By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.
OpenAIChatGPT Atlas and the next era of web browsing — the OpenAI Podcast Ep. 9
Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.
Transcript
85 min read · 16,751 wordsWhat ChatGPT Atlas is: a browser built around natural-language interaction
Andrew Mayne 0:00 Hello, I'm Andrew Main, and welcome to the OpenAI Podcast. There have been a lot of exciting releases from OpenAI recently, including GPT 5.1, Sora, and one of my favorite new applications, ChatGPT Atlas. Today, we're going to be talking to the team behind it, Ben Goodger and Darin Fisher, and explore some of the reasons for why OpenAI decided to make a browser, what the future of agentic capabilities mean, and where everything's headed next.
Darin Fisher 0:27 [upbeat music] The time is right 'cause it's actually how people should be starting their journey.
Ben Goodger 0:32 We're moving to a world where you can just tell the computer what you want.
Darin Fisher 0:35 So I think it's kind of powerful, this idea that it, that the agent has its own workspace.
Ben Goodger 0:39 My view for this has always been that this is like a long-term investment.
Andrew Mayne 0:45 Let's begin with, what is Atlas and why?
Ben Goodger 0:50 So Atlas is a new kind of browser, uh, for an era of the web where people are interacting with new technology in natural language. And so it's a- the kind of browser where you can just tell it what you want, whether it's to find the next outfit that you're gonna buy or to help you solve a really hard problem, and then it can help you harness the web to get a bunch of stuff done. Um, and so central to this, this idea is that if we take ChatGPT and make it the heart of your browser, not just an add-on, uh, it's something that can actually help you make sense of the content that you're seeing on the web. It's something that can help you take action on the web. It's something that can, you know, learn from your browsing to personalize your experience and help you with tasks that aren't just done in a few minutes, but might take days, or weeks, or months, or just generally help you, uh, become a more curious, more effective person.
Darin Fisher 1:48 And it can help you come back to a task that maybe you've not had a chance to work on in a while because it will remember what you were doing for you and help, help you, uh, get right back into where you were.
Why build it now: models, computer-use, and the “slope” of capability
Andrew Mayne 2:00 Why now?
Ben Goodger 2:01 I think the, the progression of technology, you know, with, with these AI models has been really stunning to watch over the past couple of years. Uh, and it feels like we're at this sort of sweet spot, where the capabilities of, of, um, not just the LLMs that have powered ChatGPT, but also sort of this new area of, uh, computer use and some of the other surrounding technology, is at a point where we can build some really compelling experiences for people, so we wanted to give it a shot.
Darin Fisher 2:31 You know, like Ben said, the models have gotten so much better, and they continue to get better, and you see the slope of innovation there and the pace of, of improvement. If you look back at the beginning of the year when Operator first came out, for example, and it sort of like hints at some of the potential, and now you fast-forward to where Atlas is with agent, agent and how much faster it is, how much more capable it is, just look at that slope, and you start to project, "What about ne- what's it gonna look like next year, five years out?" et cetera. And to get this sort of-- get that foundation in place, that's what we were excited about, and it felt like the right time. Uh, for me, personally, you know, I, I felt like, you know, I had made that transition to seeing how ChatGPT makes so much sense in my life to- and how much I was using it and feeling like I'm putting ChatGPT at the core of a browser, not just another tab that you have to go to, to... but to have it be at the core and part of that flow. The time is right 'cause it's actually how people should be starting their journey, and so we want to make that, uh, just so natural and easy. And, and, and, uh, so I'm really excited to-- that, that we've been able to bring Atlas up, and I'm excited to bring it to more platforms.
The browser isn’t going away: the durable, open platform for work and research
Andrew Mayne 3:40 You two have quite a bit of history working in browsers. Uh, Netscape, uh, you've worked on Firefox and Chrome, and I'd like to kind of understand where you think we are right now with browsers. We've got, like, over a 30-year history of these, and now it seems like-- they seem like they've been changing a little bit incrementally, and then all of a sudden, now we're adding AI to it, et cetera. And how do you look at the browser landscape?
Ben Goodger 4:04 Well, I think we have, um, entered this really exciting time on the web, where we've added, um, this, this very, uh, human form of interaction, uh, in the form of these large language models, that you can just speak to the software and have it do the right thing for you. Uh, and so I think that's really gonna transform the way people get stuff done online. We've gone from a world where you've had to remember website addresses, you've had to go and search for them, and now you're just gonna be able to ask for the task that you want to get done, and you're gonna see it get done.
Andrew Mayne 4:35 I've noticed that a lot of people were thinking, like: Will we still have browsers? That was a question, "Are we gonna have browsers?" And it seems like browsers are gonna be here to stay for a while. Is that something you both feel?
Darin Fisher 4:44 It's a tool that people reach for a lot.
Ben Goodger 4:47 Mm-hmm.
Darin Fisher 4:47 You get a lot of things done on the web and using a browser. It's, uh, it's hard to imagine that not being a, a big part of, uh, how, how people use, uh, their computers. Um, I think that, uh, you know, I've been- we've been through many phases of the internet and many phasors- phases of browser development. There was a time when, you know, "Here comes mobile. Why, why would anybody use their, their laptop anymore? Why would anybody use a desktop computer?" And yet people continue to use desktop computers. They reach for it for very different things, maybe, and now that they can also reach for their phone for certain things. But the, the web browser continues to be, like, such an important tool on your computer for how you get work done, how you do research tasks, how you, you know, look for information and do, um, that kind of work. You know, substantive work happens within the browser, happens on the web. Don't really see that changing. If anything, I see that growing, um, because it's a- it's this, like, conduit to all the world's information. It's such an easy platform for people to bring experiences and make it available to everybody, and the browser just makes that so easy.
Ben Goodger 5:51 There's something sort of, um, interesting about it, where, like, every-... well, you know, of all of the sort of technological advancements that we've had over the past, you know, twenty-five, thirty years, you know, with the web, um, there's something very durable, durable about the browser-
Darin Fisher 6:04 Yeah
Ben Goodger 6:04 - where like, even if you look at this most recent wave of generative AI, you know, with, with ChatGPT launching, it launched on the web.
Darin Fisher 6:11 Mm.
Ben Goodger 6:11 It's, uh, you know-
Darin Fisher 6:12 It's true
Ben Goodger 6:12 ... I think, a very powerful reflection of the capabilities of this platform.
Darin Fisher 6:15 And the platform's itself is amazing. I don't really need to recount all of the ways it's amazing, but the fact that it's this very inherently open platform, you know, content is published to an, an open un, you know, platform, an open internet, where anybody can stand up a browser to consume that content. There are really no gatekeepers when it comes to the web, which is a really remarkable aspect to it. So people can freely publish information, and p- people can freely go and find that information, and so it's, it allows it to just blossom and grow and evolve in different ways. And then it's very natural that you might want to take something like an LLM and point it at it, because now it can, on your behalf, try to understand it and help you navigate it. So the idea that, uh, it can do all that means it just makes it so much easier when you're trying to take advantage of all that information that's out there as a user. Here it is, it can go and find it for you or understand it for you, or explain it to you, or... I just want to say, I think it's been really interesting to see the evolution of the web. You know, we got started in an era where, you know, it was the coming off of the dot-com boom, right? The, the, the, the needs of the browser back then were different. The kinds of things people were doing on the internet were different. They were totally exploring, and then over the course of time, as we've worked on browsers, the kinds of things people tried to do in the browser was just so much more. If you go back to, like, um, early two thousands, you had the evolution of, like, these more advanced web applications. Like, you, you know, I remember marveling at Google Maps when it first launched.
Ben Goodger 7:40 Mm-hmm.
Darin Fisher 7:41 The fact that you could just scroll and pan through a map so effortlessly, and then it goes on from there, you know, uh, all the different kinds of web apps that people take for granted, things like YouTube or, you know, I'm going to list a lot of Google apps 'cause I worked at Google, but-
Ben Goodger 7:54 Mm
Darin Fisher 7:54 ... you know, Gmail, Google Docs, all these things. The kinds of things that, uh, you mean I can do all of that in my browser? It's kind of amazing. And-
Ben Goodger 8:01 It's become sort of like this, this operating system for your life-
Darin Fisher 8:04 Yeah. Exactly
Ben Goodger 8:05 ... on your laptop kind of deal.
Darin Fisher 8:06 Yeah, and it pushes the, uh, what the browser needs to be able to do, right? And so this era that, uh, we, uh, you know, developed Chrome in, was an era where people were already pushing the boundaries of what, uh, you could do inside of a browser. But they had good reason to do it because the web being this platform where it's so easy to put new experiences onto the internet, meant there was a lot of motivation to do that, and you wanted-- you know, developers are being very creative in how they could push the bounds of the, what the browser could do. But fast-forward to today, and it's, it's like you have all that, that's this foundation, and now, uh, just think about, like, how, you know, it's gotten-- the world's gotten more complicated. You know, I think there's a lot of opportunity for, um, it to feel overwhelming, the amount of complexity for people.
Ben Goodger 8:51 Well, actually, I, I think, you know, even going back a few years, like I remember when I was a kid, uh, and my, my school friends and I, we'd trade, uh, shareware with each other-
Darin Fisher 9:01 Mm
Ben Goodger 9:01 ... on floppy disks, as you did back at that point. Um, and my mind is just not meant to retain certain types of information, and so all of the things like the command line for how to, to run, like your unzip tool or, you know, however that worked, this just was just something that felt entirely bizarre to me. And so in that sense, the web was kind of a really refreshing take on that. It was something where I could just go and click on things and, and explore without having to sort of understand the underlying nature of the machine.
Darin Fisher 9:27 And view source.
Ben Goodger 9:28 That's, that's true. [chuckles] Um, but, uh, what I, um, what I find with, uh, you know, where the future of technology is going, um, is that, uh, you know, when we have these AI assistants that are attached to your computer, I think we'll find that we make that computing capability much more accessible to more people-
Darin Fisher 9:47 Mm
Ben Goodger 9:47 ... who aren't necessarily experts. You know, even in, you know, maybe not just even in terms of, like, how an operating system works or how a browser works, but even how individual websites work. You can express yourself more naturally as to, like, how your int- what, what your intent is, what your goal is, and then the system can kind of figure out, you know, how best to accomplish that, uh, for you.
From “semantic web” dreams to models that meet the web where it is
Andrew Mayne 10:08 It seems like there was a lot of ideas about what the web could be, and, and part of it is that a lot of standards were things that were sort of decided after the fact, or we had to sort of go here and do that, and then there was things that would've been nice to have, like the semantic web.
Darin Fisher 10:22 Mm.
Andrew Mayne 10:22 If we made sure that everything was sort of annotated, and you did that, but just in the real world, in the corporate world, it's often hard to do that. And now, did you think you'd see a world where, hey, we could have LLM sort of understand this and make that possible?
