00:00:00
Yo, what is up people of the internet? Welcome to a bonus episode of the Waveform podcast. This is a fun one. A
00:00:06
lot of you have been expecting this one because we shouted it out in our last Friday episode where we talked briefly with Josh Miller, the CEO of the browser
00:00:13
company. This is the full unedited hour-long interview which has a lot of interesting perspectives in it. It was
00:00:20
good for all of us to get to sit down and talk to him about browsers, about Arc, about DIA and to actually get
00:00:26
concrete answers on some things we were wondering about. What if Google tries to do exactly what you're doing? How will
00:00:32
you get people who love Arc to switch to DIA and vice versa? And could features bounce back and forth between them? A
00:00:37
lot of interesting stuff. So, I'll just get right into it. This is a full conversation with
00:00:46
Josh. First of all, let's introduce you cuz you're the CEO of the browser company. And the browser company is a
00:00:52
pretty self-explanatory name, but break down what you do, what the company does, and then we can chat about this stuff.
00:00:58
Thank you so much for having me. Uh, when we started the browser company, people warned me that nobody cares about
00:01:04
web browsers. So, didn't think this is why I would be here, but it's awesome to have made it on Waveform, uh, making web
00:01:09
browsers. So, thank you for having me. Uh, we make two browsers. We make a browser called Arc, which is awesome to
00:01:14
see on both of your screens. You probably noticed it. I appreciate it. Um, and then a new AI browser called DIA, uh, which I'm excited to talk about
00:01:21
today. Is that a reference to DIA Beacon? By the way, naming's hard. Yeah.
00:01:26
So, for one of the people, Dia Beacon had had a connection to it, but uh, yeah, Dia Beacon is an uh, it's a art
00:01:33
museum in Beacon, New York. Yeah. Yeah. I think I remember saying a while ago, Arc is a good name, too. I do remember
00:01:39
that. I definitely remember that. It's a good name. Yeah. And also, Dia is kind of like a new day, a new dawn.
00:01:46
If there's one thing that is hard at startups, naming anything and creating a logo for anything is a religious debate.
00:01:52
Totally fair. We got here. So, I mean, as we've talked about on this podcast, there's been like a lot of chatter in
00:01:57
the browser world. You know, obviously, we're a more niche group of people really into tech, but it just feels like
00:02:02
there's movement with browsers and with people trying new things and new concepts and new ideas of what the
00:02:07
internet is and what search is, and it's all very interesting. And so I figured it would be cool to have you talk a
00:02:13
little bit about number one how we got to arc and then number two how you moved
00:02:19
from arc to dia being the second big focus and big project. Sure. Yeah. So
00:02:25
our original observation that led to the browser company was that it was 2019 2020 and I was actually working at a VC
00:02:32
firm for two years and I was noticing that all of the kind of hot new startups that were coming in they were all web
00:02:38
apps for the first time. You know, for the 10 years before that, everything I'd been working on were mobile apps. Yeah.
00:02:43
And now these companies were coming in and showing these wild new reinventions, documents. Exactly. Yeah. And then what
00:02:49
struck me was every time they came in, they were in this rectangle that was Chrome. And that thing hadn't changed at
00:02:54
all in decades. But the thing that really got me excited about browsers was that observation. And then my wife got a
00:03:00
job with a 76-year-old artist in Flagstaff, Arizona, the least tech ccentric workplace you can imagine.
00:03:07
Yeah. And I saw her in her new job. She never left Chrome. Even in the art world, you know, she got sent PDFs from
00:03:13
galleries and they were URLs that she opened in Chrome. She spent hours and hours every day in Chrome. And myself
00:03:19
and my co-founder and uh the early employees were all consumer backgrounds. People that worked at Snapchat and Instagram were really interested in how
00:03:26
software can touch people at scale. You know, all the people in our life. And we thought, wait a minute, browsers are one
00:03:31
of the most consumer pieces of software imaginable. We you don't think about it like that. like the way you feel about
00:03:36
Instagram or your iPhone. No one cares about Chrome or Safari in that way. And so that to us felt like, oh my god, this
00:03:43
is this dry utility that hasn't changed in two decades that you use for hours
00:03:49
every single day. And everyone's like, yeah, I don't really have an opinion about it. Like where else in tech do you have that? So that was really the simple
00:03:55
idea was okay if our browsers are now really these effectively operating systems with our apps and files and you
00:04:01
don't care about it but you're spending hours every day. What might it look like to build a piece of software that people
00:04:07
cared so much about and made them feel good? Which kind of gets back to where I started is people warned they don't
00:04:12
care. And I think part of the reason we're here today is now people care so much about Arc that I'm coming on the
00:04:17
Worn podcast. So be careful what you wish for I guess. Yeah. So we we ended up in a place where like we all are
00:04:24
super heavy tech users and and a lot of us here I think all of us actually still use Arc like really love the browser and
00:04:31
the product and the features and a lot of the UI. People can scroll back to previous episodes where we've talked about it and so it's it's developed a
00:04:37
community where yes people do care about this web browser. Yeah. And now there's this new project you're working on which
00:04:43
is DIA which is super early. We've all started playing with it and it's it's very different. It's an you've already
00:04:48
said it's an AI based web browser. explain what DIA is and then we can figure out like how it exists in the
00:04:54
world of browser. Yeah, actually and before getting to DIA, I'm curious. I went back and rewatched the first time David brought up Arc on this podcast.
00:05:01
Nice. Which by the way is embarrassing because I had no idea how to explain it.
00:05:06
I was like, it removes my tabs when I don't use them. Arc got introduced to this whole company because David has a
00:05:14
tab hoarding problem and it had a feature to do that and that was the feature that sold David off. that was only one of the features, but then it
00:05:21
slowly grew to multiple people in the office and it's become like a favorite. But I remember being so excited to hear
00:05:26
that David brought it up on this podcast and I remember watching the video and like David, I knew he knew about it and
00:05:32
you really did stumble. I just was like, well, it does this and the tabs are on the side. It's kind of it's weird. And
00:05:38
and it's interesting like hearing you talk about the way that people want to have like an emotional connection with their browser because again even
00:05:45
something uh as simple as having the tabs on the side of the browser. That's something that is very native to Arc
00:05:52
that I think a lot of people are and we'll get into this later I think but like are frustrated about with like DX.
