Limitless Podcast - AI Browsers Will Completely Change How You Use The Internet

Episode Date: July 30, 2025

In this episode, we explore the rise of AI browsers and their potential to transform web navigation. Reflecting on the static nature of traditional browsers, we question if AI represents a genuine evolution or just a cosmetic upgrade. We share our experiences with Perplexity's Comet and Microsoft’s Edge and debate the future of intuitive AI interfaces. Don't miss our upcoming interview with Perplexity's CEO as we consider whether AI browsers are revolutionary or simply a nostalgic echo!------💫 LIMITLESS | SUBSCRIBE & FOLLOWhttps://limitless.bankless.com/https://x.com/LimitlessFT------TIMESTAMPS0:00 Intro1:23 The Evolution of Browsers2:34 The Rise of AI Browsers3:27 Players in the AI Browser Market5:31 Personal Experiences with AI Browsers11:09 Examples and Use Cases14:50 The Future of AI Browsers17:45 Envisioning the Next Generation of Browsers26:51 Next Up on Limitless------RESOURCESJosh: https://x.com/Josh_KaleEjaaz:https://x.com/cryptopunk7213------Not financial or tax advice. See our investment disclosures here:https://www.bankless.com/disclosures⁠

So since the year 1990, that's 35 years ago, pretty much everyone who's listening to this, both of us, we've been using an application exclusively for navigating the web, and that's the browser. Right? We've all used a web browser before, whether it be Chrome or Safari or the millions of other options that we've used. They've all been pretty much the same. If you look at a browser from the 1990s versus browser for today, it's really just a URL bar and a screen. And the screen displays the information. And that's how we've interacted with the internet forever. But just recently, thanks to the invents of AI and all these companies who are building really interesting applications on top of it, we now have a new type of browser, an AI browser, an agentic browser that will do a lot of the browser tasks for you. And that's going to be the topic of today's episode is AI browsers. It's like the browser you know and love, but it is super smart, super enhanced, and we'll do a lot of the tedious work that you normally do all for you. Ejazz, you've been going deep in this world of AI browsers.

Can you lay out the space, how it's working, who's involved, what it actually means to be an AI browser? I was just thinking about your comment on, we've been using browsers for about 35 years. Do you remember the Internet Explorer days, Josh, where we had to double-click this golden or blue e-icon and then kind of wait 10 seconds for it to load? Then we had to manually type in a web address,

and we've just come like incredibly far since then. And yeah, to answer your question directly, I think that this is a natural progression of where AI needs to be in everyone's lives. So to kind of like zoom out for a second, we had this massive explosion when GPT2 or GPT3 came to life from OpenAI. And it was this magical moment where we had this kind of like chat messenger and we could ask the smartest person in the room that, being GPT, any question, and it can give us an answer. It teaches us things. It can sometimes do things for

us and connect us in different ways with information. Kind of like how Google did when they created their search engine. It became the doorstep to the internet. It's the home page. It's the thing that everyone opens when they want to search something, interact and socialize, or even buy something, right? So it makes sense that AI is now coming to where most of the context and personal data around someone is. And I think if I were to zone in on why I'm so excited about this, Josh, is it's exactly that. I think it's the most publicly accessible data set for who you are, right, Josh?

Like, if I go on your browser right now, maybe you can share your screen or whatever that might be, I'm going to see a bunch of different tabs. I'm going to see your Spotify playlist that you're bumping tunes to. I'm going to see probably your Amazon list that you're kind of like shopping or shipping items to your apartment. and maybe a bunch of other things. It tells me who Josh is as a person. More so, it tells me how he's progressed

over the history of time, over that day, over that week, over the last 90 days with your browser history. So I think embedding AI into the browser experience is that next step up to getting a more personal AI face. Do you agree with that, Josh? Or what are your thoughts? Yeah, I think that sounds right.

I am actually a big agent AI browser hater. I think we're going to get into that. I agree that it's very cool, but I'm not sure the actual use cases apply. Before we do go there, I want to lay out the landscape of who's actually making these browsers, who is in this space that we're talking about because there's a notable company missing. So we have perplexity, which has the comet browser. We have Microsoft, which has Edge. We have OpenAI who is probably going to launch one.