Darin Fisher 10:33 Well, I think it's amazing to see-- this is, like, kind of the magic of, of these modern AI models, is that they are really, um, able to interact with things the way we interact with them-
Andrew Mayne 10:43 Mm
Darin Fisher 10:43 ... so systems, and that's, that's, that's- of course, the world is designed for people with e- eyeballs and, and mouth, and on the com- on, online for people with, uh, mice to click on things or-
Andrew Mayne 10:53 Mm-hmm
Darin Fisher 10:54 ... you know, fingers to tap on things, and so when you talk about the semantic web, of course, it would be really nice if people would publish websites that were more, you know, inherently understandable by ma- machines, but their motivation is to go where the users are. And as much as we, we make a push for websites that are more accessible and, uh, and, uh, you know, screen readers and whatnot, the reality, of course, is that that's just not where, um, developers, uh, spend their energy first and foremost, right? And it's usually an afterthought to make sure that you make everything extremely accessible. So kind of the beauty of these AI models is that they kind of meet the technology where it, where, where users are. You know, technology is designed for, uh, people to consume, so you take a look at the way that it, uh, interacts with language and understands language. It can interact with language the way we interact with language. Um, self-driving cars, they can interact with the roads and the, and the systems of transit the way that we interact with them. Um, and I think that's kind of the beauty of these AI models, is that they can, they can be developed to, uh, for this world that was designed for humans, and so that extends naturally to the browser.
Ben Goodger 11:58 I think it's, um-... It's not necessarily that we predicted exactly a hundred percent how all of this would play out, but there are some very powerful ideas in that original internet.
Darin Fisher 12:07 Mm-hmm.
Ben Goodger 12:07 Um, where, uh, like there was this idea that it was, you know, structured data, uh, that would be read and interpreted by a machine, and then presented to the user in some way. And so with the original web, there was this idea of a user's agent, a user agent-
Darin Fisher 12:21 Mm.
Ben Goodger 12:21 - and that's a browser.
Darin Fisher 12:22 Yeah.
Ben Goodger 12:22 Uh, and then that, you know, takes that, that machine-readable content and then applies some presentation preferences to it. Maybe you like your font to be a little bit bigger, so it's easier to read. Maybe you like it a certain style or weight or other stuff like that, and it does it. That was back in the original web-
Darin Fisher 12:37 Mm
Ben Goodger 12:37 ... was that idea. Uh, and so I think that that carries forward in today's era actually very, very well, and you can view where we are today as just sort of the natural, um, endpoint or a continuation of that, that journey.
Darin Fisher 12:49 Evolution of the user agent, right?
Ben Goodger 12:50 Mm-hmm.
Darin Fisher 12:51 So back in the day, you, even in Netscape browser, you could write a cut-- what was called a custom style sheet or a user agent style sheet. You could override the colors of any website, and people maybe were, uh, who were more sophisticated would know how to do that. Later on, browser extensions kind of made that a more universally available kind of thing. People could write an extension, share it with other people more easily. But it just makes a lot of sense now to, like, empower an LLM to be able to go and on your behalf, as, as Ben said, to, to really supercharge that user agent, to be able to do more things on your behalf.
Andrew Mayne 13:23 Yeah, there's, there were a lot of interesting ideas, I think, going back at kind of at the dawn of all of this, and you look even the names of some of the tools like Gopher and Watson-
Ben Goodger 13:30 Mm-hmm
Browsers are complex “mini operating systems,” and Atlas goes beyond a plugin
Andrew Mayne 13:30 ... and Sherlock and whatnot, and how it was kind of more proactive, the idea that, as you mentioned, these are sort of tools that don't just, aren't document viewers, and I think that we kind of take the browser for kind of granted, and the idea that it just sort of like, it just shows me the website. And I can see where it's helpful for you two probably haven't spent most of your careers working in the space of browsers and trying to understand that, and it seems like there's like an inordinate amount of complexity there that's invisible. Could you give me an example of like, you know, the kinds of things you have to deal with when you're trying to figure out how to make something work?
Darin Fisher 14:01 Oh, man, browsers are maybe surprisingly complex. I don't even know where to begin. I'm, I'm amazed at how much, um, work there is and how much technology goes into building a browser.
Ben Goodger 14:12 It's basically like a, an app platform or a mini operating system that's running on your desktop. So everything, every discipline of computing, it feels like there's like you can nerd out on some-
Andrew Mayne 14:22 Mm
Ben Goodger 14:22 ... aspect of the browser.
Darin Fisher 14:24 I was just having a conversation over lunch with one of the engineers on our team, explaining how, um, our Owl re- uh, Owl, Owl works, which is our embedding of Chromium that runs out of process.
Andrew Mayne 14:34 Mm-hmm.
Darin Fisher 14:35 Uh, I was explaining the rendering model for this, and it kind of led to a conversation about how Chrome first worked when we first built it, and then how, uh, the advent of, uh, GPU accelerated rendering e- evolved. And now fast-forward to the way it all works today and what we're trying to do with Owl, and, um, and just sort of the depth of complexities there, and, and, uh, yeah, I could go on and on, actually. [chuckles]
Andrew Mayne 14:58 Well, it, it is interesting because I think that people kind of overlook, like kind of the browser wars led to technologies like, uh, running node on servers-
Darin Fisher 15:09 Sure
Andrew Mayne 15:09 ... to actually do stuff, which is, I think, nobody even thought would happen, but because just trying to make these things much more efficient and faster. And I, I, again, I, I kind of, you know, marvel that anything works at all, and looking at some of the technical stuff you have released on ChatGPT Atlas and understanding it's not just a plugin, it's not just a thing that adds a ChatGPT sidebar, could you explain a little bit more about the architecture?
Ben Goodger 15:36 So from a design perspective, I would say that we wanted to imagine the entire experience with ChatGPT at the heart of this thing.
Andrew Mayne 15:43 Mm-hmm.
Ben Goodger 15:43 Uh, not just, I mean, w- you know, we have a, a ChatGPT extension as well that you can install on Chrome-
Andrew Mayne 15:48 Mm
Ben Goodger 15:48 ... for example, but there are some limits to what that can do. And so when we approached this from a design perspective, we wanted to just be really empowered to look across the entire browsing surface, and I think, you know, maybe like, you know, two to five percent of what we want to do [chuckles] is there today. Um, but we think that this being able to, to own the whole browser experience in this way gives us the opportunity to, to weave that ChatGPT magic, uh, in throughout, and that includes things, for example, like on any, um, uh, uh, text field, uh, you can sort of invoke ChatGPT and have it help you write.
Andrew Mayne 16:21 Mm-hmm.
Ben Goodger 16:22 Uh, and then as it is helping you write, that is, you know, your personalized ChatGPT. You know, it can sign, uh, your email as you because it knows you from your other use of ChatGPT. Uh, and so being able to, to build it in a way that en- enables these very like richly integrated use cases, uh, felt, you know, it felt very important to have a browser as opposed to, um, you know, just an add-on-
Andrew Mayne 16:43 Mm
Ben Goodger 16:43 ... for an existing app.
Darin Fisher 16:44 And it gets to some of the, like, the foundational stuff. When ChatGPT is at the core of this thing, and you, you enable things like, um, the memories feature-
Ben Goodger 16:52 Mm
Darin Fisher 16:52 ... it means that you can ask this thing, "Hey, what was that thing I was looking at again?" And it's gonna know, and it's gonna help you. Like, uh, I, who hasn't had this, this sort of experience of: "Oh, yeah, I remember I saw some video or I saw something. What was that again? How do I get back to it?
Ben Goodger 17:08 Mm.
Darin Fisher 17:09 I want to share it with somebody else." Traditionally, you might go through your browser history or your YouTube history and scroll through there trying to find it, or you're like: What was that tweet again that I saw? Or I was looking at a recipe with my son on the weekend. It, what was, what was the third buffalo wings recipe that we found-
Ben Goodger 17:26 [chuckles]
Darin Fisher 17:26 ... that I wanted to make? How do I find it again? And just to be able to easily recall it because this thing is able to do that so effortlessly. It's a side effect of it just being there.
Agent workspace: separating your tabs from the agent’s tabs (and why it matters)
Ben Goodger 17:35 Probably the most, and probably the, um, probably what I would say is the biggest advantage for doing this, and especially in the way that we've done it, is, um, how we thought about integrating this agent-
Andrew Mayne 17:47 Mm-hmm
Ben Goodger 17:47 ... capability. Uh, and this is really where having our own concept of browser-
Andrew Mayne 17:53 Mm
Ben Goodger 17:53 ... including what it means to have a collection of tabs. If you think of your regular browser window, that's a collection of tabs, then you can also imagine that your agent has a collection of tabs. Maybe each instance of the agent that you've chosen, that you've asked to go off and do something for you, you might have, like, five of them, each running on, on different problems, and each one of those has its own collection of tabs. And, of course, they're not showing up on your top tabs because you didn't open them, uh, but it nonetheless has them, and it's working through them, and it's getting information from them, and it's processing it and taking action on your behalf and clicking on things and all that sort of stuff.... that is the sort of, um, functionality that you can have when you go and you design a system like this, sort of end-to-end. You can invent all these abstractions.
Darin Fisher 18:33 And you had this in the very, very first version. I remember when I joined Ben to work on this, and he had, uh, already had sort of this idea, this, this idea that we would be able to segment tabs between the tabs that are the user's tabs and that tabs that are- that the agent is working on, uh, for your behalf. And that kind of shows up in the product today. When you, uh, start an agent task, you go- it goes off, and it's gonna work on whatever you asked it to work on, and it might need to open some additional tabs. And instead of those tabs just appearing in your tab strip and, and perhaps, uh, feeling a little d- uh, discombobulating because, like, where are all these-- what are all these tabs?
Ben Goodger 19:10 Mm-hmm.
Darin Fisher 19:11 Um, it, it has just sort of accumulated some work in the background, and when it's done, then it presents it to you in a tabular form. You can go and click through and see what it actually did, or you're just happy with the outcome, and you really didn't need to see all the intermediate steps. And so I think it's kind of powerful, this idea that it- uh, the agent has its own workspace.
What counts as an “agent task”: acting on the web while you watch and learn
Andrew Mayne 19:29 I would say, for many people using this, on the surface, it seems pretty easy to understand. I've got a browser, I've got ChatGPT, but we also have agent mode. What is an agent task? What is not? What would be a ChatGPT task? Could you explain that?
Ben Goodger 19:41 Yeah. So, um, uh, agent mode is basically you're inviting ChatGPT to take action on the web, uh, on your behalf. Uh, and so maybe you are, um, looking at a website, uh, and you would like to do something on that site, but you're not quite sure how to do it. What you can do is you can just ask ChatGPT directly. Uh, if you're-- let's say, you're on a spreadsheet, and you want to synthesise a pie chart, and you don't know how to do that, you can just say: "Hey, make a pie chart with this data." Uh, and it'll go off, and it will figure out how to use that software. Now, if you could think about sometimes some of the software that you use, it can be pretty complicated. So just being able to ask in natural language, in your own words, what you want to see, and then you can just sort of si- sit back and watch it take over, and it starts moving the mouse around and doing stuff like that for you. It's pretty, pretty amazing.
Darin Fisher 20:24 And, and, and you, you could see how it's going to do it.
Ben Goodger 20:26 Mm.
Darin Fisher 20:26 So actually, you can learn how to now-
Ben Goodger 20:28 Mm
Darin Fisher 20:28 ... make that pie chart 'cause it's gonna show you, which is pretty cool.
Andrew Mayne 20:33 I found it pretty useful. Like, I like, I like to study memory methods and stuff, and I have to do a thing where you have to have, like, a system for decks of cards and stuff, and I didn't want to have to paste in a bunch of, like, card emojis, and I'm like: "Can you just go do this for me?" And pff, it's like magic. It's like witchcraft.