00:05:57
It's back to the Chrome experience. But uh yeah it's just there have been many things about Arc that have just like got
00:06:04
their hooks into us here at the studio. Yeah. And so the the reason I bring it up though is because on that podcast you
00:06:09
were I would say open-minded but very skeptical. And I'm curious for you what
00:06:15
shifted because it it it relates to the DIA story that we'll get to in a second. I'm sure. I think it's one of those things where I didn't care too much
00:06:21
about the browser either. I was using Chrome and then I was switching to Safari on the laptop because the the
00:06:26
battery life was better and I was just kind of agnostic and was willing to try something. So I tried something and I
00:06:31
think it was a couple features that I did like especially for whatever reason the side tabs it's a widescreen machine
00:06:37
like it just made too much sense and I just got hooked. So Arc like quickly became my home and then the syncing
00:06:43
across devices I just like started using Arc a lot. Y awesome. Okay so to answer your question about D and then I'm sure
00:06:49
we can come back to that. So there my personal life and professional life I
00:06:54
keep very separate. You know my personal life it's people I went to college with, high school with they're not in tech. they work in manufacturing or art and I
00:07:02
feel like in the past six months, nine months, all of them have started talking to these AI chat tools for everything in
00:07:10
their life. You know, my friend who works in manufacturing that I was talking about, I was hanging out with him a couple weekends ago and he was
00:07:15
saying there's not a professional project or personal chore that he does not turn to AI in some way to get help
00:07:21
with. And the last time, there have been two times in my life that has happened where I I woke up one day and all of my
00:07:27
personal friends. When I was in middle school and everyone got an A IM and MySpace and
00:07:33
Live Journal and Facebook and then in college starting with a Blackberry and then the iPhone when we started doing things like Instagram and then stories
00:07:39
on Snapchat. This is that third time in my life. And I feel like because we're so in the tech world talking about AI
00:07:46
and I I actually think I did a poor job of leading us about a year ago because I felt uncomfortable about
00:07:51
this in our industry. AI is like a political topic. It's like politics where you have these two extreme parties
00:07:58
and voices that dominate the conversation because they have the time to care about it. You have these AGI maximalists that are like get ready for
00:08:04
UBI. And then you have people that that actually has a reaction to this and I was part of that at the beginning are
00:08:10
like it's slop. It's garbage. It's a dystopian. It's going to write our emails to our loved ones. Like how dumb
00:08:16
it is. And I think when you get out of the building, which we've tried to do for the past year and a half, and talk to people that are not in tech, it's
00:08:22
somewhere in the middle, but I would actually say slightly closer to the AGI side of it is rewiring how people
00:08:28
interface with their computers. Uh there was this University of Nebraska student I interviewed and she was talking about
00:08:34
I turn to it for meal planning and for help with outfits and friend advice in school. And so the foundational kind of
00:08:41
observation of DIA was that what is a browser? It is technically a user agent. Your browser is designed to represent
00:08:48
you to web pages and web servers and bring stuff back on your behalf. And so it seems so clear outside of the tech
00:08:54
world and Arc and all that stuff aside that people wanted to interface with the internet, not just with web pages
00:08:59
anymore, but with AI models and probably in the future agents like deep research. and shouldn't your interface to the
00:09:06
internet be able to both handle web pages and chat and models and agents?
00:09:11
And so that was the observation that made us so excited to work on DIA. We can obviously talk about ARC as well and
00:09:16
the challenges we had there, but that really was the the inciting observation that led us to say, wait a minute, if
00:09:22
we're really building a browser, a user agent for the next 5 10 years, which we have to think about, the world is so
00:09:28
clearly going there, whether or not the AGI versus AI slop debate gets talked about the most. I think the big question
00:09:34
that opens up though is that like the browser company, a lot of people felt very passionate about the browser
00:09:39
company because you guys put so much effort into making the Thursday updates like an event and the the onboarding how
00:09:47
it was so colorful and there's just like a feeling that everyone really got close to and associated with. And so I think
00:09:53
the feeling that a lot of the community is like well if you're going to do this fine but why not just integrate that into Arc? like what's the point of like
00:09:59
starting a whole new browser instead of just putting those AI features into Arc
00:10:05
itself. Yeah. And I want to touch on the Thursday updates and the feelings cuz it's really important. You all were actually influential here. Um what
00:10:11
people miss was the first year of Ark. The first year of Arc did not have a lot of the stuff that people love. And the
00:10:17
way that we got to that place was by testing it early with people who were like, "Nah, I don't like it. Here's what I don't like about it." Man, this is
00:10:23
missing the soul. Like where is that feeling? Where are the craft details? where is the browser company's spirit
00:10:29
and then us integrating it. You better believe that's in the version of Dia when we launch. We just like testing early and often. Now the question of why
00:10:36
not integrate into Arc. Boy did we try like a year of my life trying to figure
00:10:41
out how we could take where we thought the world was going and put it in Arc. There are two problems. I wish there
00:10:47
were three because the rule of three, but there were two. The first is there's this novelty tax that you get when you
00:10:54
try a new product. Not you guys, but the average person that has a job and they have stuff going on in their life and
00:10:59
someone's like, "Hey, you should try this new thing, Arc." They have like 30 seconds that they're willing to give to it. And as David talked about stumbling
00:11:07
over his original pitch, they're like, "Okay, so there the tabs are over here and there are these things called spaces
00:11:12
and then there are pin tabs, but the pin tabs are different than bookmarks and these ways. You got to try split screen." And it's just like people just
00:11:17
couldn't handle how much there was to learn that was new. Mhm. And so we just
00:11:23
felt if the world is going to as profoundly shift as we think it is because of these AI models, how are we
00:11:28
going to teach you how to interface with AI models and agents, whatever the heck comes while also teaching you about all
00:11:34
of those novel concepts? It's a lot. It it just seemed I mean before AI that was our biggest problem. That's why we came
00:11:40
to the office to try to get you on board you know two years ago is like you had someone you worked with saying it was
00:11:46
amazing and you work in tech trying new products you're like no I'm good. Yeah. So that was problem number one. Problem
00:11:53
number two is it touches me and our team that people love Arc so much. It also had performance issues. I think it had
00:11:59
way too many features. We built it in a very prototype experimental way. And what we learned over time is the
00:12:05
importance of speed and the importance of reliability and just your browser feeling snappy. And there were architectural decisions we made in Arc
00:12:12
and then layered and layered and layered over time that even if we thought we could solve the novelty problems, it
00:12:18
would have been really challenging to hit our bar for speed and performance and other things that were important to us. Yeah, I don't quite know how to
00:12:24
verbalize this and if you all think about this with your video creations, but it also felt like Ark was not
00:12:32
finished, not perfect, but it had the right components. It was what it was meant to be. It didn't feel like it was
00:12:38
missing things. And if you go back to our YouTube comments maybe a year ago or so, every time we would push a new
00:12:44
feature, people would say, "I don't want a new feature. I want like Android support. And I don't want a new feature. I want this to be faster." And so I feel
00:12:51
like we don't also don't talk enough in the software world about like when a product is not done, but in the state it
00:12:58
is meant to be. And it is what it is, and you can do with it what you want. And to me, again, people that used Arc,
00:13:04
I'm sure you didn't have that many complaints with the tab model or the craft details. It sort of felt like the
00:13:09
product was what it was meant to be and there was something to just like come to terms with, you know? Does that make sense? Yeah. No, I think that was one of
00:13:16
my questions was essentially is ARC a finished product because it is so different and it is so built and it has
00:13:22
these new fundamentals and now it kind of does this new thing and obviously there's optimizations and and speed
00:13:28
improvements and little things here and there but essentially as an idea is it done? Yes, I feel like it kind of is. Uh
00:13:35
at least that was my take which is why you know I think people who use Arc are are kind of just asking for the little
00:13:40
things now instead of just like massive new ideas. Uh but yeah, go ahead. Yeah. And the analogy that we use internally
00:13:46
is uh or one of them is, you know, Frank Ocean's Channel Orange. There was a mixtape that put him on the scene before
00:13:53
that and that only appealed to a certain type of audience. It was a great mixtape, but Channel was done. It was a
00:14:00
finished work of art. It was a complete thought. And Channel Orange was a how do we make something? I'm not sure this how
