Arc creator Josh Miller on why you need a better browser than Chrome

Today, I’m talking with Josh Miller, cofounder and CEO of The Browser Company, a relatively new software maker that develops the Arc browser. David Pierce, my Vergecast cohost and Verge editor-at-large, is a big fan of Arc and has written about it quite a bit for us. You can read his review here. 

Basically, Arc is a ground-up rethinking of the web browser. Most modern browsers started as simple document viewers and grew to support running complex apps. Arc’s main conceit is that it’s designed to make running and using all those apps as simple as possible. You’ll hear Josh describe it as an operating system several times, which is a pretty big claim to make, and he and I got into what that actually means for a web browser.

There are some AI tools built into the Arc browser, but the company also has a mobile app called Arc Search that does AI summaries of webpages. That puts it in competition with OpenAI’s forthcoming SearchGPT and Google’s Gemini-powered AI Overviews in its search results. At the same time, it also puts Arc right in the middle of one of the fiercest debates in tech and media today: whether AI companies and products are boosting content from the open web and then turning around and selling it to consumers — all without paying the people who produced that work anything at all.

We’ve been talking about these topics pretty much nonstop for the last year here on Decoder. So I was really excited to have Josh on the show to explore why he built Arc, what he hopes it will accomplish, and what might happen to browsers, search engines, and the web itself as these trends evolve. 

I wanted to know how Josh is thinking about competing with Chrome on the desktop and Apple’s Safari on mobile, and especially how he plans to monetize Arc. Chrome and Safari are a lot of things, but mostly, they’re developed by some of the richest companies in the world and given away for free. Josh says the plan is to keep Arc free but monetize a mix of customization, automation, and productivity tools that will make users’ lives so much easier that they, or the company they work for, pay a subscription fee.

It’s a bold idea to bring competition back to the browser market, and early reception to Arc has been positive. But you’ll hear Josh and I go over some of the major challenges they’ve faced so far, like having to teach people all-new sets of metaphors and design language around what browsers should be doing and why you would even want to use a web browser to run apps the way Arc is suggesting. (Or why you’d want to use a new browser at all.)

I also asked Josh about his take on the controversies swirling around generative AI and whether the web as an information distribution system is going to survive a major plundering of all its pages. Josh is pretty candid about what he does and doesn’t know about how this might play out, and he’s also more open to changing his mind than arguably any tech CEO I’ve talked to about this subject. It’s a good back-and-forth, and I’m curious for your feedback on it.

One quick note before we start: after we recorded this conversation, The Browser Company disclosed a pretty severe security vulnerability in Arc that could have let attackers insert code into other users’ browser sessions. It was patched a day after a researcher made the company aware of it in late August, and the company says no users were affected. But it’s a significant issue, and in a statement released last week, the company said it marks “the first serious security incident in Arc’s lifetime.” 

We tried to get Josh back on the show to talk about it, but he was unavailable the day the flaw was disclosed to the public. The company does say it’s making a lot of big security improvements. And in a separate statement on X, cofounder Hursh Agrawal said, “A heartfelt thanks for all the concern (and even outrage) you’ve all expressed about this incident, and for holding us to a high standard.” He went on to say that he and the company will “be using this opportunity to grow as a company, as an engineering organization, and personally as a founder.”

Okay, The Browser Company CEO Josh Miller. Here we go.

This transcript has been lightly edited for length and clarity.

Josh Miller, you’re the cofounder and the CEO of The Browser Company. Welcome to Decoder.

Thank you for having me.

I’m excited. We’re in the studio together in New York. It’s a rare occurrence on Decoder. Thank you so much for being in person with me.

Oh, it’s so fun. I was hoping you’d go easier on me, but I was told that it is absolutely not true.

Oh, no. When you’re in person, it’s even harsher because I can smile at you while—

While plotting.

Yeah.

Well, it’s great to be here. I think I’ve probably listened to almost every Vergecast and Decoder interview episode for the past few years.