Ben Goodger 20:49 One of the fun, fun things I've seen people do around the office is they'll have written a doc, and then they'll ask it to take a review pass on their doc and, like, add some comments.
Andrew Mayne 20:58 Mm.
Ben Goodger 20:58 And so it'll actually go into your doc, and it can use sort of the integrated commenting tool of whatever document editing system you're using, and it will just add comments as if it was a collaborator. [chuckles]
Darin Fisher 21:07 Yeah, it's really amazing. I mean, just we-- of course, being software engineers, we experience the model, um, critiquing our code.
Andrew Mayne 21:13 Mm-hmm. Right.
Darin Fisher 21:13 And over the past year, just amazing to see how much better it's gotten at that. Like, in the beginning, and it was not always the case, that it would find things that were useful, uh, but it's-- these days, I'm like: "This thing is sharing important, like, amazing nuggets that are saving me from, uh, you know, shipping bugs." And I can just- you know, I can see the analogy to, like, reviewing any document that I might have and just asking it to, to go and give me some, some style feedback or, you know, maybe some grammar suggestions or like, you know, tone suggestions, and I, I just- it's, it's very exciting to think about how any tool you're using on the internet, you could ask this- invoke this agent and ask it to interact with that in the same way that I would or somebody else would, and just see what it would do. And maybe you learn from it or, or, or, or whatnot.
Building Atlas with AI: Codex-driven productivity and faster platform expansion
Andrew Mayne 22:00 One of the things that's come up a lot during my conversations with the, the teams inside OpenAI is how much they're using the tools. You know, they're using GPT-5, GPT-5 Codex to do that.
Ben Goodger 22:13 Mm.
Andrew Mayne 22:13 And how has that affected you, and do you think that's gonna create an accelerated product cycle?
Ben Goodger 22:18 So, uh, so absolutely. [chuckles] A couple, a couple of, uh, anecdotes on this one. Um, you know, uh, one of the top Codex users at the company is on our team, and they're just sort of like r- raw productivity in terms of like PR output is like off the charts, you know, as a result of using this tool. And, you know, it's really exciting to see what, what-
Darin Fisher 22:38 Good PR output.
Ben Goodger 22:39 Yes.
Andrew Mayne 22:39 Yeah.
Ben Goodger 22:39 So, so it's really exciting to see what experienced engineers can do with these tools because they can- it can both sort of explore an area, help you explore an area, decide if something is worth doing, uh, and then you apply your judgement, you kind of tell it what you want it to do, and it goes off and does it. And then for folks like me, who maybe spend less time coding than I used to, I can also have it go off and, and, and prove some stuff out. So, like, everyone on our team is able to contribute. You know, our product managers are producing PRs, our designers are producing PRs, um, because of these tools, and so I'm like a true believer.
Darin Fisher 23:11 Yeah, I did a, uh-- for be- before GA, I did this, um, week-long refactoring to try to unlock, uh, a certain, uh, feature that we were trying to ship, and then I had one more to do, and this time I asked Codex to do it, and it was done within- the whole project was done within an hour, and it was of similar scale. And I was able to tell it, "Hey, just do this other one, kinda like I did that one." So I had shown it the way, and I just asked it to do this other task that was very similar, and it was, it was almost one shot.
Andrew Mayne 23:38 One of the, the promises of really capable code tools that are able to write multiple languages, porting code from one language-
Darin Fisher 23:46 Oh, yes
Andrew Mayne 23:46 ... to another. We saw with Sora and how, you know, they said, "Hey, Android is coming-"
Darin Fisher 23:50 Bringing it to Android. Yep.
Andrew Mayne 23:51 Yeah, and people were like: "Oh, great. When will that..." They're like, "Oh, no, it shipped."
Darin Fisher 23:54 No, no.
Andrew Mayne 23:54 It was weeks.
Darin Fisher 23:54 I mean, actually, ChatGPT's been amazing at doing, um, cross-language translation for quite a while-
Andrew Mayne 24:00 Mm
Darin Fisher 24:00 ... I mean, coding languages. So, um, and, uh, you know, we're building, bringing this product to Windows now, and, uh, we're actually gonna be using Swift, which because we are a bunch of, uh, the team is full of Swift experts, and we, we, we are excited to have a shared common code base and, and, uh-
Andrew Mayne 24:16 So you're gonna use Swift in the Windows?
Darin Fisher 24:18 Yeah, Swift-
Andrew Mayne 24:18 Interesting
Darin Fisher 24:18 ... in Windows. And so, uh, the-... I'm not, I'm, uh, you know, I think it's very, in my p- past experience, I've s- I've been- I was marvelling even years ago at just how good ChatGPT was at being able to essentially generate a code for me in Swift that was, uh, that did not exist on the internet.
Andrew Mayne 24:37 Mm-hmm.
Darin Fisher 24:37 So it was taking code that maybe was written for .NET-
Andrew Mayne 24:41 Mm-hmm
Darin Fisher 24:41 ... and, uh, could, could translate it to Swift for me, and I was just, I was marvelling at its capability or, or to generate a bunch of obscure WinRT code that's norm- normally a very tedious C code with, uh, lots of GUIDs and all kinds of things that are very detailed, uh, but it was all just spitting out this code and saving us just an enormous amount of time.
Andrew Mayne 25:04 Yeah, I had Codex CLI spin up a Swift app without ever having to go into Xcode to paste anything in, and it was worked just right out of the box.
Darin Fisher 25:12 Yes.
Andrew Mayne 25:13 It really impressed me.
Darin Fisher 25:13 So it's, it... I mean, it, it's, it's like, kind of like one of the strengths of these models. So the fact that, you know, if you can ask the right question, get the right prompt-
Andrew Mayne 25:22 Mm
Darin Fisher 25:22 ... and it can, and if it's on the right path on, like, how to build something, uh, it can do it in any language. That's really not a problem.
Safety and user control in agent mode: sensitive tasks, stop buttons, and auth choices
Andrew Mayne 25:29 So I've been using this, and I've been switching it to agent mode to do stuff, and I know sometimes I can just leave a tab and go off and do something else, and sometimes it's like, "Hey, if you leave this tab, I'm gonna pause." What's going on there?
Ben Goodger 25:41 Yeah, so this is, um, sometimes you'll have asked the agent to do something that's very sensitive. You know, an example is it's looking at your email-
Andrew Mayne 25:48 Mm.
Ben Goodger 25:49 - uh, and we would like you to keep your eyes on the road, so to speak. You know, I have a car that has, um, sort of an auto-drive feature, and it wants me to pay attention to the road, and it's, it's helping me drive, but it's not gonna let me, like, check my phone or, you know, take a nap or something like that. And so it has a little camera that's watching my eyes, uh, it's making sure that I'm paying attention. So you can kind of view this sensitive mode and, and agent mode as kind of like that. It wants me to pay attention to this tab while it completes, just so that I feel like I have a good level of control over it. And in fact, if you look at the bottom of the tab, you'll see that there's a little bar that has a big red stop button in it, and if you've ever, ever been in a machine shop, you'll know that the machines there-
Andrew Mayne 26:27 Mm
Ben Goodger 26:27 ... have these big red buttons on them. If suddenly it starts to do something that you don't want it to do, you just whack that button, and it stops, and so that's the idea. You just, um, yeah, just, just keep an eye on it, uh, watch it go, uh, and you can always take over, uh, if you wanna do it yourself.
Darin Fisher 26:41 Yeah, we put a lot of thought into making sure that, that these features, um, help you feel in control of the experience and take away some of that maybe uncertainty when you wanna go use them. For example, on top of what Ben mentioned, there's also the signed-out, um, uh, method of using agents. So if you wanna start without it being, uh, in an authenticated session, meaning it doesn't have the cookies required to even access your email, you could do that, and I think that can be a great way to kind of try some things-
Ben Goodger 27:09 Mm
Darin Fisher 27:09 ... and do, do it where you're learning, like: How does this thing actually work? And you might, uh, then, uh, hit a point where in order to take the next step with Agent, well, actually, it would be helpful if it were authenticated. And so then you might try, uh, doing a task where it does benefit from having your cookies.
Ben Goodger 27:26 You might not remember, probably the first time you actually run Agent, it, um, shows the screen. It explains how all of this stuff works for you.
Darin Fisher 27:33 Mm.
Ben Goodger 27:33 So you can... Like, if you read through that, you'll see sort of the choices that you have, and you'll learn about how to, to use them.
Darin Fisher 27:38 Yeah.
Andrew Mayne 27:39 I, uh, have about a half million unread email. I'm ready to go full auto. It can't do a worse job than I have so far.
Ben Goodger 27:45 You might find it-
Andrew Mayne 27:46 That was-
Ben Goodger 27:46 ... just hit Select All, Archive.
Andrew Mayne 27:48 Yeah, but it wouldn't, [chuckles] be the worst, you know?
Darin Fisher 27:50 Declare bankruptcy.
Andrew Mayne 27:51 Yeah, I could have somebody to blame. Uh, but it has been- it's been super helpful for me because there have been times when I'm trying to surface an email, and the keywords just don't work, or the results-
Darin Fisher 28:02 Mm
Andrew Mayne 28:02 ... are too many, and just being able to go in and say, "Find this thing about that."
Darin Fisher 28:06 Yeah.
Ben Goodger 28:07 Yeah.
Andrew Mayne 28:07 It's saved me numerous times.
Ben Goodger 28:09 And so one of the other things that we did, uh, with Agent as well, 'cause there are times when you'll wanna keep an eye on it, is that we, uh-- and I'm, I'm pretty proud of what the team was able to come up with here, is to make it, like, very visually compelling.
Andrew Mayne 28:20 Mm.
Ben Goodger 28:20 Um, and so there's, you know, all of the little, little sparkles and, and pixie dust and so on that appear around it as it, as it's working. Uh, it's pretty cool. Uh, so yeah, love to, to see more people try it out.
Andrew Mayne 28:31 Yeah, it's fun to watch that, and, and also with ChatGPT, too, in agent mode, I've, I've said that I could probably watch a live stream of just watching these systems solve problems-
Ben Goodger 28:40 Mm-hmm.
Darin Fisher 28:40 Mm
Andrew Mayne 28:40 ... 'cause it's, it's like how it's made, but-
Ben Goodger 28:42 Mm-hmm
Attention, ecosystem, and serendipity: expanding the web without getting trapped
Andrew Mayne 28:42 ... watching computers do it. Uh, it, it's exciting to see where this is headed. I also sort of wonder, what's it going to be like when, you know, there's gonna be a lot of different AI-powered browsers out there, and also we have to think about, like, what is, what is the ecosystem like when most of my tabs are opened by my agent and not me, and somebody's trying to capture my attention?
Ben Goodger 29:03 The way Agent works right now is that it's only g- it's only, like, running in response to your request.
Andrew Mayne 29:10 Mm-hmm.
Ben Goodger 29:11 Um, and so if the agent is doing something, it's because, you know, at some level you asked for it.
Andrew Mayne 29:16 Mm.