00:14:05
we thought about it, but like you know what I mean? It's just it's finished. Yeah. It's very It's funny because in
00:14:10
the in the video world where we make a video, we we kind of we can edit the video forever. Like I can edit one video
00:14:17
for a year and it could never be done. But at a certain point, I have to get to 97% and go on to the next one. And so
00:14:24
the video gets published and it's done and I can never finish it. In the software world, people have this expectation of updates forever. Like you
00:14:31
buy a new phone and you're like, I want to get software updates for as long as possible. if I download and install a browser, I want new features and new
00:14:38
updates for forever. Uh, so the expectation is different, but yeah, I
00:14:43
feel like it's it's maybe a little bit more of a statement of the idea being so different that it can finish and be sort
00:14:50
of published and and that's different in a software world. And honestly, I I think it's I'm surprised we don't talk
00:14:56
about software more as a cultural product, like a video or a piece of art in that way. And that I looked at my iPhone screen for example and many of
00:15:03
the apps that I still love and I've used for decades, Insta Paper, IIA Writer, even Apple Notes, they're pretty much
00:15:10
the same product and I'm happy with it. Like one of the things that I will reflect on in a decade is you know we
00:15:16
started putting out these videos very early on when other tech companies didn't do that and really not with a plan just to sort of be ourselves and be
00:15:22
open and invite the community in and just I don't know as an experiment and a consequence I think of our kind of
00:15:28
experimental prototype driven culture where every week there was new stuff and our experiment of being like let's just
00:15:34
be open about what we're doing meant that we set this expectation that like every week there's going to be some new
00:15:39
hotness and new features and that that we thrived on that for so long. Think about Safari. Is Safari a finished
00:15:45
product? How long is it updates once a year? Yeah, no one complains about that. How, you know, if you whoever you over
00:15:51
there use Chrome. How often are you like, "Man, this new Chrome feature is dope." So, I'm that's I I don't know
00:15:56
what to do with that. Again, it's a byproduct of the way we I have run this company, but I think there's also this
00:16:02
interesting bar where no one's complaining that Safari hasn't gotten like wild new features in a long time. So, that's fair. I have done a horrible
00:16:09
job in many ways. I think communicating about what the heck is going on at the browser company. So I will own that. But
00:16:14
ARC's not going anywhere. Arc is not going anywhere. Arc is not going anywhere. We're just focused on DIA in
00:16:21
terms of where most of our energy is. And then yeah, let's do Chromium upgrades. Let's do security patches. Let's do bug fixes. Let's do the same
00:16:27
things that these other browsers do. Okay. How much of your company would you say that you have devoted to making sure
00:16:33
that ARC is stable? They're probably three engineers at any given time working on one of those topics. Okay.
00:16:40
And but we're but just to be really clear, I don't want to over we are not building new features as of now for ARC.
00:16:46
So but that is very different than like this product is being sunsetted or going anywhere or anything like that. When you
00:16:52
see your two company or two browsers together, do you expect people to move from Arc to DIA or do you kind of expect
00:16:59
the Arc world to stay in Arc and Dia is a new subset of people coming to it?
00:17:04
Yeah, I think we're in a little bit uncharted territory for a software company of our stage and that you know
00:17:09
we called it the browser company for a reason. We could have changed the name to Arc many times and so we've always been excited about having a portfolio of
00:17:16
products and of browsers. Having said that, I do think a lot of the things people love about ARC will come to DIA
00:17:22
in some form. And so the honest answer is like I don't know because one of the things that's been perplexing about arc
00:17:28
is we tried to get to the root of what was powerful about arc for people. And it turned out there were like seven
00:17:33
different archetypes that used like ran you know there was the archetype that uh used the browser in zero chrome like
00:17:40
David is here and kind of flew around with keyboard shortcuts and you had like the space organizer with pin tabs and folders and renamed everything. And so
00:17:47
in terms of are people going to go from arc to dia we're not going to force anyone to do anything. I really like I'm
00:17:52
I'm an Apple fanboy. I'm inspired by that. You got the like the MacBook Pro or and then the MacBook Air and the iPad
00:17:58
Air and it's different things for different people. But I I suspect when we bring things over like a vertical
00:18:04
sidebar, when we take a novel uh take on tab management, I think a lot of people,
00:18:09
not to mention the browser company kind of design flare and craft, I suspect a lot of ARC people will prefer DIA, but I
00:18:16
get excited about having having both uh for the foreseeable future. So what's your plan to move those Chrome users
00:18:23
over then? Right? Cuz if if the whole if the whole idea is it's too complicated for most people, it's equally
00:18:29
complicated to just get someone to move on to something else at all. Like there's a lot of friction there. You're
00:18:34
a small company. There's not a lot of brand recognition for my mom, you know? What's like how how are you planning on
00:18:41
like finding those people who were never going to see the browser company at all? Because the whole point, right, is that you is that ARC is sort of a limited
00:18:48
user base because it's the power users. Y if DIA is supposed to be let's take all of Chrome's market share, how do you
00:18:54
access those people at all? Yeah. Well, that is the big question. That has always been the question for this company from day one. So I don't want to
00:19:00
purport to have like the definite answer, but the theory of DIA is so if you go back to the idea of a browser as
00:19:06
a user agent and it's a user agent for web pages and now it's going to be user agent for models and chat. I think one
00:19:12
of the things we concluded with Arc is that Chrome actually and Safari do great
00:19:17
jobs with web pages. Obviously, there are things around tab management and along sides, but like I think for most
00:19:22
people, they're fine with the way their user agent, their browser opens tabs.
00:19:27
When you talk to those people I referenced, the you know the college friend and my wife about what what is
00:19:33
frustrating about these AI chat tools, we really hear two things. The first is it's a pain in the butt to get that
00:19:40
context out of whatever app or file you're working in into the chat tool. So, it has the awareness of what you're
00:19:45
working on to do the damn thing. The second thing that frustrates them is it doesn't know anything about them. Even
00:19:52
with memory, it knows the chats that you've had, but it doesn't really know what is your taste, what is your writing style, what are the things that you
00:19:58
love, what are the things that you don't like. The browser, the traditional browser and traditional web pages solve
00:20:04
both of those issues. Because as we started with, what are your tabs in 2025? They're apps and they're files. So
00:20:10
you don't have to copy and paste and fuss around to get stuff out into chat. Let's just bring these AI models right
00:20:15
to where you are in the apps and files that you use every day. And then the second bet is that you know, you know
00:20:21
how when you use Instagram reels or Tik Tok, for better or worse, it feels like every swipe, every action like teaches
00:20:27
the algorithm to better understand you. Yeah. The way that it should feel in the future is that in an AI browser like
00:20:33
DIA, every tab that you open, it feels like this model, it's not Sam Alman's
00:20:38
GPT40. It's like GPT David. And it's getting better and better and trained for you every action that you take. So
00:20:45
that when you ask a question about something else, it does not only has the context of that tab that you don't have
00:20:50
to copy and paste into a tool. It also remembers the last seven shopping sessions that you did or the last 17
00:20:55
things that you wrote. like the way that I' I can't quite this is the first external marketing thing I've done so I
00:21:01
don't actually have my language down yet but there's something I've been thinking about that with ARC the sort of hero image was the cluttered tab bar whenever
00:21:08
we showed like seven windows with 50 tabs open people were like I hate that I want whatever this is that fixes that
00:21:15
and that's so interesting because in the old world that was a problem that was clutter that was chaos in the world of
00:21:20
AI models that is like person that's like oil it is it it is the context that
00:21:25
is missing to make these models actually understand you. And so the bet that you have to believe, which I know maybe not
00:21:31
all of you do, is that truly AI is going to change how we interface with our computers and the things that you turn
00:21:37
to it for. If you don't believe that, we're screwed. If you believe that future is going to be a reality, then I
00:21:43
think the convenience of having it right there in your tools and files combined with the compounding personalization
00:21:48
that you get from that awareness, Yeah. is something that at least from talking to people outside the building, that's
00:21:53
what that's what is on their mind right now. They're not complaining about, you know, man, when I go to Google and I
00:21:59
click this link, this sucks. Give me a browser. That's not what's going on in the world anymore. Yeah. I So, I've used
00:22:05