They've signaled the intention to launch one. And then we have the DIA browser by the browser company. And notably, the one browser we all use, we're all recording this on right now, Chrome, is missing. which is interesting because one of the things that, I mean, the reason why Google became so large, one of the big reasons was the search engine was Google Chrome and noticeably they're missing. And when you look at all the search history, the revenue from search that Google makes, it's been down only. And you would imagine from a search company who's built this browser that is the number one

company in the world, they'd probably be aiming to build an AI browser, but that's not the case. And I think that's probably in line with what we're going to talk about later of why we don't like it. But I'm curious if you have any ideas why Google is noticeably missing from this group of AI browsers. You know what's hilarious? As you were making that point, Josh, I was trying to come out with a counter example, which was going on Google's search engine. And you know, whenever you search something now, you typically see an AI summary. I was going to try and argue that maybe that was an example of them embedding some form of AI in their browsing experience, even though it's not explicitly Chrome itself, but it's not available. I literally can't have access to it. So it seems

like they've regressed. To answer your question more directly, I think they're likely going to release it. If I had to hazard a guess, and the reason why I say that is they've been so committed to kind of being on the frontier at the AI model level. So Gemini 2.5 Flash is a frontier model that beats OpenAI's chat GPT and even claude. when it comes to coding and stuff. Google has been so focused on AI agents as well, and they've also been focused on the scientific side of AI as well, which is a realm that none of the other frontier AI labs have even touched.

So it kind of seems weird that even though they control 66.5% of the browser market share, it'll seem weird if they don't come out with a product, is what I'm saying. I don't know. I just feel like it's a waiting game. It might come out like next week or something. something like that. Totally. Uh-huh. We'll see. Well, we both have tried an AI browser. We both had access to perplexity. They gave us access to the comment browser. And we had a chance to use it for the last week or two. And I would love to hear your impressions because I used it, I tried it,

I have some takes. How have you been using it? How have you been liking it? How have you been disliking it? Tell me your personal experience. I have a lot to say about this. And I'm keen to actually hear what you say. So you described yourself as an AI browser hater. I wouldn't call myself a hater, but I also wouldn't say this is convincing enough to me to ditch Chrome and to use an AI browser. So as you mentioned, we're using this new AI browser from Poplexity. Poplexity, for those of you don't know, is like a supercharged AI search engine. And they released this new product called Poplexity Comet, which is their browser. And on this browser, it looks very similar to your Chrome browser or Internet Explorer browser. You can have

many different tabs, you have a search history, you have a toolbar with different settings and stuff, but with one unique feature, which is, or rather several unique features, which is AI is embedded everywhere you go. That means if you open up a YouTube page, you have a prompt from your little assistant in the bottom right hand corner, which will say, hey, do you want me to summarize this video for you so that you don't have to spend 30 minutes watching this? Or you can go into a shopping website and say, hey, I can actually sort for you the best deals for what you're looking for, because I read your email because you connected me to your Gmail account. And so I kind of know that your mom shared a list of items for your sister's birthday.

And I know that you're on Amazon to check what kind of fleece your sister might want or whatever. That's a terrible example. I would never buy my sleeves off of Amazon. But you get the general idea. My high level takes are as follows. It's great to have an assistant that can summarize long articles and posts for me. And the reason why I say that is I spend a lot of time reading and researching, Josh. We actually do a bunch of that for this show as well, right?

And sometimes, actually most of the time, it takes a lot of intense focus and many hours of my day, right? I would say maybe 70% of my working day spent like reading a ton of stuff and trying new things out. This does reduce that now to about 20 to 25%, which is a huge reduction. Yeah, yeah, I'm spending much less time doing this. You and I have been using this browser for about, what, a week now, like as our default browser. So it's been super useful. The second thing is, and this is a smaller thing, but I think it's important, I can finally stop copying and pasting links and paragraphs of text and putting it into whatever AI

model provider that I'm using, right? So typically the format is I would copy tweet links or article links and put it in Grog4 or put it in chat GPT, and then say like, hey, like reference these links, summarizes for me, give me some other types of insets that I might have missed. And it's just, it seems so small and minor, but not being able to do that is way more convenient for me. And I just kind of like, I love kind of leaning into that. The third thing that stood out for me, Josh, is search is just such a better experience now that I have this AI that's always readily available. It's kind of like Google search, but it understands the context behind what I'm asking it or why I'm asking it.