Ben Goodger 29:18 Um, pages that it opens actually have some limitations. Um, so, like, you might be used to browsing around the web, and you'll see some page, you know, show a pop-up window saying-
Andrew Mayne 29:27 Mm
Ben Goodger 29:27 ... "Give me, like, notifications permissions-
Andrew Mayne 29:29 Mm
Ben Goodger 29:29 ... so that I can sort of spam you with updates" and that kind of stuff.
Andrew Mayne 29:31 Never, ever clicked on that anybody.
Ben Goodger 29:32 Yeah, no.
Darin Fisher 29:32 Yeah.
Andrew Mayne 29:33 Nobody's ever intentionally clicked on that.
Ben Goodger 29:34 Yeah. Um, the agent tabs can't do that.
Andrew Mayne 29:37 Mm.
Ben Goodger 29:38 Uh, they're actually blocked, uh, from doing... So there's a bunch of stuff like the design of the system-
Andrew Mayne 29:41 Mm
Ben Goodger 29:42 ... uh, to avoid you accidentally ending up in that, that state.
Andrew Mayne 29:45 Mm.
Ben Goodger 29:45 Now, of course, you're free, you know, when you browse to a website, and you're asked for a notification permission, if you want to receive updates from that site, it might-
Darin Fisher 29:52 It might be your calendar.
Ben Goodger 29:53 Right.
Andrew Mayne 29:53 Mm-hmm.
Ben Goodger 29:53 You're free as a user to go and, like, say, "Uh, yes, I want this," but the agent won't ever do that, uh, on your behalf.
Darin Fisher 29:59 Mm-hmm.
Andrew Mayne 29:59 It's an interesting world where we think about part of the beauty of the early web was the serendipity.
Ben Goodger 30:04 Yeah.
Andrew Mayne 30:04 "Oh, I found this other thing. I found these other links." But I think then that kind of got sort of weaponised against the user, where basically you try to do a thing, and it's hard to do a thing, and I think we got to a place-
Darin Fisher 30:14 Yeah, it's, it's really interesting. I don't know if this is where you're going with that, but, like, a lot of websites wanna just keep you on their website.
Andrew Mayne 30:20 Mm.
Darin Fisher 30:20 Maybe they'll run ads, which would take you off, but otherwise, they kind of keep you, you know, in that lane. One of the amazing things about the si- Sidechat or the model being present there and, and the agent even, is just that-... you can ask it questions about that site, where the answer might be something on a different site.
Ben Goodger 30:37 Mm-hmm.
Darin Fisher 30:37 And so it makes the web bigger for you and helps you, uh, not just be stuck down that, that rabbit hole that you were on, but to help bridge you to something more useful to you, maybe, or more helpful. I know you have some good stories on this.
Ben Goodger 30:49 There's something just, um, wonderful about the- I, I call it beautiful chaos of the web-
Darin Fisher 30:55 Mm
Ben Goodger 30:55 ... where you kind of don't want to always be stuck in the same place.
Darin Fisher 30:58 Yeah.
Ben Goodger 30:58 You want to be able to embrace the diversity of the web-
Darin Fisher 31:00 Mm-hmm
Ben Goodger 31:00 ... and all of the content that's on it. So yeah, I love that. I love that you can do it as well without having to leave the site.
Darin Fisher 31:06 Yeah.
Ben Goodger 31:06 You know, you can do it right there on the side, and then you can choose to, to go somewhere else. But it, it sort of, um... There's this aspect of, uh, I know with Wikipedia, you can go on these, like, multi-hour journeys through content. Um, that's really only a- that's, like, a feature of Wikipedia, whereas I feel like-
Darin Fisher 31:20 Yeah
Ben Goodger 31:20 ... the Ask ChatGPT sidebar gives you that ability for, like, the web at scale, and so it gives you the ability to ask questions about random sites and then go off in different directions and that sort of thing.
Darin Fisher 31:29 I mean, this is extremely useful if you're looking for certain kinds of products, and you find yourself onto one product page for one company-
Ben Goodger 31:35 Mm
Darin Fisher 31:35 ... or one, you know, vendor, but now you can be like: "Well, what else is out there?" And the model can say, "Well, here, here are some other sites to go check out that is related to this," and off you go. Now, your, your world has gotten bigger, right?
Andrew Mayne 31:47 I've had some-
Ben Goodger 31:48 Great
Andrew Mayne 31:48 ... wonderful discoveries with both videos and books that I couldn't find through the YouTube search engine or the Amazon search engine.
Ben Goodger 31:53 Mm-hmm.
Andrew Mayne 31:53 I found places where, because ChatGPT understood a bit more about what I was looking for, what I was really trying to find, and that was, like, it, it gave me more utility out of those sites.
Ben Goodger 32:03 Yes.
Darin Fisher 32:03 Yeah.
Browser memories and personalization: recall, context-aware search, and controls
Ben Goodger 32:03 That's actually another, um... Like, so I, I- for ChatGPT as a whole, like, the personalisation features, the fact that it sort of learns more about you the more you use it, uh, has been, like, a super popular feature of ChatGPT. With Atlas, you know, this extends to, to this, your browsing activity, uh, like your, your sort of web history. Um, and this- and, and so this allows the browser to create these, these browser memories-
Andrew Mayne 32:28 Mm-hmm
Ben Goodger 32:29 ... um, which Darin pointed out before, is kind of, uh, something you can use to help you get back to a site if you kind of can't remember it later. But it also helps in situations like with the agent. Um, you know, I'm a United MileagePlus member, and so I tend-
Andrew Mayne 32:42 Mm
Ben Goodger 32:42 ... to like to look for flights on that site.
Andrew Mayne 32:44 Mm-hmm.
Ben Goodger 32:45 Um, it would be very tedious if every time I ask the agent to go and do something like that, if I had to tell it, "And always use United Airlines."
Andrew Mayne 32:53 Mm-hmm.
Ben Goodger 32:53 Uh, but it kind of knows from my browsing, my browser memories, that I'm a frequent user of United, and so it will just go there, uh, by default.
Darin Fisher 33:00 Yeah, it helps you in the forward queries, right?
Ben Goodger 33:02 Yeah.
Darin Fisher 33:02 Because it's, it's like s- now, now this, uh, search experience has so much more context-
Ben Goodger 33:07 Mm
Darin Fisher 33:07 ... about what matters to you. So it just, it ends up being a lot more efficient.
Ben Goodger 33:12 Right.
Darin Fisher 33:12 Saves you a bunch of time 'cause you don't have to tell it as much again and again.
Ben Goodger 33:15 So I feel like that's something that helps me a lot. I think some people probably have, you know, different preferences around the- these things as well, so there are controls where people can go and see and control what, um-
Darin Fisher 33:25 Yeah
Ben Goodger 33:25 ... what memories are, are used-
Darin Fisher 33:26 You can turn it off
Ben Goodger 33:27 ... for personalization.
Darin Fisher 33:28 If you want.
Ben Goodger 33:28 You can turn it off entirely.
Search inside the browser: familiar entry points that reveal “model superpowers”
Andrew Mayne 33:29 And you also have to, when you use it, and I don't go into these tabs as much, I'm starting to do that more 'cause I realise they're there, is I can go to Images, I can go to News-
Ben Goodger 33:38 Mm
Andrew Mayne 33:38 ... and I have kind of like a search engine, and that's... The, the thing I'm trying to sort of understand is, it's a browser, but it's also OpenAI heading towards its own search engine.
Darin Fisher 33:46 Well, part of that comes from the fact that, um, when you're building a browser, people come to that browser with existing intents, like navigational intents or the, the, the idea that they do want to look for images, right?
Ben Goodger 33:57 Yeah.
Darin Fisher 33:57 Or they want to see a certain kind of, uh, subset of information, and so we brought those, those controls into the, the landing page of chatgpt.com, uh, so that it would be both familiar to people but also useful in the way that they're, they're used to, right? Uh, we want to make sure people didn't feel like they're so out of, um, you know... That, that, that we didn't want people to have to learn so many new things in order to be successful using this product. We wanted them to have a lot- a, a good dose of familiar tools and familiar sorts of things, and plus- and anyways, these are just useful. For many people, search and browse are just very connected.
Ben Goodger 34:32 Yeah.
Darin Fisher 34:32 Almost one and the same, and, um, you know, it's- so it's very important to-
Ben Goodger 34:37 Mm
Darin Fisher 34:37 ... to internalise that as we're building this experience. Uh, I think it's very powerful that, and I was touching on this before, that as people search and use the browser in maybe a very normal way, they're also learning about, uh, that there's a model there that's gonna respond to them. So you get the- a set of chips across the top, which is like quick links to go to where you were trying to go, perhaps, uh, these, these different tabs where you can click on to see, like, familiar different kinds of subsets of information, but also this model response coming in. And so you start to... vanligly using a product in a normal way, you start to learn that there's a, there's another way-
Ben Goodger 35:11 Mm
Darin Fisher 35:11 ... or that there's a superpower that this thing's providing. And it's- some of it is just the normal ChatGPT experience that people are used to, but not everybody's using ChatGPT to the fullness. And so when you- and when it's core and central to the experience, we have an opportunity to present that to people as part of their normal journey, and I think that's really cool. Similar to Sidechat, um, of course, you have to activate it, but it's right there, Ask ChatGPT, and you might be curious, and now y- you kind of unlock this superpower, but it's right there. You- the context is there.
Andrew Mayne 35:41 The initial experience for me was the, the very first day when I started using it, and I look at this, I'm under- trying to understand, okay, this is basically... It's, it's an app that has a browser and ChatGPT. It's not like we just sort of glued those things together. It's sort of like they're both there, and there's a deep connection to the ChatGPT, and I asked it to, "Could you add a bookmark for-
Darin Fisher 36:02 Yeah
Andrew Mayne 36:02 ... you know, Amazon?" And then a moment later, the bookmark appeared, and that was a really kind of special moment to sort of understand what happens when your, the LLM deeply understands the system, you know, and is able to make those kinds of changes.
Ben Goodger 36:16 I think we're, um, very excited about this. I think from a, a, just, like, a conceptual transformation point of view, [chuckles] we're moving to a world where you can just tell the computer what you want-
Darin Fisher 36:28 Mm
Ben Goodger 36:28 ... in, like, whatever way you want to tell it, uh-
Darin Fisher 36:30 Simplest way possible-
Ben Goodger 36:31 Right
Darin Fisher 36:31 ... just tell it.
Ben Goodger 36:32 And so w- what this means for making computing more accessible to more people is just, like, really profound, and that's, like, the company's mission is to make AGI beneficial to all of humanity.
Andrew Mayne 36:41 Mm-hmm.
Ben Goodger 36:41 We take that really seriously, and I think-... being able to transform computing in ways like this that might seem very small on, on face of it, they add up to something, you know, far more profound.
Darin Fisher 36:52 Yeah.
Ben Goodger 36:52 Uh, and so, yeah, that we're, we're excited about that kind of thing.
Darin Fisher 36:54 Some of my first experiences with ChatGPT as a user was really this idea Ben's talking about. You know, I was comfortable and happy just sitting there, uh, doing my Google searches, but sometimes I didn't quite know what Google query to type in.
Ben Goodger 37:06 Mm-hmm.