DIA for a little bit and I I think all these are are really great features. Being able to refer to another tab I have open and get context and like and
00:22:11
totally have this be helpful alongside the web for me has been awesome. But the
00:22:18
Chrome question still which is we were just talking about Chrome earlier in the office which is Chrome is somehow one of
00:22:25
the quietly one of the most successful consumer products of all time. It somehow managed to be the default
00:22:32
without having a mechanism to become the default. There's no like I mean there's Chrome besides money. Yeah. They found
00:22:39
their way but like when you get a new computer people just put Chrome on it. That's like the default behavior. And so
00:22:46
we watched Google IO, we watched, you know, Google have all these plans to integrate Gemini and have a a sort of
00:22:52
co-pilot, if you will, alongside the web with you. And they talked at length about how they have all this context. We
00:22:57
plug into your Gmail and your calendar and we know of all your docs and all the things you've scheduled and we can answer emails for you based on all this
00:23:04
context. So in a world where people consciously seek out Chrome and it's an
00:23:10
incredibly powerful brand, how do you sort of kick down the door and introduce
00:23:17
people to DIA and and maybe even have to differentiate DIA from Chrome where they will probably be, you know, trying to do
00:23:23
as many of these things as possible. Yeah, especially because Chrome has this like crazy ability to already have all
00:23:29
of your data. So when when Gemini gets put in Chrome, all of a sudden it already has all the context where DIA
00:23:35
you have to teach it about yourself. The the whole idea of where DIA is going very shortly is the you don't teach, you
00:23:42
just browse. You just So yes, if it's not your default browser, then we can't help you. But if you switch to it in
00:23:48
earnest, just your normal browsing is going to personalize it. But I think the deeper question here is okay, you're competing with Chrome, which has 5
00:23:55
billion monthly active users. And right, isn't that wild? Because that that is
00:24:00
the thing that people have been sleeping on is if you're interested in consumer products, don't talk about Snapchat, talk about Chrome, right? Um, so but
00:24:08
really the heart of your question is like how the heck are you going to compete? Yeah. And I'll say this has
00:24:13
been the question since day one. Prior to DIA, you know, if I was doing an interview about Arc, it was like, man, why can't they just add split screen? if
00:24:19
they just add a vertical sidebar, aren't they screwed? And so that doesn't mean it's not a, you know, a real threat. But
00:24:25
what we've seen is they have a very powerful incentive, which it which is search and search ads. And so, you know,
00:24:32
that Gemini button, we know from someone we interviewed that that thing was supposed to be the default in a much
00:24:38
more bold way, but it tanks search ad revenue. Like one of the things in DIA that people love the most is we train
00:24:44
this ondevice ML model that lives in your URL bar to make it really AI native that when you type a query it parses
00:24:51
that query to understand what is it that you're asking for and it will send you to Google a lot but if we think it's
00:24:57
better to send it to AI chat or any number of AI agents or apps in the future we just route you there. It's
00:25:02
this really powerful thing where you don't have to learn modifier keys, talk about the novel tax. You don't have to learn any new features. We just infer
00:25:09
your intent and send you to the right place. Could Google do that? Yeah. Could they do it better than us? Almost
00:25:14
definitely. You know what? That will tank dramatically overnight. Like the whole premise of an AI browser is search
00:25:21
engines are not as important as they used to be. You should probably pretty rarely go to them now. So, will Google
00:25:28
and Chrome get to the motivation where they want to do that? I'm sure at some point on the top three things that keep
00:25:33
me up at night in my next 12 months running this company. That is not one of them. I think you see in the announcement the
00:25:39
Gemini honestly I saw the video I was like oh no right. Okay. So it's a button that you have to go pay for Gemini. Go
00:25:48
pay for it by the way. So there goes your default there. Who's going to pay that? And then after you pay for it you got to go into Chrome and you got to go
00:25:54
into settings and then you got to turn on this button. You know, it it's it's such it just felt like such a Wall
00:25:59
Street gesture. So, how how then if we can agree that we have some time horizon
00:26:06
by which the search ad model and all their incentives are going to mean they're slow to truly do it AI native.
00:26:11
Yeah, we have a window. What do we have to that? We have only one shot. It's always been our shot. Build something
00:26:16
absolutely awesome that people love. Can we do it? I don't know. But like I think
00:26:21
we are here because we did it once and we built a following for software that made people feel something in a category
00:26:28
that they never cared about before. Gonna be really hard to do it again. But I have faith that if we are if we go
00:26:34
back to our roots and we're as excited and optimistic and creative and open as we've always been, that's the only shot
00:26:40
we got and it got us here. So um Google also knows that people are going to be
00:26:46
moving towards you know agents and and asking their web browser full questions
00:26:51
instead of searching in uh SEO right like we've all been programmed over the years to search in SEO because we know
00:26:57
how search engines work. People are slowly moving towards just using full questions in Google search talking like
00:27:02
a human not a robot. Exactly. And that's the reason they introduced AI mode and now they have the the Chrome a like the
00:27:08
the Gemini agent in the corner. I I hear you when you say like, "Oh, well, they
00:27:13
don't want to destroy their ads business because that's their whole business, but I think they also understand like
00:27:19
everyone else is moving that direction." So, it's your bet that you have you're you don't have any reigns on the fact
00:27:26
that Google's ad business is going to get destroyed if they move towards this business where they're just serving you
00:27:32
answers. and they're more constrained because they want to move towards that cuz they know everyone's else is going
00:27:37
to do it, but they need to figure out how to monetize it. Yeah. Let me let me tell you a story about that and then I'm going to tell you what I'm really
00:27:43
worried about so it doesn't just sound like my founder. It's like, "Oh, we got this figured out." Um, so someone that
00:27:49
previously worked at the browser company, he's a guy named Darren Fischer that ran Chrome for I think 16 years or something like that. and he told me the
00:27:55
story about there was this one feature where on the new tab page underneath the
00:28:02
URL bar they show you your most visited websites and they switched it to the favicons of the websites because it's
00:28:08
much easier to quickly notice the Twitter icon for example overnight tanked global search revenue
00:28:14
by 5%. massive freakout. Now, actually, Sundar, to his credit, Sundar was the original PM on Chrome. That was his
00:28:20
first that that was the project that put him on the map. Yeah, great guy. Sundar seems like a wonderful human. Um, he
00:28:26
said, "No, it doesn't matter. We're going to ship it." But there's a huge impact if a if that tiny change does
00:28:32
five I don't know if it's exactly 5%, whatever it was, what happens when you change the URL bar, which an Apple VP
00:28:39
told me is the most popular text box on all of Mac OS, is Chrome's URL bar. Is
00:28:44
that wild? Isn't that wild? If you change that to say, "Hey, how about 40% of the time we
00:28:51
don't send you to Google?" I just don't I don't see it. I don't see it anytime soon. Now, the thing that really worries
00:28:56
me, which I know I may not be in an audience that will believe this, and we can talk about it. I think the way we
00:29:02
use the web and the internet is ch is going to change faster than we expect. And I think if you get out of the
00:29:07
building and talk to a cousin or a niece in high school or college, I worry much more about a world where AI chat
00:29:14
interfaces like chat GBT or pick your favorite one, actually just replace the browser wholesale at some point uh and
00:29:21
sooner than we think. So what keeps me up at night is is Chrome not is Chrome going to add this button and then the
00:29:26
button is gonna be a sidebar and then they're going to get it. It's going to be like, wait a minute, are we too close to it? Like what what comes really
00:29:33
after? Because if you look at the similar waves, it wasn't the magazine on the internet on the web, right? And so
00:29:39
that that's the thing that I think about is are we being bold enough, which it has a tension with our novelty tax. So
00:29:44
our bet with DIA is make it look familiar. So in that 30 second budget
00:29:50
that we have, you get it on Tuesday at 11:00 a.m. and you can just go. We import everything instantly. It generally looks the same. You feel the
00:29:56
design, craft, and love. and then kind of reveal to you over time how powerful it is in these new ways without you
00:30:02
having to learn new things. The risk I think is actually that we even are being too incremental and we're not being bold
00:30:08
enough. Now, again, I have a bunch of reasons why I think we're in a good spot, but I think if I'm back here in a year and you invite me back and like it
00:30:15
didn't work out. I promise you it's not going to be because Chrome did something with Gemini, it's going to be because
00:30:21
even even though I'm coming on this podcast to say like, I know you think, you know, AI overviews aren't good, but
00:30:27
like people are interfacing, they're talking to their computers, they're thinking with their computers, they're doing things they've never done before.
