俱全。

So it, you know, writes the way you want it to write and and all of that.

所以这挺酷的。

So it's pretty cool.

总的来说，我们非常感兴趣的是模型如何与网页互动，以及我们如何与之互动，它如何与你正在做的事情无缝衔接。

Broadly speaking, we're really interested in the whole idea of, like, how the model can interact with the web and the ways that we interact and how can it dovetail with what you're already doing.

所以，你通过在聊天中输入/agent调用的智能体，就是这种理念的全面体现——你让它接手一个任务，直接与网页互动，点击按钮，为你完成所有操作。

So, you know, the agent is that you invoke with slash agent inside chat is, like, a very, all in sort of manifestation of that where you're asking it to take a task and go interact directly with the page and all the and push the buttons and do everything for you.

而这可能是这种理念最宏大的体现形式。

And that's that's sort of, like, maybe the grandest representation of this kind of idea.

但还有许多更小的、即时的版本，比如我们提到的Cursor Chat，或者侧边聊天具备上下文这一点。

But there's all these sort of smaller in the moment kind of versions of that, like we said with cursor chat or just the fact that side chat has the context.

当你提出问题时，它能理解你正在做什么。

And when you ask a question, it can understand what you're doing.

是的。

Yeah.

我认为这是它最有价值的部分之一，因为它在我浏览器里，登录了我所有的网站，可以代表我在多个网站上行动。

I think that's one of the most valuable parts of it is because it's in my browser, it's logged into all my websites, and it can act as me on any number of websites.

所以即使它不是我，但它拥有和我完全相同的权限和所有类似的功能。

And so even though it's not me, it's like it has all the same affordances and and and and and all that kind of stuff.

我对你们如何看待网页为此类应用如何演变感到好奇。

And I'm curious about your opinion on how the web will evolve for that.

因为目前，网页设计本质上是将机器人和人类严格区分开来的。

Because right now, there's like there's a it's really designed for this, like, bifurcation between bots and humans.

存在一种人类体验，也存在一种机器人体验——你被默认为在爬取内容，而机器人则在执行任务。

And there's a human experience, and there's a bot experience that you're you're presumed to be crawling, and there's, you know, bots.

文本和机器人。

Text and robots.

文本和所有类似的东西。

Text and all that kind of stuff.

而这是一种介于两者之间的存在：它具有个人化特征，由你驱动，但又不是你本人。

And this is sort of this sort of in between thing where it's personal, and it's driven by you, but it is not you.

那么，你认为网页应该如何为这种类型的应用发展呢？

And, like, yeah, how do you think the the web should evolve for that kind of thing?

是的。

Yeah.

所以，这确实是一个非常有趣的问题。

So, I mean, this is a super interesting one.

我认为随着时间推移，必须出现某种概念，比如一个非人类的操作者，虽然不是本人，却代表某人执行特定请求，因为我把这些事情看得很不一样。

And I I do think over time, like, there'll have to be some, like, notion of, you know, maybe a nonhuman operator that is acting nonetheless on behalf of a human, like, for a specific request because I see these things as quite different.

举个例子，比如网络爬虫。

To example to, for example, web crawlers.

你知道，网络爬虫在互联网上遍历网站，整合信息以造福大众。

You know, web crawlers are out there traversing websites and, you know, synthesizing across that for the benefit of many.

而如果你自己写一个本地的shell脚本，去获取网站内容，直接向你想要的资源发起HTTP请求等等，虽然也能做同样的事，但无疑要痛苦得多。

Whereas, like, a you know, this is you could do the same thing, you know, admittedly much more painfully if you were to write a, you know, local shell script that would go off and obtain the content of a website, maybe issue, like, the direct HTTP request to the the resources that you wanted and so on.

而这种东西更接近后者，因为它背后是你个人化的意图。

And this is much closer to that, where there is, like, your own personalized intent behind it.

所以，从我们对这些事物的概念性理解来看，这就是我的看法。

So I think just from, how we think about these things conceptually, that that's how I look at it.

从事物的发展角度来看，我认为最有趣的一点是，某种程度上，这些东西其实并不需要进化，因为我们已经有了可以自动读取屏幕、点击并完成各种操作的计算机使用模型。

In terms of how things evolve, I think one of the most interesting things about it is, at some level, like, stuff doesn't need to evolve, because we have, you know, computer use models that can just go off and, like, read the screen and click and do all that that sort of thing.