Oh, you’re ready.

The studio is as nice as you said it was.

It is. We have a fancy new upgraded studio. I’m happy you’re here. There’s a lot to talk about. The Browser Company runs a browser called Arc. You run a mobile app called Arc Search, which is browser-adjacent, I would say. It lets you browse the web in a new and different way. 

You’re obviously competing with Google. Google appears to be in a moment of change — regulatory change, self-imposed change — and then there’s AI. And obviously, Arc Search is built in as an AI product. But let’s start at the very beginning. What is The Browser Company? What is Arc?

The Browser Company is making a web browser called Arc, which the simplest way to explain it is The Verge called it “The Chrome replacement that they’ve been waiting for.” So, don’t take it from me.

David Pierce called it that. I just want to be clear.

[Laughs] David Pierce said that. No, Arc is the best browser for laptop people. If you’re someone whose livelihood is clicking and clacking on your keyboard every day, we make the best browser for you [that] keeps you focused, organized, and increasingly, we want to do your busy work for you.

Let me ask you a question about that. So, if you are somebody who makes money on a laptop, you’re presumably using a lot of applications, not looking at a lot of content. I would love to be a person who made a lot of money on my laptop just by looking at a lot of other people’s content, but I suspect what you’re getting at is this is a productivity application.

So, the origin of The Browser Company is I was a political appointee in the Obama White House and after the 2016 election, I was personally devastated by the result. I felt like technology and the technology industry had an impact on the things I didn’t like, and I was very motivated to try to do something about it.

My takeaway was, if you are not an operating system, if you’re not a platform by which your applications and content sits on top of, you don’t really have leverage to change for the better or worse the way that society uses technology. So, we decided not to start a company and do something else. And then it was in 2019 — my wife works in the art world for artist James Turrell in Flagstaff, Arizona — that I noticed that she never left Chrome. She was on this high-powered MacBook Air and never left the confines of Chrome.

So, the original observation of The Browser Company was actually our operating systems, in 2019 then and definitely in 2024 today, are actually our web browsers for laptop people. You’re sitting in applications in a browser. Your files are now URLs, too. So, the founding inside of the company was, “Wait a minute: browsers were designed for the information highway. They were designed when the web was a publishing platform. That has changed. Browsers have not. Why is that?” Spoiler alert: money. “Can we make your quality of life on the internet better?” 

So, you are correct in that relative to the origins of the web and the origins of browsers, people are not spending as much time with content as they are with opening their browser and doing their work.

So, it’s an application environment. That’s what I’m getting at, and one of the things we talk about in Decoder all the time is how the application model moved from Windows to the web to mobile, and then maybe back to the web. There’s something happening there that seems big, and it’s kind of landed on the web. Most people who want to deploy a desktop application turn to the web first. I don’t think a lot of people are deploying Win32 first anymore. Do you see your browser as having a meaningful impact on that class of developers?

Because if you’re an operating system, you have a lot of power, right? You’re like, “Here’s some APIs. Here’s some capabilities of my operating system that a developer can use.” This is what all the major operating system vendors say to their developers all the time. You’re saying my browser is an operating system and people are deploying applications to the web. Are you in conversation with those applications? Do you offer those developers new capabilities, or is it really just about the end user?

It’s a great question, and I actually think this is where Google deserves a lot of credit. I think if there’s one thing Chrome and the Chromium team specifically has done a fantastic job of is building an operating system, or an application platform, that developers love, generally speaking, and they make it more and more powerful. In fact, you had Dylan Field on this podcast; Figma would not exist if it weren’t for Google, Chrome, and Chromium making the web fantastic for application platforms.

What we’re focused on is the individual and the person at the other end. So, what we think about is the focus on developers, and the focus on publishers as Google describes them, has left the individual on a Tuesday at 2PM lacking a lot of powerful tools to make them better and faster. So, of course, we have integrations with different third-party application developers. I would love it if we could offer stuff that makes them love Arc more.