00:30:32
It's going to be because we even underestimated how much that's going to change the way our user agents and we interface. And I think that's the thing
00:30:39
that people including me missed is I was comparing these tools to Google and I
00:30:44
you know it's like I've been trained for decades to do this like robotic syntax. Yeah. And so what I was speak Yeah. And
00:30:50
so what I was doing is like I was doing that in chat GBT but that's not what people are doing net new use cases. They
00:30:56
were talking to their computers and getting help with things they never did before. And I think we don't appro like
00:31:02
I think there's this weird stigma around it where you talk to someone you get them in a quiet moment and they'll like
00:31:08
I I can't tell you the number of conversations I've had that was like man this is kind of weird I like I'm kind of
00:31:13
embarrassed to admit this but like I had like a 90minute conversation with chat last night about this emotional topic in
00:31:20
my life and I I teared up at the end and people are embarrassed there's like this social stigma because of that political
00:31:26
debate that I spoke about. Mhm. So, I just want to say the quiet thing out loud. We can think it's weird. It
00:31:31
doesn't mean people are having sex with robots, and it doesn't mean that like, you know, we're all going to be out of jobs, but why can't we be excited and
00:31:37
just name that like for the first time in like a decade or two, the way people are like using and doing things with
00:31:43
computers is changing. Like it is it is fun to think about that there's entire generations of people like we've we've
00:31:49
already done this with different like paradigm shifts but like it's crazy to us a bunch of 20 30 plus year olds that
00:31:55
there are people who don't know what a world without smartphones is like because they just became the thing that people use. And I think Google search
00:32:02
has been one of those things for so long where everyone in this room knows how to do a Google search. And there is going
00:32:09
to be a generation that grows up that never has to learn SEO speak and just
00:32:14
punches in whatever they want to know into a chatbot that can tell them ideally the correct answer but some sort
00:32:20
of answer. So that that paradigm shift I think is a fair bet to make and it's
00:32:26
just a matter of like following user behavior and building something that ideally is useful to that new behavior.
00:32:33
Yeah. And that's where that's where I think what we're excited about is not so much trying to better solve the when was
00:32:38
the what year did the you know American Revolution start. I think Google's great for that. It's what are these net new
00:32:45
things that people are doing that they couldn't have done before is where that's the heat we're following or
00:32:50
trying to find. I have a stat right here because I was wondering this and I learned this before but approximately 15% of all daily Google searches are
00:32:57
brand new which is interesting. That was already interesting to me, but I bet the number of things people are typing into
00:33:03
these chat bots that have never been typed before is skyhigh. Yeah. Is extremely high just cuz we see that
00:33:08
behavior is so new. [Music]
00:33:21
Yeah. Is AI a feature or a product? I feel very
00:33:28
confident that in five years on Mac OS or Windows, the application that has
00:33:36
default browser permission will look much more similar to an AI chat interface than a browser. And the web
00:33:43
pages that you access within them and the tabs you access within them are going to feel much more like tool calls
00:33:50
or tools that that AI chat interface can wield for you and open for you. Which
00:33:55
doesn't mean you're not opening a big Figma link. doesn't mean you're not typing in Google Docs, but I I
00:34:00
absolutely think without question the hierarchy is going to flip. I'm not telling you it's going to be DIA for
00:34:06
sure, but I feel incredibly I mean, think about all the heat that we're getting right now. That's not fun. Yeah.
00:34:13
Right. It's not It's not It's much more enjoyable to be like the art guy and everyone's like, "Oh, I love the animations." I wouldn't do this I would
00:34:20
not do this if I didn't believe in my bones that this is one of those moments where the paradigm is going to shift
00:34:27
which doesn't mean we're going to succeed but I without question and I'm like I kind of wanted to come on here and give a little more of the raw to
00:34:33
time stamp it for myself honestly like it might you know I don't know I could be wrong I could I could be I could be
00:34:40
totally naive but why I feel that conviction is this didn't come from TW I got off Twitter this is I I literally
00:34:46
went to college campuses is and just spent time with people unrel and it's just can I ask one more question? I I'm
00:34:52
here as long as this is like cathartic for me. Yeah. Um how are you going to make money? We we are going to charge
00:34:58
for DIA. That is one of the other really interesting cultural shifts is like when we started the browser company the idea that people would pay for software in a
00:35:05
kind of broader sense was not a popular idea. And one of the things with these
00:35:11
AI tools I'm an angel investor in this company called Cursor that makes it's sort of like DIA for coding. It takes a
00:35:17
traditional IDE VS code and it doesn't just add a sidebar. It kind of imbuss these AI models within every part of the
00:35:23
IDE experience. It is the fastest growing software company of all time in terms of revenue ramp in terms of the
00:35:29
amount it charges every month. Now it's very different. Those are software engineers and it's much more valuable than D is today. But I think there's now
00:35:36
finally a model where you can say hey if you really offer something that unique and valuable especially because what is
00:35:42
our argue? Our argument is going to be this thing knows you better than any other AI chat tool. It knows you better.
00:35:48
It really gets you and it's right there. Um so yeah. Now when you say charge for DIA because I know as of right now we're
00:35:54
testing a version of it that's super early. Is it going to be you're going to charge for the whole browser or are you going to charge for a feature inside of
00:36:00
DIA or what's the plan? No, we're going to we're going to charge for a kind of a premium bundle at some point. So, one of the things that's really interesting is
00:36:06
we we spoke briefly about how we built this ML uh model in our URL bar that
00:36:11
routes you to different places. There more and more of I hate the word agents, but there are more and more of these like AI apps or agents coming online
00:36:18
that are very specialized and actually often string together a bunch of different models and kind of custom prompting and tool integrations.