我认为这里的许多进化将来自于：有没有办法让这个过程更加无缝？

I think a lot of the evolution here will come from, are there ways to make that more seamless?

有没有办法让其性能更高，从而让我们能同时处理很多事情？

Are there ways to make that, you know, higher performance so that we can do many things at once?

你知道，基本上就是支持这种能力的规模化扩展。

You know, do you know, just basically support sort of scaling this up.

因为我认为我们真正想做的是，拥有一种能够长时间代表你同时执行多项任务的系统。

Because I think what we really want to do is have something that can do many things on your behalf simultaneously over the course of time.

而这将需要平台本身发生更多有趣的进化。

And that and that will just require, like, a lot more sort of interesting sort of evolution of the platform.

我认为实现这一点可能有多种不同的方式。

And I think there are probably a variety of different ways to do that.

网络的一个美妙之处在于它是一种非常声明式的媒介，我们已经开始利用这一特性，但我认为我们尚未完全发掘出网络这一独特属性的全部潜力。

One of the wonderful things about the web is that it's a very declarative medium, and so this is something that we've begun to tap into, but I don't think we've fully, like, realized the potential of that that interesting property of the web yet.

你能为正在听的听众解释一下，什么是‘声明式’，以及为什么这是一个有趣且重要的特性吗？

Can you explain, for you who are listening, what declarative means and then why that is an interesting and important property?

是的。

Yeah.

所以，支撑网页的是一种叫做HTML（超文本标记语言）的技术。

So so the web you know, powering the web is this technology called HTML, hypertext markup language.

这是一种构建所有网页的方式，也就是你今天在网页上交互的所有用户界面。

And it's a way that, you know, all of the web pages are built, you know, all of this sort of UI that you interact with on the web today.

它只是以一种特定的方式格式化的文本组合。

It's a combination of just text, formatted, in this this specific manner.

这些被称为标签。

There are these things called tags.

比如，一个按钮可能是一个按钮标签，里面包含显示在按钮上的文本。

So a button might be a button tag that encloses the text that is rendered on the button.

浏览器会读取所有这些内容，并知道如果看到像button或input这样的标签，它们都有特定的含义。

And so what the browser does is it reads all of this, and it knows that if it sees a tag that says button or input or something like that, that there's a specific meaning to it.

而有趣的是，比如表单，它实际上是用户提供数据后向远程函数发起调用的一种方式。

And then what's interesting, for example, with forms is that a form is the way that you do effectively like a a call to a remote function with some data that the user provides.

所以当我填写一个表单，比如进行搜索时，我会在某个字段中输入文本，然后用这些文本调用某个远程函数，接着就会得到另一个页面。

So when I fill out a form, for example, to run a search, I take values that I there's a text that I type into this this field, and then I call some remote function with that text, and then I get another page.

因此，这一切都内嵌于网页的设计方式之中，这也解释了为什么浏览器被称为用户代理——浏览器被设计为读取所有这些标签，并以让用户满意的方式呈现内容。

And so there's all of this sort of this is sort of inherent to the way the web is designed, and it allows you know, the browser itself is referred to as a user agent, specifically for this reason, in that the browser is designed to go and read all of these tags and figure out how to present it to the user in a way that is satisfactory to them.

所以 Atlas 本质上也是一个用户代理。

And so Atlas is sort of a user agent agent.

没错。

That's right.

你觉得我们需要不同的用户代理字符串吗？这是否是 HTML 标准的一种潜在解决方案或扩展？

Do you think we need different user agent strings for is that like a is that one potential solution or extension of the HTML standard?

我不确定。

I'm not sure.

我觉得，只要看看网页的工作方式，回想一下我们曾经开发过的各种浏览器，就会发现用户代理字符串中蕴含着很多微妙之处。

I think just, like, looking at the the way the web works, you know, there's a lot of and just thinking back to various browsers that that we've worked on, there's a lot of subtlety to user agent strings.

而且，我不想陷入一种情况，就是因为我们改动了某些东西，导致网站无法正常工作。

And also, like, the the situation I don't wanna get into is where websites don't work, because we've changed something, about it.

你知道，有时候网站会检查这个字符串中非常具体的部分，并据此触发某些行为。

And, you know, sometimes there are sites that will check for very specific parts of that string, and it will be, they'll say something, you know, they'll trigger behavior based off of that.

