The Browser Company CEO Josh Miller on Redefining Browsers for the AI Era

Hello and welcome back to Atlassian, Europe. And we're here with Josh Millers, the CEO of the Browser Company, which is in the process of being acquired by Atlassians. And we can't get into too much detail about the future plans, but welcome to the show.

Thank you very Much. Awesome to be here. First, uh, Atlassian team conference.

All right, So a lot of people are going, well, why do we need a new and different browser and what's gonna change about the experience in the age of AI? And, and so kind of educate everybody about what you guys are working on. Yeah, So the origin of the company is actually pretty humble.

My wife got a new job. She got a brand new MacBook with M1 Ship, super fancy. She worked for a artist who was 76 in Flagstaff, Arizona, not the bastion of it.

Innovation and forward thinking teams. And I noticed that she never left Chrome during her workday. Like even in an old industry, she would get PDFs from old school New York galleries and those PDFs were now URLs and her email was a URL.

And so the origin of the browser company was looking at how people were using their computers and noticing they weren't ever leaving their browser. And their browser was really more of an operating system for their day to day and the fact that their applications and files now in the browser. So the idea behind the browser company arc or first browser and now idea which will be expanding with Atlassian is how would you re-architect and redesign a browser if your assumption was not tabs or information from the Super Information highway from 2030 year, really 30 years ago.

But you said, Hey, these are actually the tools and files that I'm using every day that used to happen on Mac Os or used to happen on Windows, but it's now moved up the stats of the browser layer. So explain how that will change the way we work. 'cause I mean, today I'll open my browser and there'll be 15 tabs on that that I'm kind of trying to mentally connect.

But is there a better way of thinking about all that? Yeah, There's, and we think about it from a couple different, uh, angles. One is applications.

So again, we don't think of them as tabs. We think of them as tickets or documents or videos. And when you think about it that way, you have features like if you're in a Google Meet meeting and then you have to switch away to another tab, that might be a document that everyone in the meeting's reviewing or some research you wanna do quickly on the side relevant to the conversation, we automatically take that Google meet meeting.

'cause we don't view it as a tab. We view it as a meeting and pop it out into a little custom video player just for video calls with little meeting controls that will follow you around to anywhere you go in the browser. So that's an example of if you're someone that is in video calls a lot for your job, every time you do a call, we're making that experience more seamless and more convenient because we don't think of them as dumb browser tabs.

We think of them as smart app objects. So that's one area. The other areas with ai, and if there's one thing we've learned about AI is unreliable and can feel like slop and the things that change both of those downsides of AI are when it has a lot of context about you and your job.

Where's that context? Your context is in your tabs. Why is it in your tabs?

Because your tabs are your apps and your files in 2025. So the idea is that instead of having to export things outta your browser into some AI tool of all the context of what you're working on is right there, then can we teach an AI model when you use it natively in the browser to really already understand the things that you're working on because it's right there with you. But both of those are examples where it's not redesigning the browser for the sake of it.

It's saying the browser is designed for something totally different and this many decades later. Really people are turning to it as the heartbeat of their job. And let's make a browser that reflects that Dumb question.

But what you're describing, is it really a browser or is it something else altogether? Well, Fun fact, we called it the Browser company of New York, right? As sort of a misdirection.

So we've always viewed what we were building as a new type of operating system at the web layer. Now this is an idea that is age old. There have been so many attempts at this.

And our view is that the timing for this idea, in a world where we have devices across, you might have a Windows PC for work and an Apple device for your smartphone and your car and your, that you're gonna want a computer that follows you around all those different devices. And in the modern world where everything's moving to the cloud and there are these proliferation of devices, you're gonna want a browser or something that sits across all of them at the web layer. Now, we don't think the average person, I think you probably, your, your audience may be the one narrow niche in the world that is wants to talk about operating systems, uh, based on top of the web.

But for most other people, that's not a thing they're looking for. So we call it a browser because three to 4 billion people a day use a browser and, and they, they're familiar what to expect. So our hope is sort of like the iPhone one day someone picks it because it is better than Chrome or safari and very concrete ways that they can relate to.

And over time, if we do our job, they'll look and go like, wait a minute, I'm barely touching my downloads folder. Or I'm barely touching the folders on my desktop. Uh, and we can evolve the the browser to really reflect that new need.

Will the people who currently build browsers evolve along your path? Or are they kind of stuck in that lane because they're trying to service a million consumers? Yeah, I mean one of the things that's fascinating is, you know, a year ago we said, Hey, we're gonna make this AI browser thing.