00:36:25
You can imagine a world where just the base DIA is free, but then you know you can buy the like software engineering
00:36:31
bundle or the sales and marketing bundle or again I'm I'm making this up but I that's another thing is man this
00:36:37
industry is so bad with like naming and human words but this concept of agents
00:36:42
though it is annoying to say that word is is is really real and powerful in the things you can do when you string these models and tools together. And so I
00:36:49
think as you see more and more of those kind of verticalized applications that are native to AI, being able to unlock
00:36:55
those and route you to different places. Um, and do you believe like a mobile model works with that too? Cuz like
00:37:01
we're not, you know, there's all these ideas of this company IO taking away your phone. See if that happens. But uh,
00:37:09
do you see DIA also being on a mobile interface? You have to be, right? I
00:37:15
think that was one of the reasons that you even took a while to get on art because we weren't on Android. So I think you have to be on mobile. I think
00:37:20
the interface on mobile is much more of a remote control. So if you think about the desktop browsers primarily, go back
00:37:26
to the Tik Tok analogy. Every tab you open, it's getting personalized to you. But actually that personalization is
00:37:32
owned and controlled by you and that's why we're, you know, doing things in the browser. Well, now if that is like kind
00:37:37
of personalizing the algorithm for lack of a better framing, when you're on your phone, you're just going to want to talk to it. Uh and you're going to want to
00:37:44
get its help, but it doesn't it's kind of based in that awareness and knowledge. It's compounded from desktop.
00:37:49
Um, but I think similar to Arc, what a browser is on mobile is very different than desktop. So I do not imagine a
00:37:56
world for example where anyone is mobile only with DIA. That would miss the raw power. You know, it really needs and I
00:38:02
think if you go back to our first YouTube video and the original idea for the browser company, it was this idea of an internet computer which was because
00:38:10
the the browser is popular because it's now an operating system because your apps and files are there. What that means is your computing life is in the
00:38:16
cloud. It's all stored on a server somewhere. And if that's true, then your computer in air quotes should be able to
00:38:21
go with you to any device because it's all just in the cloud. Text follows you. Exactly. That still holds in this world,
00:38:26
which means that all right, if you're just on your mobile phone and you don't want to use D on desktop, you're you're missing the power of that. But I think
00:38:32
whether it's open AI or Google that is where you know I think the thesis was
00:38:37
right whether or not we win or not which is your computer is going to increasingly feel like this deeply
00:38:43
personalized AI model that goes with you across all devices and different devices
00:38:49
play different roles with mobile being more of like easy input probably voice centric with desktop playing the
00:38:56
personalization engine that really knows what's going on in your life especially your livelihood. I don't know what this IO thing is going to be or your TV or
00:39:02
your car, but like you better believe it's going to be there too. Good luck to them. Uh I guess while we have you, the
00:39:08
question personalization is often like on the opposite end of the spectrum is privacy where if I am using Google's
00:39:14
thing, Google has to know everything about me to deliver that level of personalization. And it's interesting hearing you guys have made your own
00:39:20
models and are going to build things that like keep all this context and knowledge about you. How much are you thinking about privacy? And if people
00:39:25
care a lot about privacy, should they like something like this? Yeah. uh our view is giving people control and giving
00:39:33
be just being very transparent. I think one thing that is true though is to get the most power from these models, it
00:39:39
needs your personal context. Like it's not even worth it if you're not willing to teach it about yourself. And so I
00:39:44
think at some point actually these these these MacBooks and your phones are going to get strong enough and open source models are going to get small enough and
00:39:51
powerful enough that they can run locally on device. In the meantime, you have to be okay sending your data to an
00:39:57
API that is going to, you know, integrate it into the model. Um, but I mean, as you know, what we found is like
00:40:03
people will make that trade. Actually, the same friend I was talking about, he's going to hate me for saying this. I had this moment where I was like, you
00:40:09
got to ban Tik Tok. Like, this is just this is just bad for like get get rid of it, right? Uh, and he's like, honestly,
00:40:16
it brings me so much joy giggling at these Tik Tok videos all day. like the CCP can have it all if like my time on
00:40:21
the toilet feels like that enjoyable. So, I would say that is on one extreme of the spectrum, but generally I think
00:40:27
if you show up to someone like, "Hey, you're busy. You got a lot of stuff going on in your life. Here's this thing that can actually help you in
00:40:33
meaningfully new ways and all you have to do is browse like you've always browsed and we're just going to use your browsing data in, you know, to to train
00:40:40
an AI model to be better for you." I actually think as many people would be willing to do that. Andrew, you had a
00:40:46
question. Yeah. Yeah. I had a question and it's not DISA specific but AI browser specific and maybe since you're
00:40:52
in this it's a question we ask all the time and might be able to shine some light on it but we talk all the time
00:40:57
about these like uh conversational like searches instead of old school links and
00:41:03
everything like that. And a lot of that comes down to like what's the best TV I should buy? you know, asking a question is consumerbased, but it's scraping data
00:41:10
from websites where a reporter spent time where that reporter gets paid by advertisers on the site. How or have you
00:41:18
thought about at all whether it's DIA or just in general like how are people like that compensated or even like you know
00:41:24
we make YouTube videos and we there's lots of stuff about chat GPT probably scraping a ton of YouTube videos so like
00:41:31
we're losing viewers the people who are doing the original personal content are
00:41:36
potentially losing out and then it gets to the old dead internet theory of later on like who's going to be making these
00:41:43
articles that AI scrapes from anymore. Yeah. So, sorry that's a very big question. No, I I don't get to come on
00:41:49
the podcast and just get the softball. I was like ask I'm I'm here as long as you guys want. Um, so there's the kind of the DIA specific what we've observed and
00:41:56
I think there's the larger ecosystem question. On the DIA side, this goes back when when we started when I first
00:42:02
had my own ideas about AI. I was thinking about it much more in the what are the things I already do in a search
00:42:07
engine or browser and how do we how does AI make it better? And that's where I was missing the forest for the trees. But we see in Diaz, people aren't using
00:42:15
it to do that sort of stuff. It's a it's almost like a companion or a partner on
00:42:20
whatever they're working on. So, you know, an example that we put in a video which came from a college student is you're buying a used car. You have two
00:42:26
links of similar models side by side and in Dia split screen, which is the same as ARCs, and you're like, "What's the
00:42:32
difference? I'm not a car guy. Like, help me understand the difference." And then you have Dia's sidebar or Dia's um
00:42:38
chat sidebar as a way to kind of like work through and understand the difference between those. That's true with writing. You have a a document open
00:42:45
and you're working through it and you have it right there side by side. So what we're seeing is the most popular use cases in DIA are less of a hey I was
00:42:53
doing this before and now I'm doing this and much more like I couldn't turn to my computer to have opinions before. My
00:42:58
computer never had opinions. It couldn't be subjective and now it can. So that is kind of what I was saying before about the narrative. we get so locked into
00:43:05
these kind of memes in our industry that I think we're missing what people are generally actually doing. So, it's less
00:43:10
of a uh I'm going to Gemini, I'm going to chat GPT and just asking random questions and it's more of a I'm using
00:43:16
my browser like I've always used a browser, but now I'm just understanding the context of what's on the page
00:43:22
better that I've already opened. Again, I haven't been on any press or anything yet. So, I I'm still working through how to talk about this. But think about it
00:43:28
like this. One of the ideas behind ARC was that how do you put what's what
00:43:33
matters to someone's life at the top instead of these tech ideas. And what I mean by that is like you got to buy a
00:43:38
used car. That's at the top of the pyramid. That's what matters to you. And so in order to do that, you need car
00:43:44
sites and you need to read things and you need to check out the photos and all that stuff. What is new is now our
00:43:50
computers can have the perception of thinking, having opinions, giving feedback, critiquing, joking. And that
00:43:56
is just a new tool in the toolkit of buying a used car. And so what's new is that sec in my opinion it's that almost
00:44:03
emotional intelligence not the IQ that is so new and what people are really turning to it for. Not just doing the
00:44:10
thing we used to do that old tool we had in a new way where there's definitely that to come back to your ecosystem
00:44:16
question which is what I think you were really answering though. This is where I think other people go on podcast and stuff. I have no idea. No one has any
00:44:23
idea and I think it's one of I don't want to I don't know if scary is the right word but like the web as we know it was broken is breaking even faster
00:44:32
and some it just doesn't make sense anymore. So I don't I don't not only do
00:44:38
I I not have an answer to what replaces the web and the incentive structure. Um
00:44:43
I haven't heard anything from anyone that suggests and and it's just this
00:44:49
complex problem where it's what's so wonderful about the web is it's decentralized and it's open. There are so many different actors and players in
00:44:54
it that kind of need to work together and in 2025 we don't generally work together on things that affect all society very well right now. So, I don't
00:45:02
know, man. But this is why it just it feels like it's coming. And our approach
00:45:07
has always been we have this value uh assume you don't know, which is like we have no idea what we're doing. So, how
00:45:13
do you proceed if you have no idea what you're doing? You build stuff really quickly. You go to the MKBHD office, you give them dia, they tell you you got a
00:45:20
lot of work to do, you go back to work, you come back. And so, like our approach is like just get in there, have these conversations, and try our best. But in
00:45:26
terms of like how are publishers like insert your favorites publisher that
00:45:31
post textbased blog posts that were accessed through Google search engines.