Darin Fisher 37:06 And, [exhales] you know, when I, when I realized I could ask sort of a really poorly formed question to ChatGPT, and it would come back with- make some sense of what I asked, what I said, and it would, it would give me something that maybe now I could query Google for. And that's how I first started using ChatGPT, and then I started to realize over time that, like, "Oh, why am I not just asking it in the first place?" You know, and, and it's sort of- I think for people, they all have like a-- there's like a bit of a journey with new technology, right? You're y- we're all creatures of habit. We're used to the things the way we work, and, and it works well for us, the things we're used to. We're used to it, so it's not a problem. And but as you maybe explore something new, you start to see, oh, there actually is a better way. And for everybody, that journey's a little different. And so for me, one of the things I was most excited about with, uh, Atlas was this idea that, uh, when you're typing into the address bar, that, that the default is ChatGPT. Because for me, that actually makes sense for most of what I'm gonna do. And-
Ben Goodger 38:02 This is one of these things where I feel like now when I don't have access to that-
Darin Fisher 38:06 Correct
Ben Goodger 38:06 ... I feel like there's, like, this little bit of, um, friction.
Darin Fisher 38:09 Mm-hmm.
Ben Goodger 38:10 Uh, like, it takes longer now, because then I've gotta f- oh, I've gotta go find my ChatGPT tab in another-
Darin Fisher 38:15 Yeah
Ben Goodger 38:15 ... browser and, like, figure out how to get at that and do that.
Darin Fisher 38:17 Yeah.
Ben Goodger 38:17 Whereas, uh, with Atlas, you can just, like, open a new tab and start typing.
Darin Fisher 38:20 The old way was a much more manual way. This way is a much more- I don't have to be as, like, clever about what I ask. I can just, I give some, some sort of-
Ben Goodger 38:28 I have a problem.
Darin Fisher 38:29 Yeah, [chuckles] I have a problem.
Ben Goodger 38:30 Yeah.
Darin Fisher 38:30 I can say it in a much more simple way.
Andrew Mayne 38:31 I- I know that I'm still having trouble kind of context switching and understanding that it's not just a URL search bar or just an empty keyword search, whatever, that literally I can ask it for things, and not just have, you know, you know, "Well, what is the capital of Nepal?" And not just have that pop into a Google search box.
Darin Fisher 38:48 Right.
Andrew Mayne 38:49 And that's the thing now, I'm like, "Oh, yeah, when I go type in the thing, if I type in the URL, I get the URL, but I can also type in kind of my query and do that," but that's still taking me time to adjust.
Ben Goodger 38:59 Well, just as a, um, [tsk] like a general rule, like I find sometimes modes can be a reflection of some of the limitations of the system underneath that, that, you know, at the end of the day, humans don't understand.
Andrew Mayne 39:11 Mm.
Ben Goodger 39:11 Um, and so I think the north star for us with so much of this stuff is, can we just help you arrive at the right place? You know, regardless of you needing to know, ah, I should put it in this mode, or I should put it in that mode. Like, that is sort of the struggle. It comes down to, like, how do you want to use this tool? And so, uh, we wanna make this thing something that, um, you know, if you just go in, as Darin was saying before, you know, you can just kind of tell it what you want, maybe this half-formed thought, and it will- will give you something good.
Andrew Mayne 39:37 Mm.
Ben Goodger 39:37 It will help you figure out the problem. Now, of course, there are ways that, you know, if you are, um, [tsk] a user that understands some of the underlying capabilities of the system, we wanna give you the option to invoke those too, to bring them down and help you. Um, and that, that's sort of an efficiency gain, uh, that you can get, but the certainly the system shouldn't require that you know all of those sort of incantations. It should be able to just take what you say to it and give you something good.
Darin Fisher 40:02 Yeah, I think it, it reminds me back to, like, the early days of browsers, um, in the era when people would install, like, a toolbar for their search engine.
Ben Goodger 40:09 Mm-hmm.
Darin Fisher 40:10 And y- you know, and that meant they had yet another box on their browser, right?
Ben Goodger 40:14 Mm.
Darin Fisher 40:14 And Firefox had a dedicated search box for doing your web search. Uh, eh, but back then, um, as much as people were very used to that and very comfortable with it, you know, you have one box to type URLs and one box to type search queries. Um, when we were working on Chrome, we were like, "Why? Why have two boxes? Why do people have to stop and think about which box-
Ben Goodger 40:33 Yeah
Darin Fisher 40:33 ... to type it into? Just give them one box." Um, now, if you look at Chrome, that's what its URL bar looks like, right? Just one box.
Ben Goodger 40:40 Mm-hmm.
Darin Fisher 40:40 And that's become the industry standard. But even on Chrome's new tab page, there's actually two boxes.
Ben Goodger 40:45 Mm.
Darin Fisher 40:45 There's one for the address bar at the top, and then there's this box in the middle that's the maybe comfortable, familiar Google box, right? What we wanted to do, and we kind of pushed ourselves with, is like this whole topic of like, hey, you might have a conversation you want to start with the model. You might be interested in, in navigation, navigational query, but really, you might not make up your mind about what your intent is until you start typing.
Ben Goodger 41:07 Mm-hmm.
Darin Fisher 41:08 And just one box is a lot simpler, and so when you open up Atlas, you just have one box on the new tab page, and that was something from a design perspective that we really, um, tried to achieve, and I think we were able to, and it keeps the whole system a little bit simpler for people.
Ben Goodger 41:22 Mm.
Darin Fisher 41:22 It might be a little, um, unusual and, and not what people are used to, but I think over time, they'll, they'll get to like it.
Favorite non-AI primitives: scrolling tabs, tab search, and performance architecture
Andrew Mayne 41:28 What was some of your favorite features, some of the things you're glad you were able to implement?
Darin Fisher 41:34 Oh, man. You know, um, it's interesting whenever you get a chance to build a new browser, uh, you, you have, have, having, having worked on quite a few, uh, you get a chance to sort of, uh, start over and reset on certain things. Um, not everything-
Ben Goodger 41:49 Mm
Darin Fisher 41:49 ... because I think one of the core tensions is that people are used to their browser the way it is, and, but you do have a chance to rethink some things. So one of the features I worked on was, uh, the scrolling tabs feature.
Andrew Mayne 42:00 Mm-hmm.
Darin Fisher 42:01 It kind of came from an insight of, um, [tsk] that, uh, you know, life, tab life could be a little better if maybe new tabs all started, all were inserted on the left-
Andrew Mayne 42:10 Mm-hmm
Darin Fisher 42:11 ... or just on one side. Um, you know, if you're a user who pins tabs in your browser, which is a pretty advanced use case-
Andrew Mayne 42:17 Mm
Darin Fisher 42:17 ... maybe a lot of people don't know that you can pin tabs to the tab strip in Chrome or Safari or other browsers, but it is a common thing, and if you pin a tab, it'll be pinned on the left side. And whenever- suppose that was like a Gmail tab. When you click links, those new web pages would open just adjacent to that pinned tab. But if you press the plus button, uh, the new tabs would appear off to the right.
Andrew Mayne 42:39 Mm-hmm.
Darin Fisher 42:39 And what ends up happening is, you're working throughout your day. You're going to Gmail, opening a, a, a URL from there. You're pu- hitting the plus button, opening buttons, uh, tabs on the right, and you're sort of accumulating old tabs in the middle.... And so, uh, it becomes a little bit painful to, uh, close all of the tabs to the right from the middle, you know, to, to-
Ben Goodger 43:00 Mm-hmm
Darin Fisher 43:00 ... to clean up those tabs because, and you just end up with a lot of clutter. So scrolling tabs was one of the innovations that we worked on to try to make tab management better, and it's like, it's not an AI feature.
Ben Goodger 43:10 Mm-hmm.
Darin Fisher 43:10 But it's like when you have this opportunity to rethink browsers, um, it's an opportunity to rethink some of these primitives and try some different things.
Ben Goodger 43:18 I mean, it's a major productivity tool, so finding these wins, uh, can be, like, really, really exciting.
Darin Fisher 43:23 Really.
Ben Goodger 43:23 Yeah.
Darin Fisher 43:24 And we think about, um... One of the things that I came to realize and appreciate only later is that if you have a browser that more naturally scales to having tons and tons of tabs, it means that certain kinds of things get unlocked for you. So everybody's, or m- many people are probably familiar with the ability to search for a specific tab that you might have open. There's a Command-Shift-A or-
Ben Goodger 43:44 Mm-hmm
Darin Fisher 43:44 ... a button for that in many browsers. Uh, what, with our, with, uh, with, uh, with our, with our system that a lo- with scrolling tabs, the fact that it can accu- allow for a lot of tabs to be accumulated without them all being in your face, you can still search across them and find these old tabs. So in a way, it's like this, this history of, uh, of things you've done in your browser is there for you to search in a very familiar way because it's your, it's Command-Shift-A, it's right there, and it-- you can have that capability without it being cluttered.
Ben Goodger 44:15 Darin's talking about tab search.
Darin Fisher 44:16 Tab search.
Ben Goodger 44:17 So you can search over your tabs, you can just type, and it will find the tab that you want. But I think the most interesting thing about this feature is the fact that you don't need to close tabs.
Darin Fisher 44:24 Right.
Ben Goodger 44:25 And so you can end up having... You know, my-- I think my browser, I've got, like, well over a thousand tabs open, and I just wouldn't think to, for that to be possible-
Darin Fisher 44:32 Hmm
Ben Goodger 44:32 ... in another browser.
Darin Fisher 44:33 Or you might think that that'd be a problem, right? But it's not.
Ben Goodger 44:34 No. Because the system manages the memory for you.
Darin Fisher 44:37 Yeah. Now, this is the scrolling tabs feature-
Ben Goodger 44:39 Yeah, mm-hmm
Darin Fisher 44:39 ... that is not on by default, and part of the reason why it's not on by default, as much as we think it's magical, and I'm a huge fan of this thing we built, it is also a little different than what people are used to.
Ben Goodger 44:48 Mm-hmm.
Darin Fisher 44:49 And we wanted people to, um, not have to learn so many new things all at once when they're approaching this browser that is, you know, bringing all these AI capabilities. But one of the amazing things when you have-- allow for thousands of tabs to be open, means not only do you get to access it again with tab search, but the model can see them.
Ben Goodger 45:08 Mm-hmm.
Darin Fisher 45:09 The model can see these tabs. It means your working set can be very large, larger than what you might keep in your head naturally, but you know that there was something there, so you're gonna ask the model for it, and it can go and interact with those tabs again.
Ben Goodger 45:21 I-
Darin Fisher 45:21 And I think that's actually pretty amazing.
Ben Goodger 45:23 I would be remiss if we didn't, um, if I didn't mention for this question as well, just the basic feature of Atlas, which is the Ask ChatGPT sidebar.
Darin Fisher 45:31 Mm-hmm.
Ben Goodger 45:31 This is something I get some value out of every single day as I use the browser. Um, I, I pull that thing open, I ask it to summarize a page if it's too long, or I wanna figure out, like, if I'm reading an article, like, how it really, um, matters to me in particular, or if I have a question about something that's going on in the world, it can go off and do some research for me and come back with stats and facts and figures. I've used it when I'm online shopping to make sure I'm looking at what really is the best deal on something that I'm looking at. I've used it to help, um, spin up agent tasks to go off and automate some of my productivity workflows. I've had it build, you know, Google Forms for me to help me, like, quiz my coworkers on the best way to design new features for the browser.