And that's really been an unlock for me, right? So an example might be I've been researching a specific technique of how to train AI models. This is something I was doing this week, right? I was trying to figure out how China's open source models was so much better than America's, even though they had like less resources and all that kind of stuff. I was trying to figure this out. and I was kind of like writing notes in my Google Doc, which perplexity had access to. I was kind of like writing up some thoughts. And then I opened up a separate tab to search for a specific term.

And it already had it preloaded for me. So again, it's like basic little things like this, but I like that it makes me feel comfortable and heard. But that's my take. Well, what about you, Josh? Okay. I've, yeah, I used it. No, I signed up for the, we got an invite code.

I signed up. It was the most gorgeous onboarding experience. I have ever had. It was designed incredibly. You type in your code. It unlocks this really fancy screed. It has nice music playing in the background. You choose your profile as it was gorgeous. And then I opened it up and it's the same as pretty much every other browser. In terms of how I used it, I actually used this example that we have on screen right now, which one of the interesting things about the AI browser versus a traditional LLM that we would use like chat GPT is it doesn't have all the integrations that a browser has. So you can connect at this AI

browser, in this case, comment, to your Gmail and actually have an interact with your email for you on your behalf. So this example here, and what I tried is I get a decent amount of emails that I'm subscribed to from something I bought one time or from a community that I used to be part of that I don't really care about. And you can just kind of ask it to pull the list of everything you're subscribed to and then choose the ones that you want to unsubscribe to. And it will do all of the work of doing that for you. So immediately after a few minutes of trying this and a few prompts, I was able to clear out my inbox from maybe 20 of these garbage emails a day to maybe just like three or four of the ones I actually want to receive. And that was really cool

to me. So in that case, it's really interesting because it has the access to integrations that other LLMs don't necessarily have, like chat GBT, like GROC. They exist in the silo, whereas this is kind of embedded across the entire browser experience. So in that case, was good. I really enjoyed the browser. Outside of that, it felt very much just like a normal browser. And it was mostly annoying that I just had to re-log into all my accounts again to do the same thing. So I went to YouTube and I uploaded a video for this channel. And I went to our RSS platform and I upload. And I was doing the normal things I do when what I realized is I really don't use the browser for all that many interesting things that require AI. When I want to use

AI, I go to a tool optimist for AI to kind of abstract away the complexities of having to use a browser. So when I want to learn about something, primarily I use GROC for that because GROC has the live real-time information from X, which is normally where I consume most of my information anyway. And I just have it pull all of the aggregated data that I want for me versus actually going and trying to find it. And as I'm going through these use cases of how I use a browser, there's less and less that are actually interesting. I think when people use browsers since the beginning of time, they've used them for two things,

right? It's like for productivity and for leisure. And when you're doing work, you just kind of interact with the specific apps. in the browser. So you're using an Excel spreadsheet or you're doing whatever, like uploading a video. And then in Leisure, you're kind of shopping for something. You're looking through clothes. You're watching a YouTube video. You're consuming media. And on the leisure end, there's no way that AI can supplement that. And then on the productivity end, well, you kind of want AI to exclusively supplement that. And you don't want to have to deal with all these annoying interfaces or