00:45:36
What happens after that? Yeah, we're about to find out. It is a big question. I was just going to
00:45:42
mention because like I one of my biggest pet peeves is I'll be on Twitter and I'll post a video and someone will be
00:45:48
like, "Grock, tell me what this video says." And then they'll like summarize the video and be like, "Great. I'm not going to watch the video now." Great.
00:45:53
Okay. So now half the people who were going to watch it lose that information bit, but you know some people will still
00:45:59
watch it for entertainment value. And you know to to use the like buying a
00:46:04
used car analogy there there are so many different behaviors that I see which is there are people who just want to speedrun learning about the car, don't
00:46:11
really care too much about it but just need to know like what's the difference between these two cars and that is going
00:46:17
to pull from some motor trend article reviewing the car passionately. that's
00:46:22
going to pull from some YouTube video of a guy who used the car for years and did a daily driver 30-inute video and that
00:46:31
kind of breaks a little bit when the incentive goes away. But it also will potentially be a world where someone
00:46:37
will go to the motor trend article and will read some of it and will ask the the sidebar chat for more information or
00:46:44
more context to better understand it or they'll watch the 30 minute YouTube video and 10 minutes in they'll pause and ask for some more context to better
00:46:51
watch the video. So there's different behaviors. There's different directions that this could go. We don't know. No
00:46:57
one seems to know exactly what's going to happen, but it's interesting. And I would take the other side on one specific part. I think that the most
00:47:04
kind of soulful original in-depth brands and publishers are going to do better
00:47:10
than ever. Right? I actually think the MKBHDs of the world, I don't think people are dumb. And I think when you're
00:47:18
going to buy a gadget or you're trying to understand something that matters to your life, I think you're going to turn,
00:47:24
especially in a world of all this AI generated everything, I think you're going to turn to the best of the best
00:47:29
more than ever. And I think you're going to be willing to pay more and do more than ever before. What happens to the longtail of
00:47:37
the blog that covers air conditioners in a kind of more SEO driven way and that
00:47:42
sort of that that type of business and media company? I have no idea. But like if I can invest an MKBHD in the world of
00:47:50
AI, I would invest a lot of money. And I'm not just saying that cuz I'm here. I really believe it. And I think in some ways you even see that a little bit with
00:47:56
the browser company. Like I've I've always thought about how people's love for maybe love's a strong word, but um
00:48:04
endearment to what we're doing and how we're doing it far outweighed our scale. You know, like we wrote a Substack post
00:48:10
yesterday that announced nothing new. It just shared some new data in a clearer way and it like was at the top of tech
00:48:17
meme which like everyone in the industry reads that is that is disproportionate with our user with everything. I it was
00:48:24
cool but it didn't make any sense to me and because I think the way we showed up in the world had a little more soul and
00:48:30
spirit and opinion and personality than other startups did. And I think that
00:48:36
across industries including media is just going to continue. And I think it's going to by the way I don't know it has
00:48:41
nothing to do with us but I just I just think people are long they can feel that character and I think in a world of AI
00:48:47
people are going to gravitate towards that more and more still breaks a lot of media you know I don't know who sells ad
00:48:52
for this podcast but they probably disagree but like I think you all are good you all are good you can rest easy that you all are good our like anecdotal
00:48:59
version of that is no matter how much we make of like a review of something Marquez will get x amount of pe like
00:49:07
dozens if not hundreds of people being like oh yeah but what do you really think of like this phone and like that feels like the type of average person to
00:49:14
me that's going to go into that AI chat and be like what does Marquez like which
00:49:19
phone should I buy and then not actually go into our video that we spent a lot of time making and and let it decide for
00:49:26
them and like when it's pulling from our stuff or when it's pulling from a Verge reporter doing a review of that that's
00:49:32
where is is your are are the graphs that you care about relative to 12 months ago
00:49:38
are they up are they in the middle or are they down. I think they're in the middle. But I've always also thought about the videos that and we are so in
00:49:44
the weeds, but it's fine. We is I think the the I think of our
00:49:50
audience as kind of two main buckets. One subscribes and watches videos purely for entertainment value and not for any
00:49:56
information or should I buy it at all. And so they'll watch I one of the most common questions I get or answers that I
00:50:01
get is, "Oh, I love the videos. I never buy anything you're making videos about, but I just like watching them." And then the other bucket is purely
00:50:08
informational. They are here because they are on a quest of finding information and figuring out if they should buy the thing or not. And I think
00:50:16
part of what we're talking about here is wiping out one of the buckets, which is that second one, which is I am here for the information. Pause the video, ask it
00:50:23
a question, get the answer, peace out. And the entertainment portion, which we happen to have an audience that's here
00:50:29
for that, too, will remain. And I think what we're probably more concerned about is those that have way less of the
00:50:36
entertainment value and are more on the SEO end of like we have a mission of like delivering information about these
00:50:41
products and people don't like show up here for entertainment necessarily. They just come here for the information.