我知道在Chrome早期，比如我们就遇到过这种情况，会导致网站无法正确渲染。

And I know early in the Chrome days, for example, we would see, you know, sort of behavior like that where it would cause sites not to render properly.

因此，从这个角度来看，由于Atlas主要基于Chromium，我们觉得从开发者角度来说，他们应该能感知到这一点。

And so from that sense, you know, with Atlas being predominantly Chromium, we feel like it's just, you know, from a developer perspective, they should perceive it.

他们应该像为任何基于Chromium的浏览器那样，同样地为它进行开发。

They should, you know, build for it the same way that they build for any Chromium based browser.

但随着时间推移，我们可能还需要找到其他一些信号或类似的东西。

But there's probably other signals or stuff like that that that we will need to come up with over the course of time.

现在对我们来说，要弄清楚这会是什么样子还为时过早。

It's just it's very early for us to figure out what that looks like.

但你最初关于网络可能会如何变化的问题，确实非常有趣。

But your original question about, like, how the web might change is a really interesting one.

我认为，随着越来越多的用户代理由智能代理或模型驱动，这可能会对开发者创建内容的方式产生一定影响。

I think as more and more of the user agents are perhaps driven by agents or models, you know, bought that not end end up having some bearing or impact on how developers create their content.

我认为，某个时候可能会出现一个转折点，看看生态系统如何演变会很有趣。

I think at some point, maybe there is an inflection point there if you know, it'd be interesting to see how the ecosystem evolves.

对吧？

Right?

你知道，人们创建内容是为了供人类消费。

You know, this people create content for human consumption.

过去，我们曾多次大力推动语义网，让网络更易理解。

In the past, we've always we've had moments when we're pushing heavily semantic web, make a web that's more understandable.

看看由此带来的所有好处。

Look at all the benefits that come from that.

屏幕阅读器会运行得更好。

Screen readers will work better.

网站会更易于机器理解。

Websites will be more machine understandable.

现在这些AI模型出现后，它们能够理解那些原本并不容易被机器理解的网站，因为这些模型以人类的方式与之交互，能够像人类一样提取信息。

What's happened in this now with these AI models is they're able to make sense of the websites that aren't very ordinarily machine understandable, but because these models are interacting with it in the way humans do, they're able to glean the information just as humans do.

这对计算机来说是一个巨大的突破，因为它现在能够理解这些网站。

And that's kind of a big unlock for the computer to help you because it can understand these websites.

对吧？

Right?

但随着越来越多的计算机模型、基础模型驱动这些系统，这些网站或许也会发生变化，谁知道呢？

But as as that unlocks more and more computer models, base models driving these systems, these websites, maybe, who knows?

也许网站本身也开始发生变化了。

Maybe maybe the the websites start changing as well.

我的意思是，这让我想起关于所有代码都由编码代理生成的讨论，编码代理在指挥其他编码代理，你知道，这一切最终会走向何方？它们该使用什么编程语言？诸如此类的问题。

I mean, it reminds me of discussions about what happens when all the code is being created by coding agents, and and the coding agents are directing the coding agents, and, you know, where where does where does everything go, and and what what programming language ought ought they use and all these kinds of things.

你开始怀疑，也许这里面有些科幻元素，值得我们畅想未来会如何演变。

You start to wonder, maybe there's some sci fi stuff there to kind of dream and imagine how things might evolve.

如果我说我清楚会发生什么，那是在撒谎，但我能想象事情会改变。

I I would be lying if I told you I know, but I can imagine things changing.

完全正确。

Totally.

有趣的地方在于，现在浏览器是一个非常好的启动方式，因为你不需要做任何改动。

That's the kind of the interesting thing is, like, right now, browser uses it's a really good way to bootstrap this because you don't have to change anything for it.

但一旦你完成了启动，所有人都开始使用代理，我就很好奇这是否真的是最有效的方式。

But once, like, once you've bootstrapped and everyone's using agents, I'm sort of curious if that is actually the most efficient way.

例如，本，你之前提到过，让代理同时为你做多件事。

For example, Ben, you were you were talking about, you know, having it do multiple things for you at once.

你知道，让Atlas逐个浏览网站有点慢。

You know, having a watching Atlas scroll through websites is kind of slow.

也许存在一种更原生的代理方式，让代理能够像MCP那样与网站交互。

And there there may be a more agent native way to allow agents to interact with websites like MCP, for example.

你们有没有在考虑这些方向，还是仍然只专注于核心功能？

Are you guys thinking along those lines, or is are you still really just focused on the core stuff?