scraping data or getting that information. So it puts this browser in kind of a weird spot where I was using it for a few days. I was excited about using it for a few days. And then I just kind of didn't really have a need to use it anymore. I do a lot of the thinking and the analyzing in GROC. I do a lot of the productivity work in chat, GAPT. And then I just watch YouTube videos on YouTube using the Chrome browser that's worked for the last 30 years. So that's kind of where it's been weird with the AI agents as it seems cool. But I'm not sure there's there's any sustainable use cases that are very exciting. Do you agree, disagree? Yeah. I mean, last week we spoke about Open AI's new agent, right? And one of the main takeaways from that episode was, this agent's pretty cool, but it's not really adding much value to my life right now. Sure, I can, like, jump into my Gmail account and read a bunch of my docs. But, okay, and then what? Yeah, sure, you can, like, research a bunch of clothes that I might want to buy, but then what? Like, it, doesn't buy anything for me. And I feel like this is the same type of case here, right, where you've added a solution to a place where maybe you don't actually quite need it, right? But I want to spend a bit of time arguing what this could be versus what it is right now, because I agree largely with you that it is not as useful as I'd want it to be. And I really think you make a good point around the leisure side of things. But I want to kind of like think

about what this might end up becoming if we imagine that everyone's going to have some form of AI agent or companion in the future, right? So number one, I think the personal AI assistant is going to be a hugely valuable market going forward.

I don't quite know how it's going to manifest. My idea is it's probably going to be on the work side of things. So there's going to be an enterprise heavy presence of AI agents. And we're already seeing that already where they kind of like plug into your workspace, your Slack, and they're allowed

to do a bunch of things. Right now they kind of suck, but eventually I'm guessing it's going to get better and it's going to get more intuitive. The second thing is there's this trend, Josh, where I think we're stepping away from a lot of the actions that we've grown so used to over the last decade, which is scrolling, searching for different apps, downloading apps, the screen and typing letters. I think this progression is going to become something more automable where we're going to be speaking into a microphone and we just kind of like look at things or AI guesses

what we already want before we even get there, right? There's going to be this like autonomous kind of flow of things. I don't think we're there. I think this is a stepping stone. I do not think browsers right now are like a 10x improvement over what we had before. So I'm with you. I'm just going to use Google Chrome for now, but it's a good show and goal. Yeah, to me, if I'm talking to Perplexity CEO, which we are talking to in a few weeks,

I think the idea is why on earth are you spending so much resources in this intermediary step? Because it feels like the end state of this is fully agentic, fully obfuscating away all of the complexes of the browser. You want something, the agent goes out and gets it to you and delivers it to you. And this is kind of this middle ground that isn't quite there. and it's not fully leaning into agent. If we had to compare the agent episode that we had last week to this week, it feels like open AI directionally is actually in a much better position.

Because that agent, you can see an end game there where it kind of stinks now, but it's the right form factor and it has the ability, it has the openness to do that, whereas the browser is very constrained. I mean, their entire industry is dedicated to unifying user interfaces across the internet just so it's easy for humans to do that. And there's so much unnecessary complexity when it comes to this, 35-year-old interface that was built for a world that is totally different than today. So the constraints that existed in the 90s that we're still using today, they no longer exist.

And it feels a little lazy to just keep the same interface and throw an AI agent on top of it. When you could really redesign this thing from the ground up, you can really truly have an agent-first world that doesn't require you to interface with any sort of browser like this at all. So I think that's the place that I'm excited to go. to you, Josh. Yeah, it's probably either the earbuds in my ear or the little hardware piece that I have and it knows everything about me. It spends time with me. It understands what I like to do. It has my whole preference stack. And when I want something or I want to know something, it has all the context to solve the problem for me and it can go and do it. And then in the case there's any clarifying questions, we can clarify, but it's really just a personal

assistant that is fully capable of engaging with the world the same way a human would be. And that to me feels like the final form of this agentic. So I'm hearing a few things. Number one, you think this is going to be a different form factor completely. So maybe not looking at a browser or maybe not even looking at a screen at all, but maybe it might be some kind of earbuds, as you said, we're kind of like opined on what the new hardware device that Open AI is rumored to be building with Johnny Ive is going to be, maybe it's something that sits on your desk and listens and sees everything that you can

see. But that's one thing, a different form factor. Number two, it sounds like you're describing a very multimodal world, which I agree with, right? So it's not just something that can understand text, but it's something that can see that is very visual. It's something that can hear the same things that you hear and that understands the context of conversations that you're having with other people or why you do different things in the world. So it's kind of expanding from your digital presence to also be aware of your physical presence, right? And then the third thing that I'm hearing from you, Josh, is we're just kind of at that midway point where we kind of have something

or we're throwing a bunch of AI slop at the wall and hoping something sticks. But it's not cohesive right now. We just have this like cool AI chatbot and we've slapped it with a bunch of tool access, but none of the cogs work together. None of them match. So we just kind of have this really clunky kind of machine. am I kind of like painting the picture accurately? Yeah, there's two approaches, right?