And everyone's like, you're nuts. Like you've been working on this other browser arc. Why are you going to try to invent this new category?

What does that even mean? What is an AI browser? Now there are five or six vendors rushing to build AI browsers, many of them very similar to RSD.

So we view that as validation that actually there is going to be this new category of browser, pseudo browser, pseudo AI is new. Is it the old thing? Updated?

There are gonna be a lot of people. And what we're deciding to double down and focus on is an AI browser for work. 'cause our view is that in 2025, when you open your laptop, you're probably doing your job or school because on the weekend you probably wanna be on your phone or on Apple TV or out in the world doing a activity.

And so if you're opening your laptop, it's probably 'cause someone's paying you to do that. And we wanna make you better at that where others focus more on the broader consumer market. How unique or custom can the experience get for each company or even for each individual user per se?

Uh, that, that is one of the things that is when we started the browser company, we spent the first year actually rebuilding the infrastructure needed to build browsers on top of chromium, which is the web and uh, web rendering engine. And we did it in a way that allows us to make much more novel and creative interfaces on top of chromium than what other browser vendors think can provide. And so if you go look at our first product arc, you'll see that, uh, it looks very different than Chrome and Safari in ways that actually challenged, uh, us in terms of novelty of people learning new things.

But it's also a lot of our beloved arc features that people are fanatical about stem from that infrastructure layer we built that allows us to develop novel interfaces in new ways. You combine that with really the properties of LLMs, which are inherently generative and personalized to each person in query you, you combine those things and we are really hopeful that a, a central value prop of our browser will be and continue to be. It is personalized for you and your job and now your company after the Atlassian deal, uh, hopefully closes.

Um, but we're very excited about, uh, personalization as a, as a defining selling point of our product, especially personalization. That doesn't mean we're selling your data to advertisers or anything of that nature. So within the limitations of what you can say about Atlassian, how does what you're doing and what they do align?

Yeah, one of the things that uh, we have always focused on that is in Atlassian's DNA as well is focusing on the individual in the sense that we want, you know, Atlassian pioneered as, you know, kind of the bottoms up sass motion in many ways. And really in plain English, your, I know your, your viewers don't need to know this, but most people don't. That meant something that you would pick 'cause it was the best in class tool and you wanted to use it 'cause it was best for your job.

And if not that you at the company were using that tool, maybe they would go and buy a license from Atlassian. And so we've always been really focused on the first part of that equation, which is how do we build the most helpful browser for your job, for your workday? But there's that second part of that sentence, which is what the needs of a boss or an IT department or a security department, especially in the age of browser based security with all these SaaS apps is they need a lot of things that are not anything we have expertise in.

And so we're very excited to collectively focus on this AI browser for work focus on making the most valuable tool for the individual that they pick to use. But ultimately, if there are enough people at a company or an organization that're interested in it, what can we offer, um, the real buyers of the software and, and, and, and eventually make teams the be get better together in the browser? Actually, when we started the company was in, uh, 20 19, 20 20, like at the height of collaborative web-based software, Figma, GitHub, notion, Airtable, all these companies were coming out, uh, with very collaborative multiplayer software.

And we always wondered why do you have multiplayer and collaboration at each app? Shouldn't it be at the operating system layer? Should it be at the browser layer?

So there's teams and work, uh, functionality that, uh, we're excited about with Atlassian as it relates to learning how to distribute DIA to organizations that meet their needs. But I also think there's a lot to do, but just how do you make teams work better together? Uh mm-hmm.

Just in terms of their day-to-day work and workflows. Um, how do I secure all this? Because there's a lot of people talking about prompt injections and if your browser access is something somebody can have a malicious prompt and you're connected to everything.

So that kind of means I gotta secure everything, but how do I do that? Yeah, I mean, one of our selling points is security and privacy. So we have, uh, zero data retention, uh, on our, all of our LLM use.

Our memory functionality is end-to-end encrypted, stored on device. So though you have memories we can't see them. They are passed through an encrypted channel to an inference provider.

So everything that is best in class as it relates to security and privacy, that's available today. We do. Now there are a bunch of things that are, no one really knows how to secure yet, like computer using agents, which you're referencing.

We didn't ship them. We were one of the, I think we were the second, uh, external company on open AI's computer use beta. We were in the first five of anthropics.

We had access to these models before they were public. And we said, no, no, no, no, that's too dangerous. It's not worth the trade off.

So one of the things you'll see is a lot of our competitors that are more or less a mimicry of our products have kind of championed how they have all these computer using agents that we do not. And we look in that and say, that is deeply insecure, not really that valuable, wildly expensive and wildly slow. I'm sure all of that stuff will go away.

