OpenAI's Adam Fry on ChatGPT Atlas: the browser as the missing OS layer for AI

phenomenal response. I don't know if you saw how many folks saw the YouTube video. I think it has over a million views now. Um but we will talk to Adam Fry about it. Welcome to the stream, Adam. How are you doing? Welcome. Doing very well. Nice to meet you. Busy busy couple days for you.

Yeah, it's been a bit of a whirlwind. I think just like you said, there's been such a big outpour of a reaction. And we're so excited to see the reaction and so excited that people get to use JPT Atlas. Now, this is not GPU constrained at all, right?

Like I mean obviously it's like a it it interfaces with your core pool of GPUs, but it's not like a new like an entirely new provision of GPUs, right? Yeah, that's right. So, you know, it's it's available to everyone. So, Chacht Atlas, you can go to the download link, you can download it.

It's available free plus pro everyone. Sure. You know, the agent part of it, which I know you guys are savvy about and we've talked to our colleagues before, you know, that is more compute intensive.

And so if you use some of the more advanced features of chatpt atlas where you ask chatp to go do things that can be more GPU uh intensive for our paid users for sure. Uh is there an incentive to set people I saw a screenshot but you never know what's real anymore.

Uh that that if you set atlas as your default you can increase your chatbt subscription limits. Is there some sort of incentive? We were just talking to Drew from uh Dropbox about, you know, productled growth, how you actually, you know, increase customer adoption.

Is there something going on there that you can tell us about? For sure. So, you know, chatbt Atlas is, you know, it's a browser and that's sort of at the core of what you do every day. It's sort of your operating system for your life, uh, whether it's personal or at work.

And so, you know, part of that is we wanted to really, we felt like the more people use it, you sort of discover these AI features. I think you guys have talked about this a lot in your show, but you know this new generation of browsers, there's a lot to discover.

You have to learn how to use chatbt in your normal workflow. And so we've really wanted to incentivize to people to really give it a shot. And the best part of these new browsers is using chatbt everywhere you go across the web.

And so one of the things that we did to start this off was if you give it a shot for a week as a default browser, you can sort of extend your limits and really allow you to test out all those AI features. That's cool.

Uh what what was uh was it always obvious that Chat GBT would would build a standalone browser or was there somewhat of a debate internally? I always felt like Chat GBT was a web browser, right? It felt like it was browsing the web on my behalf and serving it up in an interface.

And so it feels like it feels like this is the second browser that you guys are doing, but you're going such a good point. You know, there's that great stat of, you know, if you ask people off the street, you know, what browser do you use?

Most people actually don't even know what browser that they use because the concept doesn't always make sense. Y and so, you know, it wasn't obvious, but what it really started from was a very specific user insight.

You know, we were seeing, you know, you guys probably felt this when you use chatt, but you're on chatb chatbt. com. You're in a million tabs. You're doing a whole bunch of other stuff. You're in your docs.

You're you're planning for your next show, and you're like copying and pasting back and forth between chatbt and all the things you're doing. Yeah.

And so we just sort of sat back and said, you know, if CH chatbt is going to be more and more helpful for you over time, and people are going to rely on it more and more for the work that they do, for their personal life, it has to be able to coexist where you're doing that work.

And that that's what the browser is, is you know, where you're doing your work. And so we said, how can we actually bring CHBT into that when you invite it in to actually have the context of what you're working on?

And I think that that spark of an insight was really once you believe that we were like oh yeah building Chachi Atlas is is of course we've got to do that. What uh how how are you guys thinking about what success looks like with the product over the next few months?

I think you have an immense amount of pressure obviously that the chat GBT retention data was flowing around the internet over the last over last and it's kind of hard to hard to fast follow that because if you set the bar you know here but how how how are you thinking about it and how um uh and and then I kind of have a follow-up question around like you have hundreds of millions of happy active users that you can be porting over to Atlas over time but um maybe first like yeah what does success look like in the first uh weeks and months?

Yeah, it's a great question. So the the start is we think of it as a multi-year journey, right?

People have been using browsers forever and you're probably your existing one you've been using for 10 20 years and so it's you know building a browser is sort of like we joke it's like uh you're moving around the furniture in someone's home.

You know you know exactly where this is, you know exactly where that is and so change is going to take time. So we measure success over the long run. We want to be as helpful to our users as possible. But in the short term, you know, we're focused on retention.

You know, we're focused on for the people who are giving it a shot, which is a lot of people because people are really excited about this. You know, are they sticking with it? And, you know, are they loving it?

Are they sort of diving in deeper to using chat throughout, you know, as they browse the web to get advice on whatever they're working on as they go? And if we see those signals, which we're starting to see, that's when we sort of, you know, pour on the fire and and and sort of grow it and grow it.

But so short-term retention, but we also understand this is this is a long-term game for browsers and you know we're investing heavily for a long time in it. Are you thinking about a mobile version? We are bringing experiences Atlas to mobile and to Windows and so the team is you know furiously working on it.