There's one approach that is the open-ended one that open AI is taking. And then there's the constrained approach that perplexity is taking, where we're just going to take this browser experience and make it a little bit better. And the other side is, hey, we're just going to throw away the entire browser experience because that no longer matters, and we're going to redesign it from the ground up using this new intelligence. And I think that's kind of the fork in the road that I see. And perplexity has taken one as well as all the other companies that have made an AI browser.

and then I would imagine Google who hasn't released a browser is probably considering this. Open AI has done this and they're kind of taking the way that feels like it has an end game where if you apply a lot more intelligence to a browser, I'm not sure how much better it gets, but if you apply a ton of intelligence to an open-ended agent, well, that's scalable to infinity. It can get infinitely more powerful. It's not constrained to any sort of interface. It can build its own. It can generate it whatever it needs.

So just in terms of structurally speaking, it feels as if, the agent is the more probable outcome to win, whereas a lot of these companies are choosing the easy win quickly, which is just, oh, here's a browser, but AI. Okay, I don't know if I completely agree, because I think a lot of these browser experiences are just going to become agentic, right? And I guess we should define what we're talking about here. Like, I guess a browser experience is, with AI, is you log onto your browser, and and then there's a bunch of summary tools or an AI search engine embedded in there.

But something that's more agentic is something that can not only understand what you're asking it, but then perform actions for you, all autonomously, in a loop, intuitively, maybe not in a loop, but like open-ended. It can solve open-ended problems whilst you're sleeping or whatever that might be, right?

And I think that eventually every browser's heading that way. And the reason why I'm convinced of this is something that I opened up the video with, which is that it has the most personal data set for you. The reason why OpenAI and ChatGPT is so sticky is because it knows so much about me, in my opinion, right? So I keep going back to it because I'm like, okay, it understands the context of what I'm about to ask it. If you think about it, browser is the next natural moat for that, right? Especially if you're trying to unlock AI that can do things for you. Where do you do most of your things? It's probably

on the browser or probably on your phone, on your browser, or via an app or whatever that might be. So I'm guessing that's probably where we're headed. Okay. Well, we will see. I want to dive into some examples. So people who are listening can actually see how they can use this today and what type of interesting things people have been experimenting with. So I see you have a couple you've pulled up. Can you walk us through examples of how people are actually using this tech today? Okay, so this is one. I think he's using perplexity on this one. This is actually something that I did. It looked very familiar.

So in this video here, he's basically trying to create some form. of a travel guide for the city that he's in. He's in Washington. And he's like, okay, hey, I want to go on a walk and see some of the major monuments

and blah, blah, blah. Again, like, I feel like you could just Google search this and find like a million different forums that could do it. But I like that it's embedded

into the app itself so that I don't need to go to a forum, find someone's random Google Maps that they're willing to share with me and then come back to this Google Maps app and it opens it up. Instead, it's all embedded into the application layer, which I like, and it saves me time. Not groundbreaking, but something that's pretty useful, right?

We have something here, which is something that I've just seen a lot of the agent demos do, Josh, which is like, hey, I have a cluttering of random emails. Love. Please unsubscribe for me, and we can just get on with my life and, you know, reduce my spam to basically zero, inbox to zero. And this is something that he demonstrates here. I tried this out as well, and I think you did as well, Josh.