我们正在思考一系列不同的技术，以帮助我们实现网页浏览。

We're thinking through, like, a whole host of different different technologies to help us drive drive, you know, web browsing.

我认为，除了Atlas团队之外，还应该更广泛地思考ChatGPT在做什么。

I think as well, like, beyond the the Atlas team, just to think broadly about what ChatGPT is doing.

比如，我们也推出了围绕这个产品的生态系统，我认为这正是我们鼓励开发者构建更动态组合世界的一种直接方式。

Like, we've also, you know, launched these these so eco this app ecosystem around around the product, and I think that that's sort of a a very direct way in which we're encouraging developers to build for a more dynamically composed world.

但这些当然都是在浏览器中实现的，可能并不属于Atlas本身的一部分。

But that's that's, you know, that's, of course, in the browser, but it's, you know, maybe not, like, part of Atlas in particular.

所以我认为，我们会尝试一些不同的方法，看看效果如何。

So I think some of these things are we're gonna try a few things and and and see how it works out.

此外，目前驱动ChatGPT Atlas代理的许多技术，其实源自OpenAI最初推出的Operator技术预览版。

Plus, you know, a lot of the technology that's powering the ChatGPT Atlas agent now, you know, has its roots in the original operator tech preview that Chat's that OpenAI put out.

如果回溯到那时，对比当时的性能和现在的表现，你就能看到进步的速度。

If you rewind the clock back to then and compare the performance then to now, you start to see, you know, sort of the rate of improvement.

在质量和性能方面，已经有了巨大的飞跃，我们正处在探索如何优化、让这些技术更好运作的曲线之上。

There's been, like, leaps and bounds improvements sort of the quality and and this performance, and we're kind of on that curve of of figuring out how to optimize and make these things work a lot better.

我认为，前方还有很多令人兴奋的工作和机遇。

And I think there's a lot of exciting work ahead and opportunity ahead.

这确实是一个值得与大家分享的重要进展，我认为它开启了想象与可能性的大门，让人们能真正看到它能帮到自己什么，就像你之前提到的那样。

This is this was a a meaningful step to share with people, and I think it's it opens the door to, like, imagination and possibilities and for people to have some real things that it can help you with, like what you were talking about earlier.

但还有更多即将来临。

But there's so much more to come.

你知道的吗？

You know?

你认为智能浏览器会让网页变得不再必要吗？

Do you think agentic browsers will make the web unnecessary?

我的意思是，你认为未来会不会出现一种情况，即待在ChatGPT内部、让代理自动完成所有浏览任务反而更好？

And by that, I mean, do you think there's a chance there's a there's a future state where it actually becomes just better to stay inside of ChatGPT and your agent is going off and doing all the browsing?

那时，也许它会根据品牌或作者希望你看到的内容，实时为你构建一个定制网站，但你不再像五年前那样直接看到渲染的HTML页面。

Then, you know, maybe it's maybe it's building, like, a custom website for you in real time based on what the brand or the writer has wants wants you to see, but you're not actually, like, seeing rendered HTML in the same way that you would have been, you know, five years ago.

我个人不这么认为。

I don't think so myself.

也许这只是因为我还没足够想象出ChatGPT未来会走向何方，但我确实认为，随着这些工具变得越来越强大，人们会越来越多地将任务委托给它们，而在未来十二个月内，它们将会变得极其强大。

And maybe maybe this is just me not not not being imaginative enough yet about, like, where ChatGPT will go, but I I do think that there's an aspect of I think we will see people delegate a lot more to these tools, especially as they grow more powerful, and they're gonna get, like, amazingly powerful over the next twelve months.

但我仍然认为，人们还有很多事情想自己动手做。

But I still think that there's a lot of stuff that people wanna do themselves.

无论是娱乐，还是购物、旅行规划之类的方面，我都希望深度参与其中，这可能至少源于我内心的好奇心，然后我会到网上去寻找答案。

And whether it's, you know, even just things like entertainment or, like, you know, there's aspects of of, you know, shopping or trip planning that I'd like, I do wanna be deeply involved with, and it's probably gonna start at least with some, like, curiosity that I have, and I'm gonna go out there on the web and find it.

我认为网络最令人兴奋的一点就是它拥有如此丰富的内容，所以我总是对探索它充满期待。

And I think the one of the most exciting things about the web is it has so much stuff on it, and so I'm always, like, excited to explore it.