We want to bring it to as many people as we possibly can. That's why we brought it to all of our free users, all of our paid, all of our pro right away globally. Yeah. Um we think we've got a great browser. We're really excited for people to keep using it. Very fun.

Yeah, it's going to be uh big shoes to fill since Sora rocketed straight to the top of the app store when you go when you go live on there. I'm sure it'll be I think I think add Adam Adam can take it to the top again. I love it. I'm a believer. Yeah, I I I I love the framing of it being like a longer a longer journey.

Like it just does seem like uh there's not going to be like the Sam Alman stealing GPUs moment for the browser where it just goes viral and everyone's like I got to check this out, right? it's very hard to like accelerate something that's a productivity tool. And so having that mindset um makes a ton of sense.

I'm wondering about new new patterns, new language, new archetypes of features that uh we might be talking about in a few years after this all rolls out. if you see any glimpses of this.

The thing that I'm thinking about is I was talking uh to someone on our team about Chrome extensions and I was thinking about what the extension looks like in the future and if there's some sort of like I've I've prompted an agent to do a certain workflow so many times that then it kind of collapses it into some UI and it places it somewhere in the browser and it's like it's almost writing like the custom software like there's you see I'm kind of like painting a bunch of splatter on the canvas But there's something there.

How do you think about like these new patterns? How much will your thumb be on the scale or do you want the community to customize things? Like where does the line draw and how are you thinking about that? Yeah. So there's probably three three big things when I look forward a couple years that I'm really excited about.

So the first is, you know, part of this Atlas launch was the agent in your browser where where it can take action. It brings up a cursor. It does things for you. I'm really excited for that as it gets better.

So, it's still early research preview, but it's starting to do tasks where we've seen it actually be helpful in a short amount of time. And I'm really excited for proactivity around that. Eventually, you can actually suggest, hey, I've already drafted five emails for you. What do you think?

You don't even have to task it. And so, there's this sort of proactive world of the agents that I think is going to be really really interesting and valuable for folks. The second is working with developers.

I think you know as it's more computerto computer interaction if you have an agent that you're delegating to to go you know write a document or go to a website those websites will evolve to work better with a software that is actually going and clicking around and we just are sort of scratching the surface we're teaching these computers to click around the way that we have but I think eventually the internet and website will evolve to co-work with agents to make things more productive for everyone and then the third is you know I think eventually models will be really good to create their own sort of applications within the browser.

And so I think there's so many amazing places that this is going to go and this is just sort of the beginning of this journey. And I think to Jord's point earlier, chatbt is already so retentive.

If you can bring chatbt into the browser like we have with Atlas today, it's already going to be so helpful to your life and then it's just going to get better from there. Yeah, you guys need to figure out you share any numbers with us. Well, you got to we got to I was going to say like the number of YouTube videos.

I think I think the uh you know the the most recent comp that we have to like a scaled consumer tech company launching an ambitious product in a super competitive market was was Meta launching Threads. Yeah.

And when there's this early excitement around a product, you can see, you know, massive amount of like downloads and then people were ready to call threads, you know, basically say like, okay, it's over. Like good, nice, good effort basically.

And then Meta figured out kind of the right um flow to get people to, you know, cultivate like a unique community there, get people moving over there consistently. And so I I would expect to see something similar out of Atlas.

And that I think a lot of downloads and excitement and usage early some people will churn but you have that massive base of users that you can continue to you know show them like if they do the right prompt in chat GBT it's like you know I can do this on your behalf over here if you like if if if you'd like.

So excited to see how this plays out. Yeah and I think the key thing there is focus too and investment and you know I get to speak on behalf of the team that's working on this but they're over there working on it right now just making it better and better.

So if you stick with it and you look at the small signals, the small paper cuts that people are telling you about. We just keep filling them, keep getting better, it's that's when your retention goes up. That's when people really enjoy your product. So you know, the launch is one thing. We couldn't be more excited.

The reception to your question has been immense. Lots of downloads, but uh not sharing any numbers today. Okay. But uh you know, it's really about do people stick around and and we have to keep working on that to earn that from people. Yeah, totally. Makes no sense. Well, we'll hit the gong once you're charting. Okay.

Okay. I I was ready to go to YouTube. Thank you so much. I mean, thanks for breaking it down and congratulations on on the launch. Yeah, this is great. Thank you guys for having me on. It's it's such a blast. Appreciate it. Awesome. Always a great time. We'll talk to you soon. Cheers.

Um, our next guest is Ian Rogers from Ledger. He believes he's already in the reream waiting room. Ryan in the chat. I have to address it. Well, Taylor says he laughs and says it's a good chromium rapper, sir. I think uh I think big things have small beginnings. Big Chromium rapper company of California.

Um, no, I think I I I think that they have the incredible advantage of being able to have, you know, a massive user base and get people over there, continue to iterate, and um, it just, you know what, you know what else is crazy?

I mean, the VS Code story like like the the browser wars look a lot a lot less crazy when you think about what happened to VS Code. Open source code editor, you get Windurf, you get uh,