Darin Fisher 46:11 I really like that example-
Ben Goodger 46:12 Yeah
Darin Fisher 46:12 ... because if I remember correctly, you said you also asked, you asked Sidechat to help come up with the outline of the-
Ben Goodger 46:18 Right
Darin Fisher 46:18 ... of the survey, and then you said, "Hey, can you just put it into a Google Form for me?"
Ben Goodger 46:22 Mm-hmm.
Darin Fisher 46:22 And it did it.
Ben Goodger 46:22 Right.
Darin Fisher 46:22 And that's, like, really cool.
Ben Goodger 46:24 Yeah. So it's just, um, you know, we talked before about, um, bringing the power of ChatGPT with you everywhere you go on the web, and I think that sidebar really, it's, it's like having ChatGPT sitting on your shoulder, just right there to help, you know, give you some advice, uh, wherever you might need it.
Darin Fisher 46:38 Uh, and sometimes even just simple things, like, um, I was in Slack, and, uh, there was some, uh, some- somebody shared some text that was in another language, so I just selected it, and I right-click and, uh, asked Sidechat about it, and, and it translated it for me. And it was so much easier than having to-- You don't have to sit there and copy/paste, right?
Andrew Mayne 46:59 My, my favorite use of that so far with the, in the agent mode, and I, I won't name the cloud provider, but it's a very big company that often you find out you run a lot of services, and you forget what those services are, and at the end of the month, you get a bill, and it's a very confusing bill because you're trying to figure out, "I thought I shut this down."
Darin Fisher 47:16 Isn't this all of them, or...? [chuckles]
Andrew Mayne 47:17 Yeah, well, some of them are a little bit-- or maybe been around longer and trying to parse through that's like reading a Soviet tractor manual. And I went in, and I said, "Hey, I got this bill. I think I shut everything down. I don't know what's going on. Can you help me with it?" And I watched it navigate-
Darin Fisher 47:32 Yeah
Andrew Mayne 47:32 ... through the website, go to the page, find the different things I was doing, explain to me what the service was doing. I'm like: "Can I shut this down?" I'm like: "Yeah," shut it down, and that was, like, a hundred dollar a month bill that was just saved through-
Darin Fisher 47:43 Wow, that's really... That's awesome.
Ben Goodger 47:45 Another one I had, um, as well, I was, uh... Actually, I had some, some, uh, medical tests done recently.
Andrew Mayne 47:54 Mm-hmm.
Ben Goodger 47:54 And, uh, sometimes it can take a while for the doctor to come back and explain to you, like, what they mean.
Andrew Mayne 47:59 Sure.
Ben Goodger 47:59 And then in the me- in the meantime, you have the patient portal there, and you can access the-
Andrew Mayne 48:02 Mm
Ben Goodger 48:02 ... sort of doctor language stuff there, and I can't read that. It's not written in English, normal English. And you can have the thing-- You can ask, and that will tell you kind of like, what, what that means for you, and I found that to be really helpful at least.
Darin Fisher 48:15 Yeah.
Ben Goodger 48:17 So saving you money, helping you get some answers, this thing feels very, very... I, like, I'm pretty convinced that this is the way, increasingly the way that people will interact with information.
Darin Fisher 48:25 Or you're using some, um, very popular, yet complicated HR tool or something like this-
Andrew Mayne 48:31 Yeah [chuckles]
Darin Fisher 48:31 ... and you're like: "Where is that thing again?"
Andrew Mayne 48:33 Yeah.
Darin Fisher 48:34 And it, and it, it, of course, has studied the manual for you and can go and show you the way.
Andrew Mayne 48:38 I-
Darin Fisher 48:38 It's kind of remarkable.
Andrew Mayne 48:40 I feel like it, it took me a long time to realize that once I had, you know, an iPhone, that I always had a camera and a flashlight in my pocket. And there are many situations where I'm like: "Ah, what was the name of this thing I saw in the store?" It's like, you could have taken a photo, you know? Or, "Man, it's dark." It's like, you got a flashlight. And I feel with these tools, there's a lot of capability there that... We even saw that, too, with Search. You know, some of us were power users of Search, and other people, are-- it was a complete mystery.... And do you think we'll see a faster acceleration here, that people are gonna start sharing and understanding how to use this?
Ben Goodger 49:11 Yeah, I think the, the stage that we're in with Atlas right now is, um, we think this is a really powerful tool, but we don't know all the ways in which people will use it.
Andrew Mayne 49:19 Mm.
Ben Goodger 49:19 And, and it's kind of like the internet in that sense. Um, and so one of the reasons why we wanted to get this out, you know, when we did, is we just wanna see how people use it, and hear from people, you know, where it works well, where it sucks and needs to get improved. You know, um, but I think over the course of time, we'll get a better feel for that. I also think we'll need to help explain in more cases, like, when the right time to use it could be.
Andrew Mayne 49:41 Mm-hmm.
Ben Goodger 49:42 Um, because I think there's a part of building something that feels like magic, and sort of making that magic real for more people in more situations. And we don't wanna have to have-- rely on people to always think-
Andrew Mayne 49:52 Yeah
Ben Goodger 49:52 ... I should ask this question at this time.
Darin Fisher 49:54 Yeah, it's really easy to just-- We're creatures of habits. We use the browser the way we use the browser. We use our computers the way we use them. Uh, we don't always realize when there's a better way to do something or a more efficient thing that we could be doing. I, I feel this way about when, you know, the process of learning how to use ChatGPT in the first place. It's like- it's just like realization, "Oh, I should just ask the model for that. It'll save me time." But I-- it takes a little while, and there's a bit of a tipping point for people, where they start... At some point in their journey, they're gonna learn that how to use these tools. And there might be some people who are early adopters, and they can show the way and figure things out and share those ideas, but, but also, it's, it's kind of like I think a lot of people haven't yet found, uh, found their way to how to use these tools in the best way.
Andrew Mayne 50:39 I, I find myself still trying to shut down tabs because I'm still-- You know, I started using browsers in the era of getting the pop-up message, "You have many-- too many browser tabs open."
Darin Fisher 50:50 Yes.
Andrew Mayne 50:50 And now the, you know, compute, and the capability, and the management internally of these things is way advanced. And so I think about, like, y- you know, I'm not optimized in many ways to use this.
Darin Fisher 51:00 Yeah, so, uh, a- also another, like, um, say, non-AI feature of our browser was that we kinda took a, took a page out of the playbook of, of mobile browsers, uh, recognizing that, um, you know, your laptop's, uh, computing resources are not really limited.
Andrew Mayne 51:14 Mm.
Darin Fisher 51:14 You have a battery you care about. Um, you know, so w- we put a limit on how many, uh, tabs would actually be backed by a live web page.
Andrew Mayne 51:23 Mm-hmm.
Darin Fisher 51:24 Instead of trying to, um, might be the more traditional approach that desktop browsers would take, which is to just try to mitigate the cost of those background web pages that you haven't used in forever, uh, we will just close them down.
Andrew Mayne 51:37 Mm.
Darin Fisher 51:37 And if you go back to that tab, it'll get reopened. And we keep a reasonable limit there, and we, uh, apply, you know, somewhat of a clever-ish, uh, caching algorithm-
Andrew Mayne 51:48 Mm
Darin Fisher 51:48 ... to try to, um, be smart about keeping- making sure that the tabs you care about are kept in memory, uh, so, so as to sort of lessen the burden on your computer. Um, you might notice also with Atlas, that it restarts super fast. Uh, when you restart Atlas, um, because of the way it's structured, uh, the Owl process is separate from the Atlas process, Owl being our embedding of Chromium.
Andrew Mayne 52:11 Mm-hmm.
Darin Fisher 52:11 So the two can start up in parallel, and we can restart Atlas, uh, very quickly with all your tabs and the data associated with them. Um, but the web pages aren't loaded yet.
Andrew Mayne 52:20 Mm-hmm.
Darin Fisher 52:21 Um, and when you click on them, it'll bring those web pages back. Uh, but this way, the whole system can stay fast, lightweight, and, um, as we were able to build Atlas as a separate, um, application from this, from Owl, uh, Atlas is controlling Owl. Owl is projecting data into, you know, the, the rendering of web pages into Atlas, but Atlas itself can stay a relatively thin, uh, Sw- Swift application.
Why Chromium (and its lineage): compatibility, extensions, and where innovation sits
Andrew Mayne 52:44 Mm. Why Chromium?
Darin Fisher 52:47 That's a fantastic question. I [chuckles] I answered this question on, on, on the f- site formerly known as Twitter, um, by saying that, uh, you know, uh, web compatibility.
Andrew Mayne 53:01 Mm-hmm.
Darin Fisher 53:01 So it turns out, unfortunately, or for better or worse, a lot of websites are only really designed to work with Chromium. [lips smack] Uh, there are f- features of major websites, which I, I won't get into the naming names, but that, that are just not present if you're not using Chromium-based browser.
Andrew Mayne 53:17 Mm-hmm.
Darin Fisher 53:17 And the other reason is, um, Chromi- Chrom- Chromium extensions. Extensions built on top of Chromium are very popular, and when you build your browser on top of Chromium, it means those extensions will just work. And so we wanted to make sure that we were building a browser that, first off, works for people-
Andrew Mayne 53:34 Mm-hmm
Darin Fisher 53:34 ... that all the websites they care about will be supported, and all the features of those websites will be supported. And we want to make sure that they could, uh, install any of the extensions that they care about and that they're used to using in their browser.
Andrew Mayne 53:44 And it seems like there's also, you know, we, you know, kind of non-technical, when they hear Chrome, they hear Chromium, but not understand there's like a, a really deep lineage that even goes further back.
Ben Goodger 53:54 Oh, yeah.
Andrew Mayne 53:54 You also have WebKit, and KDE, and whatnot.
Ben Goodger 53:57 So what I'd say is, and, you know, I think there's, like, a lot of excitement for, you know, from among the community for, for-- to see, like, new rendering engines come about.
Andrew Mayne 54:05 Mm-hmm.
Ben Goodger 54:05 Uh, and that's certainly been part of the DNA of the web too, over, over the years. Um, at the same time, just like Darin said, um, you know, when you build a new browser and you don't have that many people using your product just yet, it's, it's-- you kind of just want the web to work-
Andrew Mayne 54:18 Mm
Ben Goodger 54:18 ... as people know it today. And actually, back when we were starting working on Chrome, we had the same concern. Like, the, you know, Chromium today, you know, has Blink, which is sort of its own rendering engine that sort of diverged from, from that lineage.
Andrew Mayne 54:31 Mm.
Ben Goodger 54:31 Um, but at that point, like, there wasn't much appetite for taking risks like that. Uh, and so that, uh, the Chromium rendering engine is based on WebKit-
Andrew Mayne 54:39 Mm
Ben Goodger 54:40 ... uh, which is the Safari rendering engine, which is open source, and that itself was based on an earlier rendering engine from, from the Linux, Linux world-
Andrew Mayne 54:47 Mm
Ben Goodger 54:47 ... uh, called, uh, KHTML. Uh, and so yeah, it's really interesting to go back in time, so you can see how these sort of open-source projects fork, and branch, and so on.
Darin Fisher 54:55 Yeah.