But in fact, we think that what was missing was looking at my wife using her laptop on Tuesday at 2PM [and realizing], “Wait, that’s what she’s doing? We can do better. Computers can do more than that.” So, that’s the orientation we take to our work.

One of the big questions when you’re starting a new browser company is, one, how will you take share from Google and Microsoft in Safari, particularly in iOS? And then two, what engine are you using? 

Because you’re not going to write a new browser engine that seems like a massive undertaking. You’ve landed on Chromium. It seems like the whole industry is headed towards Chromium. Microsoft famously uses Chromium now. Was that a big decision? Was that a little decision?

To be honest, it was an intentional decision, but it was a little decision. And for better and worse, the theme of my answers in this podcast will probably be, “We come back to the individual at 2PM on a Tuesday.”

If there’s one thing they want from their rendering engine, if they’re familiar with what a rendering engine is, it’s that their web apps work. So, for example, for whatever reason, we have a lot of teachers and people in education using Arc. A lot of software at school districts are optimized and actually only work in Chrome-based browsers. It was intentional in that we wanted to make sure we could strip out a lot of the kind of tracking and nefarious parts of Chromium that at least don’t align with our values.

But once we realized we could do that, we thought, “Hey, almost every website will work on this rendering engine. We want to make your day better at 2PM. Let’s jump to that part,” to the end person-facing part of the software.

So, you’ve got Chromium as a rendering engine that’s the same as Chrome. Arc itself is the Chrome around Chromium. This is just the language. So, you built a wrapper around Chrome — that’s a pretty familiar idea. And then the idea is all of those things will make productivity, particularly productivity for knowledge workers, better on the web. But you’ve invented a lot of terminology.

There’s a sidebar, there are spaces — there’s just metaphor after metaphor in Arc that are different from Chrome, right? There’s boosts. There’s just a lot of words and concepts in this browser, which are interesting, but a lot of them are, “We have to teach people a new metaphor for using the web or thinking about this browser as an application layer in their computer as opposed to just a web browser.” Where did the genesis of this come from, and how did you go about, honestly, just picking all these names?

Yes. To be totally honest, I regret many of those words. I wish we didn’t have so many new concepts. And I think it’s too complicated of a piece of software for many people, and I think we have to make it a lot more simple. But where it came from was, “Wait a minute: if you look at someone’s tab bar and they have 50 tabs open, and they’re really teeny tiny, and there are a lot of duplicates — why? You don’t need five versions of the same Google Doc. How can we solve that problem?”

So, most of the new concepts came from the perspective of “what is broken, what is wrong with the way people use their browsers today, and can we invent a way to alleviate that problem?” And what that led to, for better or for worse, was a lot of small features and a lot of small ideas that make your day just a little bit better, save you a couple of clicks, that I think is built a very cult-like following in the software but has made it a bit too unapproachable for the average person in that it is a lot of new ideas.

That’s part of what we’re working on now: how do we strip away a lot of the experiments that didn’t work or didn’t work as strongly as we hoped they would to something a lot more focused and a lot more essential? Which, right now, is focused on how we do your busy work for you. Because people I think love our features, like when you’re playing a YouTube video and you click away, we automatically open a picture-in-picture player. Or if we notice you have tabs open that you’ve had open for a long time that you haven’t used, let’s just tuck them away neatly for you. And so we’re going to be focused on trimming down the product even more and really try and enhance the bit that does your busy work for you and has these little moments of delights.

That seems like the challenge. You have identified one set of users that already knows they’re using a web browser as a productivity platform, that already knows that all their apps are in a web browser. And then there’s another class of users that is just using Safari because it’s what came on their Mac, and you’ve got to get more of those people in order to grow your user base. How do you balance the two? It feels like you already have the power user problem.