我认为这一点永远不会消失。

And I don't think that will ever go away.

也许它会变得不一样。

Maybe it will be different.

也许会有一些人，比如今天的孩子们，他们从未生活在没有这些技术的世界里。

Maybe there will be folks that, you know, maybe they're like the the kids today that haven't sort of lived in a world without some of the stuff.

他们可能对此有不同的看法，但这只是我的观点。

Like, they may have a different view on it, but that's just mine.

我不知道你怎么看，达伦。

I don't know about you, Darren.

我知道。

I know.

我觉得人们喜欢逛商店。

I I think people like window shopping.

我觉得人们喜欢浏览。

I think people like browsing.

我觉得人们喜欢这种感觉。

I think people like that sort of thing.

或者你知道，我喜欢坐Waymo，但我也喜欢开我的手动挡汽车。

Or, you know, I love taking Waymo, but I also love driving my stick shift car.

而且你知道，有些时候两者都很重要。

And, you know, there's gonna be moments when both are, you know, important.

有时候我想要坐Waymo，有时候我想自己开车。

There there's moments when I want the Waymo and moments when I wanna be just driving myself.

你知道吗？

You know?

我认为未来就是这样子的。

And I think that's kind of the future is gonna always be that way.

而且，这也取决于你想做什么。

And, also it depends on what you're trying to do.

我觉得这些模型在为你合成信息方面可以非常出色，从而引导你进入手动模式。

You know, I think that these models can be just incredible at synthesizing things for you that might lead you onto the manual mode part of it.

对吧？

Right?

你可能会以一种非常自然的方式将这些技术融入你的生活。

And and and you're probably gonna just incorporate these things in a very natural way in your life.

你会在适合你的时候在两者之间切换，人们也会逐渐明白这一点。

You're gonna go between them where where it makes sense to you, and people are gonna figure out figure out that.

但始终会需要与网络应用，或者说应用程序进行互动，而网络是分发这些应用的绝佳媒介。

But there's going to be there's always gonna be a need to interact with with, you know, web apps, if you will, or applications, and the web is a tremendous medium to distribute those things.

电子商务，网络是这方面的绝佳媒介。

Ecommerce, the web is an amazing medium for that.

是的。

Yes.

你可以让模型帮你准备一个购物车商品列表，但你还是想亲自去看看那些商品。

You could ask your model to please prepare you a, you know, a shopping cart of items, but you're gonna wanna go look at it and you're gonna wanna go see things yourself.

在大多数情况下，你不会只是说，好吧，帮我买下它，而不先看一下。

You're not just gonna be like, yeah, buy that for me without seeing it in most cases.

所以，我认为我们正逐渐进入一个融合的世界。

And so, you know, I think there's there's kind of this blended world that we're probably coming by.

AI 有一个方面，就像一个你可以委派任务的同事或伙伴。

There's an aspect as a of of the AI as a you know, like, actually, like, a a workmate or a coworker or something that you can delegate to.

而 AI 还有另一个方面，是作为思维伙伴或合作者的角色。

And then there's an aspect of the AI as a as a thought partner or a collaborator in that sense.

我认为这两种角色实际上优雅地共存，不是非此即彼。

And I think that these worlds sort of are actually elegantly you know, it's neither one or the other.

它更像是两者兼而有之。

It's kind of both.

是的。

Yeah.

对。

Yeah.

作为思维伙伴，这些模型已经实现了这一点。

Definitely as a thought partner, this is already the case for these models.

当你在家研究某件事时，向聊天机器人提问可以节省你一些时间，我只是把它当作一种探索方式；当我编程时，我也会通过它来激发想法。

You know, when you're researching something for at home, you know, asking the, you know, the chatbot about it saves you some time, figure you know, just as an exploration, bouncing ideas off when I'm coding, I'm doing it that way.

我生活中很多地方都引入了ChatTPG，帮助我梳理思路，明确我正在处理的问题类型。

So many things I'm bringing ChatTPG into my life to help me sort through my thoughts and and what kind of problem I'm working on.

我认为，这就是我所想象的，未来还有很多类似的情况可以类比。

And and I think that's that's sort of what what I imagine you know, I can imagine lots of parallels to that in the future.

我好奇的是这一点。

So here's the thing I'm curious about.

当乔什和赫希刚开始创办浏览器公司时，我和他们已经是很久的朋友了。

When Josh and Hirsch were first starting the browser company, I'd been friends with them for a long time.