Ben Goodger 54:55 Mm.
Darin Fisher 54:55 There's code in Chromium that comes from the Mozilla project, too.
Andrew Mayne 54:58 Mm-hmm.
Ben Goodger 54:59 You're going back to the 1990s, uh, you can find this- [chuckles]
Darin Fisher 55:02 Yeah
Ben Goodger 55:02 ... uh.
Darin Fisher 55:03 This is true.
Ben Goodger 55:04 This lineage.
Darin Fisher 55:05 So, you know, browsers are this sort of, um, this layer cake of technology that's been built up over time. And, you know, really, where we are, uh, wanting to innovate is that next layer.... where the, the AI model comes in, and how it's articulating and interacting with the foundational layers. And so whereas building on top of Chromium is like, gives us this well-known foundation, uh, we built on top of Chromium in a very different way than normally browsers would. Most browsers are just taking Chromium and forking the UI, or, uh, layering another, uh, you know, UI on top of Chromium, but running right in the same process as Chromium.
Andrew Mayne 55:41 Mm.
Darin Fisher 55:41 Uh, what that means is that, uh, if Chromium's doing work, your application's not doing work. And so in them- the structure that we, we set up with Owl, uh, it means that, uh, Atlas is able to work in parallel with the rest of the, all the activities related to rendering the web and, and producing web pages, which is pretty cool.
Andrew Mayne 55:59 So if the browser part crashes-
Darin Fisher 56:02 Yeah, if the, if the Owl part crashes, if something goes wrong with that piece, um, then y- uh, Owl, uh, sorry, Atlas can restart Owl.
Ben Goodger 56:10 So that's actually a really interesting thing, 'cause I remember back when we were originally doing the design of Chrome, there was this thought of, like, well, the web page might crash, and therefore your browser should be around. And then because, you know, Chromium has sort of become this very sophisticated platform for web pages, it itself has become super complicated, and now, like, Atlas-
Darin Fisher 56:27 Yeah
Ben Goodger 56:27 ... is this very lightweight frame around the outside that really is about, like, that core productivity use case of using a browser, um, with ChatGPT, kind of as this tool that you can bring down on, on, onto any page, and that's really where its focus is. Whereas, like the Chromium, i.e. Owl piece, is able to focus on being that platform, and then both parts are not really-- They're, they're resilient to each other's, you know, difficulties- [chuckles]
Darin Fisher 56:55 Yeah. [chuckles]
Ben Goodger 56:56 ... that they might run into-
Darin Fisher 56:56 It's true
Ben Goodger 56:57 ... uh, in that sense. Um, yeah, just actually, like another fun fact about, like, OpenAI and the benefit of the system that, that we built, um, every engineer that starts at OpenAI merges code on their first day. And, uh, if you think about, you know, how massive Chromium is as a platform, you know, it's, it's really super powerful, but it, it's a lot of code. It takes a while to get all that code onto your device.
Darin Fisher 57:21 Just a little bit complicated.
Ben Goodger 57:22 Yeah.
Andrew Mayne 57:22 Mm-hmm.
Ben Goodger 57:22 And, and it takes a while to build it all. Uh, and so we tend to get our new hires in the afternoon as they've gone through all of their onboarding training, um, and, uh, then they have to merge a change. Uh, so to be able to check out all of that code and build it, and then make a change to the code in your first afternoon, can be pretty tough if you have to do all of that. But because we have structured this in a way that they don't have to, you can go and make a change to the, um, to the Atlas side, get that checked out, and built very, very quickly. Uh, our engineers are able to be productive right away, and, uh, merge code on their first day, and like, ship features on their first day.
Darin Fisher 57:55 Related to this, um, so always when you're starting a new project, you get to make new technology choices. You get to- When we were starting Chrome, we got to say, "What is the latest and greatest way to build code?" Right? Uh, fast-forward to starting Atlas, we're like, "What is the latest and greatest way to build a native app on macOS?" So of course, we're gonna be using Swift, we're gonna be using SwiftUI where, where it makes sense.
Ben Goodger 58:16 Mm-hmm.
Darin Fisher 58:17 And we're gonna be, you know, using all these, um... You know, the app is built in Xcode, and just done in a very familiar way. So people who are f- are used to doing Swift development, maybe 'cause they're building iOS apps, they can come in and just be instantly productive because this is not a foreign code base, not a foreign system, you know? And, uh, and yet, it's, it's harnessing the power of Chromium at the same time, which is super complicated.
Long-term bet and what’s next: platforms, mobile UX, and the agentic web in 5–10 years
Andrew Mayne 58:40 When you look at this from the outside, a lot of people draw comparisons. They go, "Well, you know, there was Operator, now they're doing ChatGPT Atlas. You know, is this gonna be a real thing for them, or is this just another experiment?"
Ben Goodger 58:54 So a browser, um, you know, I think it's a, it's a super, um, core tool for productivity, um, and it's something that you need to be able to count on.
Andrew Mayne 59:05 Mm-hmm.
Ben Goodger 59:06 And so my view for this has always been that this is like a long-term investment, and so that's the way that we're approaching it. Um, there's a set of functionality that we've launched, which is sort of like the first phase. If I come back to browser history, I sort of say it's the, you know, Netscape 1.0, if you like-
Andrew Mayne 59:20 Mm-hmm
Ben Goodger 59:20 ... of this new era of, of web browsing. Um, [lips smack] so there's gonna be a lot of imp- future improvements to come, features that we're building because people have told us about it from the initial set of feedback, things that we come up with, you know, through our partnership with research, a whole host of new functionality, uh, that will come out over the course of time. The other thing that we hear from folks is they wanna see this across different platforms.
Darin Fisher 59:43 Different platforms. [chuckles]
Ben Goodger 59:43 Yeah, basically, "This browser isn't available for my Windows device or on my phone," or that sort of thing. So these are things that we're, we're thinking about and, and, and working on. So there's, like a, a long roadmap of enhancements, and so we want people to, um, both feel confidence that this will get better over the course of time. Uh, actually, it gets better every week when we push an update. Um, and they will see it increasingly on more of the surfaces, uh, where they are, so definitely a, a long-term investment for us.
Darin Fisher 1:00:10 Yeah, we got this a- app to the point where, um, internally, uh, the users at OpenAI were, um... We were seeing them, uh, enjoy this product, and, you know, and we wanted to- We were at a point where, uh, the kind of feedback we were getting was, uh, was, was, was, uh, you know, uh-
Ben Goodger 1:00:30 Why haven't you shipped it yet? [chuckles]
Darin Fisher 1:00:31 Yeah, why haven't you shipped it yet? [chuckles] That's exactly what I was thinking, 'cause it's like, we weren't getting new feedback.
Ben Goodger 1:00:36 Yeah. Mm.
Darin Fisher 1:00:36 We, we were, we... And we realized we're ready to ship this. We wanna share this with the world. We wanna, we wanna hear how other people are experiencing it. Um, and I think that, you know, this feedback's been all kinds of amazing. Uh, you know, obviously, there's been, uh, you know, people who have their pet feature that's missing, or like when Ben said-
Ben Goodger 1:00:53 Reader mode. Reader mode
Darin Fisher 1:00:53 ... "How come I can't-"
Ben Goodger 1:00:54 The same. [chuckles]
Darin Fisher 1:00:54 ... You know, "Where's Wind- Windows When?"
Ben Goodger 1:00:56 Mm-hmm.
Darin Fisher 1:00:57 Uh, but, you know, the other day, my, my 14-year-old son came back, and he's like: "Oh, man, my friends at s- school, they love this browser," you know, kind of a thing. And, and I was asking, like, "Really? What do they love about it?" And he's... And then they- this- he just was, like, talking about all the different AI features that they were checking out. I think it's really interesting to see, um, that sort of spark of fascination from people, um, whether it's kids or, or, uh, you know, people we work with, or friends as, as they sort of share their experiences with this thing, and-... I don't know, I just love to see how people, um... My w- my wife was so giddy when she f- first got to try Atlas when, uh, when we did the friends and family testing.
Andrew Mayne 1:01:34 Wow!
Darin Fisher 1:01:34 And just to be able to go and explore, researching some task and asking, uh, um, the Side Chat about, uh, what she was looking at, and just she had so much fun with it.
Andrew Mayne 1:01:43 My wife loves it. This is not a lie, an exaggeration. The night that it came out, we're sitting in bed, I'm reading, and I look over, "How are you doing?" She says, I swear to you, I swear to you, she goes, "I can't stop thinking about ChatGPT Atlas." [laughing]
Darin Fisher 1:01:55 [chuckles]
Ben Goodger 1:01:55 [chuckles]
Andrew Mayne 1:01:55 Because for her, it was her first use of an agentic system like this, and to be able to go do these things was a huge unlock. Was her favorite tool is-
Darin Fisher 1:02:02 Yeah
Andrew Mayne 1:02:02 ... ChatGPT-
Darin Fisher 1:02:03 Yep
Andrew Mayne 1:02:03 -now connected to the browser.
Darin Fisher 1:02:05 Yep, exactly, because, um, just the tediousness that it takes away when you- you're-- I'm looking at this website, or I'm doing some research, and I can just, uh, ask it now-
Andrew Mayne 1:02:14 Mm
Darin Fisher 1:02:14 -about the thing I was talking about it, to it before, and now it can take this as context, and we can kinda keep going and research and whatever the topic is, it just becomes a lot easier when the AI, when the model is right there with you.
Ben Goodger 1:02:26 I think a lot of folks like struggle with, you know, how to do, you know, e- you know, sometimes, like, what seem like very complex tasks on certain websites. You know, one of those websites, I think is, you know, if I'm on a spreadsheet, on-
Darin Fisher 1:02:36 Mm-hmm
Ben Goodger 1:02:36 -a web-based... Pick your favorite web-based-
Darin Fisher 1:02:37 Mm
Ben Goodger 1:02:37 -spreadsheet program, how do I visualize this data in a certain way? If you just have a tool there that you can ask, you know, drop in and, and help in a very consistent way, I think that's really, really interesting. Um, yeah, the other piece that I think is interesting, you know, related to, uh, to your wife's reaction is, um, this is a lot of net new capability, and I think for a lot of folks in the world. Um, I'm-- what I'm really excited about with this tool is that, you know, our, our model capabilities are always evolving.
Darin Fisher 1:03:08 Yeah.
Ben Goodger 1:03:08 Uh, so at any given point in time, it's not like sort of the ultimate state of it, uh, but we kind of get to show people how this stuff works. Uh, and I think it-- with that, build some more understanding, some more trust, um, about, you know, how this, this, this technology is working on, on your behalf, and I think it's, you know, even if we came up with the world's best model, um, uh, tomorrow, that solved every problem perfectly in the way that you personally would have wanted that problem solved, you probably would still want to be able to come along for the ride-
Darin Fisher 1:03:38 Mm
Ben Goodger 1:03:38 -and understand how that was actually happening. Uh, just, you know, for, for your own education. I think over time, I think when as your trust level builds, eventually you'll be- you'll feel comfortable fully delegating very complex tasks, uh, to this technology. But we're not at that stage yet. Um, and so I think one of the things I'm really ha- really happy and excited about is that people are able to come and observe the sort of the next step of this technology and, you know, watch it, see how it does, you know, tell us, you know, yell at us when it doesn't work, you know, that kind of thing. Uh, but I think you can kinda get a feel for how it works, and as a result, like you will know, like what it, what it's gonna be capable of, and you'll kind of know what the controls are, like, where you want it to stop doing something, or you want to do something differently, you can just tell it, and, and so on.