The way that we started building this product was through the lens of problem statements and that’s how we ended up with so many different solutions and so many different words. But I think the byproduct of that is four years later, I think we have a much crisper understanding for the average laptop person — who, again, doesn’t know what a rendering engine is and honestly probably isn’t reading The Verge and isn’t an early adopter — what are the most painful, annoying, tedious parts of their day on the internet? Where if we just focused on them they’d say, “Wait, I want that.”

I do think as much as there are some things that may be power user-y, there are other ideas in there that you talk to 10 out of 10 people in this demographic, and they go, “Yeah, I have seven windows and 87 tabs, and it’s a mess, and it’s chaotic, and I feel overwhelmed.” And so we’re going to be focused on trying to build an antidote to a few specific problems.

I feel like web apps in general require people to understand new metaphors. We often write and talk about how younger people are not as aware of file systems as a concept — they grow up on iPhones and iPads and ChromeOS devices using something like Figma, which requires a bunch of people to accept a bunch of new metaphors. And then you’re trying to change the metaphor around all of those metaphors. Is that going better or worse than you expected?

Honestly, it’s going better than I expected, but I think we’re going to hit a plateau. Our ambition really is to change the way people use the internet and improve it. And if we really want to reach out of that early adopter crowd, we have to simplify. But I think one of the really exciting things is the most-used text box on a Mac is the URL bar in Safari.

And so what we’ve realized is we kind of spread out and we built all these new surfaces and all these new nouns and all these new spaces, but if we just focus in on a few points that people are familiar with and use a lot like the text box, like the URL bar, there’s a lot that a lot of power we can pack in that.

And actually that Verge article shared a lot where, no, people don’t really want to manually organize stuff in file systems anymore. They want to tell the computer what they need, and they want the computer to go get it for them. So, I think you’ll see us pack a lot of the ideas behind some of our power user features in a much more approachable and familiar interface, which is the Command+T text box that you go to all the time to ask for things, now you can ask for a lot more.

Well, you started this whole conversation by saying you were distraught that an election had been lost and computers were maybe responsible or not, and the operating system is where the leverage is. How do you turn all of that into the leverage you’re seeking? Is it, “We’re not going to show you some websites.” Is it, “We’re going to make you have a healthier relationship with Instagram?” Are you just going to pop up a warning that’s like “You’re on Instagram?” How do you actually use the leverage of owning something that feels like an operating system?

In the same way that your background as a copyright lawyer informs a lot of the work that you do, I want to take a minute just to talk a little bit about my origin story because it relates to the answer to the question. When I was a senior in college, I didn’t know what I wanted to do, and I was a sociology major, and I went to a lecture by a professor named Robert Putnam about his book Bowling Alone. After the lecture I went up to him, I said, “Professor Putnam, what should I do with my life?” He’s like, “I don’t know you, so I have no idea, but if you like my book, there’s an entrepreneur named Scott Heiferman that started a company in New York City called Meetup after he read the book. Maybe you should go work for him.”

So, I went to get a job at Meetup, and on my first day of the internship, Scott gets up, and he says, “We’re going to turn away from the banks, and we’re going to turn to each other on Kickstarter, and we’re going to turn away from big box retailers, and we’re going to turn to each other on Etsy.” And he went on and on, and it was deeply inspiring, and it was that part of me that fell in love with tech and the idealism behind it. To me, that shows two things: one, I have always been motivated by people at the other end, and two, Scott was totally wrong. 

I love him,but I think, honestly, the part of me after the election that said, “I got to fix something, we got to do something, we got to fix democracy with technology” — I’m still an optimist; I still care about people, but I think we now have right-sized what our role should be, which is instead of saying in that moment, “How do we as some tech company with 20 people fix democracy or improve our civic society?” It’s just as worthy and ambitious to say, “My sister-in-law who’s a teacher and spends hours every day copying and pasting between different software to be a teacher, let’s get rid of that busy work for her.” That is just as ambitious, and that’s just as worthy.