00:50:47
Yeah. And now those people aren't going to come here anymore. Yeah. So that's that's one of those scary questions I think for that specific type of
00:50:53
publisher. Yeah. I would say I I mean I think it depends on the query. So I think for a query about the American Revolution is different than I'm going
00:50:59
to buy something in my life. At least what we've seen, it's very anecdotal, is I think people treat the first result
00:51:05
back from an AI chat tool in the same way they treated a Google like results page. It's a jumping off point. I don't
00:51:11
think that's true for all types of queries, but for these types of queries, um, but I, you know, David Pierce from
00:51:17
The Verge has said something again and again that has really stuck with me that is, and I'm going to paraphrase, and
00:51:22
sorry, David, you did this on your own podcast about something I said, so I I got to do this Pierce stuff. Um, just
00:51:28
like media companies need to make the best products and they need to think about what they do as products. And so I
00:51:34
have no doubt that whatever is going on in your graphs, whatever you're worried about, this crew and this office setup
00:51:40
is awesome. You're going to make a better product. And so that's actually part of the reason I'm kind of like doing this and doing the Substack post
00:51:46
is I regret not being unapologetically uh open to the idea that this was all
00:51:53
going to change even if I didn't buy into this AGI and all this other stuff. And like what do we got to do to be
00:51:59
ready? And so I feel like in some way it's like okay if that's what you're worried about you have so much
00:52:04
creativity in the tools here that I have no doubt you will make a better product for I want to buy X how do you do it
00:52:11
than whatever thing is going to pop out of some AI model. Yeah, I think yeah, there's gonna again we don't know what
00:52:16
the behavior of the user in the future is. We know what we see now and we know where the trends are inflecting towards, but how much they change and whether
00:52:23
they totally flip on their head is a completely unanswered question, which I think is something we might lose a
00:52:29
little bit because we're so focused on the now. And I think it's an interesting thing and it's and it's it's it's scary and it impact again like it it
00:52:35
challenges a lot of people and companies and but you know when we started the browser company the reason we called it the browser company of New York was
00:52:42
because we were so bored of startups and startup names and startup products. It was just like a boring moment, Quibby.
00:52:48
Right. Yeah. And so the reason I say that is like I think it's it is exhilarating if not a little bit scary,
00:52:56
but it's just there's a moment of creation and rebirth happening here and and in in some ways that are a bummer
00:53:02
and in some ways that are like, man, I'm I got a I got two sons that I'm like, man, I'm not excited for how we deal
00:53:08
with that in a bit. And in other ways it's like I bet you come here every day to do things like how do we do this new
00:53:14
product challenge? you know, like you wouldn't be doing this. You would have sold this company or done this a lot differently if you didn't enjoy that
00:53:20
challenge. So, that's how we're trying to show up to this is like, yeah, we have no idea, but how privileged are we
00:53:25
that we are at the age we are in the industry that we are in this moment in time? Like, yeah, I know that sounds
00:53:31
very cheesy, but to ground us a little bit, um, do you have like a timeline of
00:53:37
when DIA will feel like this product that you are sort of pitching? Because I
00:53:42
think right now I think a lot of the backlash from Arc being I don't know
00:53:47
about Sunset but maintained and DIA coming out is that be like part of the
00:53:53
reason that people love Arc is because you built in the open right and now you're building DIA in the open. Um I
00:53:58
think a lot of those ARC users love the browser company so they were frustrated that Arc was being sunset and not being
00:54:05
sunset not being maintained journalism sorry being maintained. uh is that the
00:54:11
version of DIA that people have access to right now is very bare bones and doesn't necessarily meet the the more
00:54:17
grandor uh ideas that you have for how you're going to interact with this AI browser.
00:54:22
So, do you have like a timeline of when people are going to be able to play with something that has those capabilities?
00:54:28
Yeah, it obviously depends on what bits you're talking about, but I'll I'll try to answer. Yeah. Um by the way, yeah, at
00:54:33
some other point over beers, I'd love to talk about the pros and cons of building in public and being transparent in public on YouTube. Um, but yeah,
00:54:41
actually when my second son was born, I like left the hospital in Paris, actually uh to go get food for my wife
00:54:47
and someone's like, "Hey, Arc, I love it." I was like, "Whoa, this is like a new this is not what I thought was going
00:54:52
to." But any event, to answer your question super directly, um, I'd say if you're someone that tried Arc or has
00:54:58
never tried Arc and you try Dia to get to the bar where you're like, "Oh, this feels better than Chrome 6 weeks." You
00:55:04
are on an old version of it. I feel good about that. terms of the both I'd say grander of what I'm talking about which
00:55:10
again was always there with ARC we're going to be internet computer operating system for their web the grander and the
00:55:15
I'd say ARC members feeling like it has enough of the basic you know the things that they the vertical sidebar I'd say
00:55:23
somewhere between Labor Day and Thanksgiving um but again on the hey you
00:55:28
just browse like you normally do and this like model self-personalizes to you that's going to be a many many year
00:55:34
thing but what we did with ARC was just like let's be honest about where we're going and build it in public and around
00:55:39
people like you know the other day one of the things we're going to release soon in in the kind of when we give it to ARC members um very soon is there
00:55:46
were these college students that were hacking our personalization features to make these mini apps like almost like AI
00:55:53
apps so they created this syntax where they would be like when I do back slash
00:55:59
gadgets I want you to do these 17 things and so then they'd go they'd hit a new
00:56:04
tab and do back slashgadgets and then whatever they wanted to do and they kind of like made their own little apps and
00:56:11
we're like well and multiple did for different different use cases and this goes back to the like native to the
00:56:16
technology native to the phone we're like well that's pretty wild and so what we've been sprinting to do you know that
00:56:23
we weren't going to do five weeks ago is like man we got to formalize this and see if we make it even easier what other people do and so that's all to say part
00:56:29
of the reason that you know the grander or it won't be there you know right away is the stuff's hard and it takes time
00:56:36
But part of the reason is the reason ARC is so beloved is because we were like we have no idea what we're doing. Let's put some stuff out there, see what people
00:56:42
do, react to it. And so we also want to have time to say maybe the big idea here is actually AI native apps aren't going to be these agents. They're going to be
00:56:48
these like user created and shared little mini apps that it's an idea. It's an idea. Definitely an idea. It's a
00:56:54
definite idea. I guess uh I'll throw out one more and maybe we end this with this is a a feature suggestion please. So I
00:57:01
think you mentioned earlier, you know, Arc is maintained, DIA is rapidly evolving and maybe adds features from
00:57:07
Arc. Yes. And I think you you spoke a lot about the novelty factor of like trying to keep it familiar, easy to
00:57:12
understand, doesn't look too crazy. There's this setup process on a couple different phones. I think Asus phones is
00:57:17
one of them. Super niche. But oh yeah, I log into the phone, I set up the phone, and then I land on a splash screen. And
00:57:23
the splash screen is a direct fork between how do you want this to look? Just like Chrome or just like Arc? And I
00:57:29
think if I if I was logging in for the first time and I was going to set up DIA and I was showing this to my cousin, he
00:57:35
would just fork over and use the version that starts off looking just like Chrome. And I think a lot of people would see the make it look give me the
00:57:42
arc stuff button and they would click that button and they would love it. I think that would be a hero button for
00:57:48
DIA if it you know obviously I think you want it to be a mass product and you it'll be a focus for the for the future.
00:57:53
I think a lot of people would love a lot of these features in DIA with just a click. I love that. I mean, and the
00:58:00
original like internal saying we had was we for ARC was we want to make it your home on the internet. So, it doesn't feel like you're browsing in some
00:58:05
generic hotel room, but it's yours. And I think this is very much in the spirit of that, which is like your browser, your user agent should look like
00:58:11
whatever you want it to look like. So, yeah, I love that. We'll come back on the pod when that happens and and give you a shout out. Bet deal. All right.
00:58:17
Thank you so much. Thanks for joining us. Awesome. It is. Thank you. All right. That is it. Thank you again to Josh for joining us and let us know what
00:58:23
you think. Are those answers convincing? Do we think this browser has a future? Would you use something like DIA? Would
00:58:30
you use something like Arc? Are you still going to be a Chrome person? Let us know in the comments section below. But either way, we'll see you guys very
00:58:36
soon back with your regularly scheduled programming on Friday. See you then.
00:58:41
Peace. [Music]