And it takes like a couple of seconds, right? So you basically, two clicks, you like connect PopplexDxD to your Gmail account or whatever email account you use. And then you write a prompt in the search bar being like, hey, like,

I just want to get rid of all my spam emails. And it just, it does all of that subsequently. This is something that I feel like you and I could use sometime. Josh, I didn't actually try this, but, you know, creating content ideas for your YouTube channel.

The difference between opening up a conversation with chat GPT and saying, hey, I'm a video creator. Can you tell me more about some ideas that I could potentially produce versus this? It actually scans your YouTube page itself whilst you're actually on it. Again, I don't think this is a novel experience

because you could technically just copy and paste the link to your YouTube page into chat GPT, do the same analysis. Maybe it takes three extra clicks or prompts, but you kind of get that. at the end of the day. And then this is an example that I keep seeing pop up, Josh,

because OpenAI demonstrated it during their live demo for their new agent release, which is, hey, can you go through my LinkedIn network, accept any kind of requests that you think would be useful towards furthering my career? And also maybe reach out to a bunch of people that I might want to connect with in the future. It's not quite at that capability.

It's still kind of like whatever. I'm not really shocked by this, but I'm guessing it's useful for some. And the final one, which is something that perplexity keeps kind of like talking about, is the fact that it has awareness of all the tabs that you have open. I don't know about you, but I have probably around

kind of like taking a scroll here, but like see different tabs open. And they're all depending to different things, right? And sometimes I lose tabs or most of the time I just forget about a bunch of tabs and why I need to use it and they just kind of like remain dormantly open a praying that I reopen them again, right? What we have here is he's kind of like looking for different bike products that he wants to buy.

He has a series of tabs open on different windows and all that kind of stuff. And what he does is he just goes to the search, sorry, the chat interface for complexity and puts in his request saying like, okay, right, now I need to figure out what model of bike that I want to buy, what kind of parts do I need for it? I'm going on this particular trip. It's going to be, whatever, 150 miles. help me sort this out and it kind of leads him in a step-by-step

process. Again, none of these kind of get me upright in bed. Do you know what I mean? I'm not kind of like, yeah, this I'm being like, wow, you know what? This is magical. This isn't like an whole iOS moment or an Apple Watch moment where I'm like, oh wow,

it's a new device, is a new thing. So yeah, I'm just kind of like left unsurprised, I guess. So I guess to wrap this up, I have one more question for you, which is, A year from now, what do you think you are primarily using to interface with the internet? Is it going to be one of these browsers, or is it going to be an AI tool like chat, GPT, like Rock, like Gemini? What's the time frame?

A year from today. What do you think will be more powerful to use? I will not be looking at a laptop, Josh. If I'm looking at a laptop by that time, we have completely and utterly failed. So I expect by that time to have some different kind of form factor where it's just my eyes and I see everything that I need to that is digital. I hear everything that I need to. If I want to watch a YouTube video, I kind of like look in a particular direction and it pops it open. I hear it. If I want to call my friend, I'm just one to two taps away from doing that or I just say, can you call Josh please? I need to talk to him about A, B, and C.

And it just does that. So it's not a browser. Okay. Yeah, I feel the same way. I don't imagine a world in which browsers are exciting, but I am very excited to ask the person who's making this about why he is investing so much time, energy, effort, money into building this browser because we are having the CEO for Plexi coming on in a few weeks.

We are going to talk to him about exactly this, and we're going to hear the rebuttal from the person who is actually responsible for making this happen. So I know, I'm certainly excited. I'm sure you are too. I cannot wait for that episode. But in the meantime, we'll be dabbling. I'll be trying to find more interesting use cases, more fun things to use the browser for.

If anyone else gets an opportunity to try it, please share in the comments. Let us know what you think about the browser. Are we just missing the plot? Is there something that we're not seeing because this is some amazing new tech? Or is it actually just like, well, it's Chrome with like an AI extension built on top of it. So I'd love to hear your thoughts. Thanks, as always for listening. If you enjoyed the episode, please share like.

If you have any nerdy friends that you want to impress, send him this podcast. It's pretty cool. We have a good time here. So as always, thank you for watching. We'll have another episode coming later this week, and we will talk to you guys in the next one.