So, honestly, there’s kind of been this personal transformation from early ‘20s, the internet is going to fix everything to, “Hey, let’s just make our friends and our family and our lives a little better every day.” So, don’t get me wrong, I still have that part of me that is as idealistic and hopeful that the web and the ideas behind the internet can improve these top-level ideas, but we are much more interested in almost like the anthropological approach to “Nilay’s day, how do we make it a little bit better?” and find worth in that. 

There’s a little bit of tension here. You described Arc as being an operating system. You obviously want, in some end state, for application vendors to be talking to Arc as an operating system and maybe leveraging some of your capabilities. You’re talking about end users making their lives better.

But you live on another operating system; the applications inside Arc or whatever other browser are doing whatever they’re going to do. How do you balance that role? It feels like there’s only one stakeholder whose experience you can actually improve or adjust, and Apple might just make it much harder for you because you run on a Mac, or Microsoft is going to put Edge pop-ups all over Windows, or Figma is going to strike a deal with Chrome to use some cutting edge API that you don’t have access to. There’s a lot of dependencies there. How are you balancing all that?

This is where I’m just a big believer in the web. As tricky a moment as it is in many ways, I believe the web has won, is winning, and will win. And I think in the web, there are enough parties involved and there are enough incentives where it’s not really about The Browser Company — it’s about betting that the web is an application platform, and the decentralized nature of it will mean that people will still keep building for the web. 

As long as people are building applications for the web and the center of gravity — especially in this world of AI, love it or hate it, is heading even more to the web — I think there’s enough incentives in the industry, in the ecosystem, to suggest that if we build one user agent for it, there’s really good work we can do there.

I want to talk about the web in detail, but I think this brings me to the Decoder questions. This is a big ambition. How big is The Browser Company now?

Eighty people.

And how is that structured?

We have kind of functional teams — design, engineering — but we really like to organize in deeply cross-functional pods. So, we hire people that tend to be mutts, as we like to say in endearing ways. They come from different backgrounds with different skill sets beyond just whatever their title is, and then we put them together in these little pods of five people and give a prompt like, “How can we help make the experience of Shopify sellers, how do we make it easier to use their tools every day?” And we give them six weeks and say, “Go.” And they try a bunch of things, and we see what happens.

When you have a prompt like that, do you say, “Okay, you came back, you have an answer. We’re going to go find a bunch of Shopify sellers and try to market Arc to them specifically.” Or is it, “We’re going to abstract the solution to a bunch of other use cases and market the abstract product that you’ve invented”?

It depends, but actually, it’s reversed in the order we do it. So, one of our first hires was a woman named Adena [Nadler], and she runs a team now called the membership team. So, what we start with is actually conversations with Shopify sellers, and we watch them use their computers. We ask them about their problems, the things they do every day, and we actually try to abstract solutions for them based on that.

Sometimes we focus on individual tools. So, we built this feature called GitHub Live Folders that, if you’re a software engineer and someone needs a code review from you, it’ll just automatically pop up and say, “Hey, Nilay needs you to review his code.” That’s something specific for GitHub. And other times, we’ll take an idea and abstract it to something that can work everywhere.

We heard the story from a teacher last week actually, where she said she spends an hour every week taking attendance logs from a Google Sheet that she has and copy and pasting them into a school district-wide CMS of some sort for attendance records — and it takes her an hour. That makes me so mad. We can send reusable rockets to space apparently, but we have teachers spending an hour doing copy-paste, copy-paste, tab switching.

So, Nate on our team last week prototyped this mass-paste idea where in one fail swoop you can take a bunch of data from one tab and paste it in a very formatted structured way into another tab. So, there’s an example relative to GitHub where the seed of the idea was this teacher with this very specific piece of software she has to use for her very specific job, but in it is this much larger relatable idea of we can all relate to copy and pasting back and forth between tabs incessantly. So, it’s a little bit of both, but it always starts with a person. It always starts with people and always starts with going out into the world and trying to understand. Sometimes, it’s a family member. Sometimes, it’s a cousin. Sometimes, it’s a stranger. What are they experiencing on the web every day?