I'm an optimist and a futurist, but it's not there yet and it's not worth the trade off. So, um, your question about prompt injections is exactly why we don't do anything related to computer use. And as it relates to kind of general assistant usage, we have a five person security engineering team that has been red teaming for nearly a year, just those scenarios.

And so you're not gonna get 'em all, but I'd think if you stack us up against our competitors, we are ahead of the pack as it relates to security and privacy. And after this deal closes, it'd be even more of a focus. So how do you envision the browser kind of getting distributed?

And I asked the question because so many of the browsers we use today are, you know, they're essentially embedded in the box that it comes in or you just click on it and it's automatically installed. Um, are you envisioning that average end users are gonna go download this, or a company's gonna say, we want a browser that is specific to our work environment and our functions and we're gonna distribute that to our Employees? I mean, I think there are a lot of interesting thing, I think this world might change in a, in a world where we're a part of Atlassian, but to speak to what we've done historically, what we've found is if you are someone who spends eight hours a day in your browser for your job, then seemingly small quality of life improvements that make your workflows better are enough to get people to switch.

So for arc, we, we had consistent organic word of mouth growth that was really from people screen sharing something in ARC and someone going, what's that thing? And it starts a conversation about arc. Um, so now Chrome Safari, they're trying to be a browser for literally anywhere, anywhere doing anything on the web.

And so that means a lowest common denominator product that is very difficult to differentiate because you gotta be a great browser for my mom and for me and my little cousin, we're approaching it very differently, which is say we don't need 4 billion people using our browser every day. We want everybody that is in this conference hall to that who works in a browser for their job to go, oh, this is the professional tool made for me in my job. And that sort of software, as we know from this conference hall and how big it is, is if you make teams better and you make developers and individual workers better, they're willing to switch to new tools and they're not representative of the average Joe.

Uh, but that's not who we're going after. We're going after professionals who get paid to work on their computers. So this new project that you're working on, how far along are you guys exactly?

Uh, so DIA is actually available, uh, in general availability today for anyone to download on Mac Windows coming in the next year. Uh, so we had started working on DIA about a year ago. We've been working on arc, our original browser for three or four years, DIA for the past year.

Uh, went into private beta for DIA this past June and it's just sort of been a, a whirlwind since then. Um, yeah, when we released it, everyone, uh, I don't think anyone really knew what we had been up to and what's an AI browser and you know, in the last three to four months, you're seeing, uh, the market kind of surround us and, and, and take their own, uh, attempt on the, on the category. So in some senses it feels like we've been doing this a long time and then you kind of take a step back and go like, oh, we're just getting started, others are just getting started.

We think that the winner of the space is gonna be crowned in the next 12 months. So I think if we're here in a, in a year, we're either gonna have champagne or tears, but it's gonna be a big year and we're really excited to go harder. Are the users starting to come together and be self-supporting?

I'm reminded of the early days of the Mac when the IT department kind of ignored it. So it was just a bunch of end users supporting themselves. Is that kind of the same phenomenon we're seeing here And Yeah, I'm not.

So what we are seeing so far, and one of the reasons again we're in this conference hall is we're seeing, uh, a, a critical mass of employees at companies independently use DIA and their IT teams or their security teams pop and going like, wait a minute, what's this thing? Like, how does this work? Can we get, and so part, so one of the big things we're excited to, to address with Atlassian is, uh, how do we formalize what we're seeing organically develop and how do we just go as fast as we can to bring this to as many people as possible?

And for us, that always starts with shipping excellent software that is actually useful to people in their day to day and have trust and try to encourage that sort of word of mouth growth that comes from the best software tools and technology in your life. And that's a high bar and it's really hard and we don't always hit it. Sometimes we get high on our own supply and like think that we've built something great and we really haven't.

But when you, man, when we built this, you know, integration for, uh, your calendar that pulled out video, uh, URL links and made little nice buttons to join the meeting two minutes before the word of mouth chatter we got from that. Because if you're hopping between meetings all day, every time you're like, thank you ark, thank you a thank you Dia. And so it's gonna be hard, it's gonna be really hard.

There's a lot of competition, but we're, we're encouraged by the groundswell we're seeing and you know, hopefully this deal closes, we'll be able to go even faster. All right. Hey folks, you heard it here.

The way we work is changing and it, it may be in a browser or it may be we call it something else someday, who knows. But one thing for sure, we're not gonna be watching a bunch of tabs from now on. Yeah.

Hey, thank you very much. All right. Thanks for coming back and we'll be back in a minute.