Darin Fisher 1:04:19 I think you can kinda dial up and down how much you use the model-
Ben Goodger 1:04:22 Mm
Darin Fisher 1:04:22 -and the ways you use it in this product, but it's kinda like right there, and it's easy to try again. And I think sometimes the magic, the magic I've felt with ChatGPT is, is when it, when it really works ever so well, that I, I'm like: "Oh, I'm gonna go back for that," right?
Ben Goodger 1:04:35 Mm-hmm.
Darin Fisher 1:04:35 It's you have an- you have those magic moments where it's like: "Oh, my gosh, I, I, I'm gonna change my habits because of how this works," right? And I think, I think in this case, it's like easy for people to use familiar patterns, right? I open the new tab page, I do a search, it lands me into an experience that includes some links I can click on, but there's also the model response. And so you start to learn that, like, uh, maybe instead of uh, that website I was looking for that was gonna answer my question, it's just right there. And, and I can go and explore that, and I can ask it another question. You start to learn the power of, of this model just by virtue of using the product in a familiar, normal way. And so a- a for me, that's, that's kind of exciting to see how people-- it might open the door to people, um, [lip smack] starting to un- uh, realize just, uh, you know, w- what the capabilities of this model are.
Andrew Mayne 1:05:26 Looking to the future, and first, let's start, uh, short term. I can understand how a Windows version, you're gonna pretty much want parity of what you have right now in the Mac version. But when you talk about mobile, and it's one thing when I have a lot of desktop space, and I can put a sidebar and have the chat thing there, but when you're talking on uh, mobile browsing, as y- you guys know, especially, is a very tricky thing, and moving a search tab from the top to the bottom or whatever, seems like a revolutionary change. Uh-
Darin Fisher 1:05:55 Mm.
Andrew Mayne 1:05:55 How are you thinking about that, the mobile experience? And also, is it gonna maybe... Or are we gonna be thinking more agentically in how we use these things?
Darin Fisher 1:06:02 Well, so maybe fun fact, some of our initial explorations were actually on mobile, and, and part of the way we were thinking about it was really: what does it mean to bring the model to the web?
Andrew Mayne 1:06:12 Mm-hmm.
Darin Fisher 1:06:12 Right? And that can take many different forms. And of course, ChatGPT is- exists as an app on, on your phones. You can imagine ways in which you might, um, share, uh, to that, the kinds of memories that have been, uh, generated by using Atlas, right? So there's a lot of different flavors and forms in which mobile could come, or how this could manifest on mobile. But at the basic level, it's like, what are some interesting... You know, we're gonna be looking at, like, how can we bring the web to the model-
Ben Goodger 1:06:42 Mm
Darin Fisher 1:06:42 ... and what does that make- what makes sense on mobile?
Ben Goodger 1:06:44 The UX, um, you know, may look a little bit different. You know, obviously, as you point out, there's a sort of a different form factor there. We've got very talented designers, though I'm sure that they'll come up with a good way to solve some of this stuff. Um, d- definitely, I think on mobile, or at least I noticed through my own use, um, like my use case, is far more information retrieval. Like, I have a question, you know, actually, I use the ChatGPT app a lot-
Darin Fisher 1:07:07 Mm
Ben Goodger 1:07:07 ... so if I have a question about something, I'm around, or I point the camera at the thing, and like, "What is this?" You know, that type of thing. So I think that there are a whole host of, um, situations like that, where there is web content that is part of that journey, and we wanna make sure that, that, uh, that user flow that you have with Atlas, where you can view some web content, where you can ask follow-up questions, and then go back and look at the content again, that that feels very good. Um, that's something that we're sort of in the midst of figuring out-
Darin Fisher 1:07:34 Mm
Ben Goodger 1:07:34 ... right now. And so, you know, like, not too much more thought on it than that. Other than that, it's, um, something that we're gonna wanna make sure is, like, very, uh, feels very good.
Darin Fisher 1:07:43 ... Yeah, I think we hear from people that the, the, the importance of mobile, 'cause, uh, you know, they're doing work on their laptop, and they want to continue in some fashion on their phone, right? So you can start to imagine, uh, the kinds of, uh, space of possibilities there.
Andrew Mayne 1:07:58 Where are we gonna be in five years with how we're using the web and how we use tools like this?
Ben Goodger 1:08:02 So I would love to be in a place where people think less about the particulars of the tools they're using and are more just expressing what they want-
Andrew Mayne 1:08:14 Mm-hmm. Yeah.
Ben Goodger 1:08:14 - to the system, and then the system is smart enough to understand how to respond to that, uh, in a good way. And so in that sense, we can be as, you know, as humans, we can be focused on the highest-
Andrew Mayne 1:08:28 The questions [chuckles]
Ben Goodger 1:08:28 - you know, the questions, the highest order, um, piece, which is like, what is the most, um, interesting for me to do? You know, the model maybe can take over, an agent can drive maybe the less appetizing part of the work, the more, more-
Andrew Mayne 1:08:41 Mm
Ben Goodger 1:08:41 - I use the word toil to describe some of that sort of grunt work of pulling information from a bunch of different sources. Um, maybe it can do on- do a bunch of things that just seem very difficult to you because you, you've not done them before. It, it sort of knows how to do those things, and then you can be focused on the things that you want to do. Uh, so I imagine a world where actually there may be a lot of internet traffic in the future that is, um, that is agentic.
Andrew Mayne 1:09:04 Mm.
Ben Goodger 1:09:04 It may even be most of the internet traffic. I still see, um, people as doing a bunch of tra-- generating a bunch of traffic as well. Um, but, uh, that should be, you know, it should be efficient. It should be... People should be focused on doing the things that they want to do and delegating more of this work, uh, to, uh, more of this toil, really, uh, to, to agents that can take on a lot of that, that other stuff. And then, like, if you have to, uh, make a decision on a project, maybe your agent comes back to you and, and gives you some choices. "You know, if you're gonna take a vacation, do you want to stay at this hotel or that hotel? You can pick, uh, between them. You get to make a choice." Um, but all of the sort of grungy clicking around and scouring the internet, uh, for, for these things, maybe it, it, it took the first pass and presented you some choices.
Andrew Mayne 1:09:49 We, we talked a bit about how, as users and developers of this tool, you look at it. If I was somebody who had an e-commerce site, and if I was looking at I'm gonna be putting information on the web, and I know that, one, there'd been conversations, you know, for a while, like, what happens when LLMs go search the web? And now we get into the world of agents, when agents are using this and plus, you know, LLM-powered browsers. What advice, what direction would you be telling people to think?
Darin Fisher 1:10:15 You mean, uh, the publishers?
Andrew Mayne 1:10:16 Yeah.
Darin Fisher 1:10:17 Yeah, I mean, I think that, um, it's really interesting, like, in some ways, I think about the, the maybe a little bit related to what Ben was talking about. Like, um, you know, you see the more recent models, they've, they've learned how to, uh, based on the query, decide, "Hey, I should actually look at the internet to, um, to, to answer your question," right?
Andrew Mayne 1:10:37 Mm.
Darin Fisher 1:10:37 And so I think it's really interesting how these models can, um, help you connect to publishers or in, in web apps and whatever content that's out there, either, um, you know, giving you, uh, you know, a snippet and, like, a citation to it, so you can go deeper or, or even just connecting you to it because that's actually what you... You know, if your intent was to navigate to a site-
Andrew Mayne 1:10:59 Mm
Darin Fisher 1:10:59 ... then it can help you get there. I think one of the things we've been exploring with Atlas is how to, for example, better handle and better serve navigational intents.
Andrew Mayne 1:11:07 Mm-hmm.
Darin Fisher 1:11:07 Sometimes people come to their browser with absolutely the intent of, "I want to buy this product on this site," you know, and that's our job just to get you there-
Andrew Mayne 1:11:16 Mm
Darin Fisher 1:11:16 ... as fast as possible. And so it's been actually, an el- an element of building Atlas, was making sure we're serving those kinds of queries well. Uh, other times, you just want to probe the knowl- the, the knowledge of the model and have it go and research something for you, and sometimes that involves it needing to invoke tools to do that on your behalf. And so again, sort of to Ben's- what Ben was talking about, it's I... You know, I imagine a world in the future when you don't have to be so prescriptive about what tool you want the model to use, but rather it s- has this incredible palette of tools that it could draw upon, and some of them can be, you know, actuating your browser, things like this.
Andrew Mayne 1:11:52 We're gonna be using web pages in ten years?
Darin Fisher 1:11:54 I think so. I- you know, it's kind of a, um, it's kind of this, this fabric of like, of this world where people are publishing through this, this, this, this- it's the core primitive-
Andrew Mayne 1:12:06 Mm
Darin Fisher 1:12:06 ... of how people are putting content out there, right? So it's kind of the, you know, the, the, the internet superhighway, all that kind of, uh, those analogies, but it is like this, this, uh, open fabric for on which people can publish, and I don't see that changing.
Ben Goodger 1:12:19 The world's largest, you know, as you mentioned before, most open platform, and I think some of that power of openness is always gonna make it attractive-
Darin Fisher 1:12:27 Mm.
Ben Goodger 1:12:27 - for people-
Darin Fisher 1:12:28 Yeah
Ben Goodger 1:12:28 ... to put content on.
Darin Fisher 1:12:29 The way I look at it is these tools, yes, they are able to understand that fabric, understand that internet, understand the content that's out there, but they're also able to bring that content to people and connect people to that content.
Andrew Mayne 1:12:41 Mm.
Darin Fisher 1:12:41 And that can be very powerful. And again, it's all in service of what is that user's intent. I think it's fairly interesting to think about how we can do a better and better job of that, and really serving the needs of those users. And ultimately, as people are putting content out there that's intended for people, if you're, you're, you're putting out content that's your, you know, it's a gallery for somebody to go shopping or something like this, uh, we wanna help people find that, help people get connected to that, help people with the journey that they're on, and whatever that may be.
Andrew Mayne 1:13:08 This is exciting. Uh, thank you for sharing this. Um, any last suggestions, any power user tips?
Ben Goodger 1:13:15 Yeah, definitely, definitely, the scrolling tabs feature is, like, a favorite, uh, for both of us. Um, yeah, I would just say, like, uh, challenge yourself. Like I said, at this point, very early stage, but challenge yourself with your curiosity on any given page. Ask a question on more pages, uh, that, that you visit, uh, and you might be surprised with what you come up with.
Andrew Mayne 1:13:35 Awesome.
Darin Fisher 1:13:36 And we'd love to hear from everybody about, you know, how they're, how they're experiencing the product, so please keep the feedback coming.
Ben Goodger 1:13:41 Definitely.
Andrew Mayne 1:13:42 Ben, Darin, thank you very much.
Darin Fisher 1:13:43 Thank you.
Ben Goodger 1:13:43 Thank you.
Episode duration: 1:14:21
In the uListen app, the transcript lights up word by word as you listen. Get uListen
Transcript of episode WdbgNC80PMw