You’ve got kind of an interesting challenge there because mass-paste seems pretty abstract. “I’ve got two tabs, I’ve got two sources of data. I just need to move them over.” Maybe Chrome will build that feature — maybe they won’t. At least you’re competing with another browser entirely. With something like a GitHub notification, it seems likely that GitHub might build that feature and send you a notification to a mobile app or send you a notification to whatever web-based notification system that the industry will eventually adopt. How do you think about that? That your features might get adopted by the very applications that you’re trying to support?

If you talk to these application developers, one of their complaints is actually browser vendors are pretty restrictive about what they can do in the browser. So, one of our popular features is in our Command+T text box, you can type “new Notion document,” and you can hit enter, and it’ll create a new Notion document. Notion loves that. Notion can’t do that in Chrome or Safari because Google’s trying to protect its search ad revenue. So, there are examples of places where we’re actually giving developers more access than they would in other browsers because we’re not optimizing for search ads. And then there are other examples where they’re actually things that you can only do at the browser layer that exist across multiple tabs.

So, if you think about the teacher example, the things that the developer of Google Sheets and the obscure public school district CMS application would need to do to have an integration, that’s never going to happen, but at the browser layer, because we sit underneath all of it, we can actually do those things very easily. So, it obviously depends on the feature, but generally speaking, because other browsers are designed to be, essentially, big search boxes for the search ad business model, there hasn’t been as much innovation at the interface layer or the operating system level of a browser such that application developers, I think, are very excited about the access that they will be able to have, and there are things we can do across web applications that would be difficult otherwise.

You’re really describing the browser as an application layer. This is the model for apps going forward, and you’re drawing a pretty stark contrast to Google, which is “search for some stuff and we’ll show you some documents.” The web is in a moment of pretty intense tension between these ideas. You mentioned AI — all the AI applications are deployed to the web because they want to skip the app stores in one way or the other. Crypto, for better or worse, was mostly a web phenomenon because they didn’t want to pay app store taxes, either. Do you think the web is headed toward being more of an application system as opposed to a document storage system?

I’m curious, what do you think?

Well, I have a lot of feelings about the web as a publishing medium, but I think the pressures on the web as a publishing medium are not insurmountable, but unavoidable and certainly changing the economics of the business there. Whereas the pressures of app stores, on mobile phones in particular, are potentially devastating, and that’s why you see so many applications on the web. So, it feels like unless someone actively stops it, documents will move off the web and applications will move off the phone, but I’m not 100 percent sure it’s actually happening. You have a vantage point — I’m curious if you see it.

I would say unequivocally, putting aside my own feelings about it, that the web, since we started the company five years ago and the trend lines have continued, is becoming more and more of an application platform. I think that’s undeniable. I think it’s very exciting. I think it poses some problems in the context of publishing. I also think, as you mentioned, there are these words, there are these phrases [like] application platform. My wife, in her job, has things she has to do. I don’t think it is going away that sometimes she needs information, and actually, frequently she needs information.

So, I think what has changed is, as you know, the origins of the web, were a publishing platform — they’re actually closer to TikTok or Twitter in many ways than an application platform at the time. What has changed is that the mix has moved toward more applications, but the idea that as part of your job, as part of your personal life, you need to find something out or learn about something, that’s not going away. But I think the trend lines are toward it as an application platform.

Do you think that mix is shifting? If I were to start a tech website today, I probably actually wouldn’t start a website. I would almost certainly start a TikTok channel and just show people whatever I was covering. I see that as some amount of platform economics but also a lot of web economics. The desire to put new information on the web first is fading, whereas the desire to deploy applications to the web is rising, and that mix is shifting, and maybe it feels like your entire company is a response to that mix shifting, but I’m wondering if you actually see it day-to-day in how people are using the browser.

Yes, absolutely. And in fact, keep in mind, I’m 33, I grew up on the desktop web. That’s where I got lost as a child in my curiosities. And so, in fact, it’s been a process for me to admit to myself that this thing that I loved about the web and I wanted from the web that — if you look out again from a sociology, from a human perspective — we’re not seeing it as much. A thing you said that I also think is true and makes me so mad is, yes, if you are going to start, put a piece of information out, you probably should start a TikTok channel. I don’t like that, but I think that is true.

I think one of the interesting things, though, is if you go back to the origins of the web as a publishing platform, what we’ve learned about publishing platforms in retrospect is it missed two big things: distribution and discovery. We now know that the most powerful part of any publishing platform is discovery, and the web publishing platform didn’t have that built-in. Google’s a hack in many ways for that. TikTok’s a hack for that.

The second thing it didn’t have baked in is payments. Can you imagine the iOS ecosystem If Apple didn’t have native payments that were easy and seamless? Think about what that’s done for subscriptions and purchasing apps. Yes, there are a lot of challenges with 30 pecent taxes, but it enabled this thriving marketplace. And so if I look at the trajectory of the mix shift on the web toward applications, there are reasons people are rushing toward it.

And if I look at the reasons that information or publishing has faded, I think it can really come down to those two missing elements. I wish I knew what you could do about that because, again, the web is a decentralized protocol, but I think you can look at those two factors and explain a lot. I’m curious if you agree or if you have thought about that. 

Well, I agree on the diagnosis. I’m not sure what the cure is, but I asked you that question because if the browser is the operating system and you control that, well, you could be the Apple that introduces a payments layer to the web. Famously, Marc Andreessen thought the web would be powered by micropayments when he did Netscape, and it just never occurred, and then crypto arrived, and we had to listen to it.

Probably not the right idea, but the idea is cyclical. The idea that we’ll have payments on the web in some way is cyclical. And if you are controlling the browser, I’m wondering if that’s something you could introduce to fix the document-side model of it or if you’re staying focused on the application side?

I would love nothing more than to get involved with that. Because another thing we think about are the fundamental economics of browsers and the web itself, which is so dependent on ads, and I think, often, these conversations are binary “ads are bad” or “[ads are] good.” That’s not what I’m saying, but I think there’s so much more potential in the ways that browsers and publishers to the web and applications to the web could monetize if payments were built in. I think that’s extremely exciting. It’s a great example of somewhere where it’s sort of a win-win-win. If you make payments easier, the individual’s happy because it’s easier to make payments — you don’t have to pull out your credit card. The merchant’s happy because you grease the wheels — it’s easier to have transactions, and whoever’s connecting the two is making money as well. So, I find payments fascinating. I think it could do so much good for the web.

The flip side of believing in the web is we are a minnow. We’re barely a minnow, and so one of the interesting tensions we feel in this conversation — I’m sure we’ll talk about Arc Search — is we’ve got ideas we’re excited, but we’re not at Chrome scale, we’re not at Safari scale. So if we ever have the privilege of getting to a place where our voice can move the ecosystem in some way, I think adding payments natively to the browser in that layer of the stack would do wonders for the ecosystem. And I hope that we or someone else gets there because I think it would be fantastic.

How does The Browser Company make money today?

We don’t currently charge for anything, but we, as part of this kind of 2.0 product that’s coming out soon, we’re going to be charging individuals and businesses for a plan that does more of your busy work for you than the default plan. But we don’t have anything concrete to announce.

So a subscription. A subscription browser is where we’re going.

Potentially.

When you say plan, that usually means recurring revenue, not “we’re going to sell you a browser one time for $49 in a box.”

Yeah. So, the honest answer is we don’t have the specific details yet, but what we are sure of is we want an exchange of value, which is we do your busy work for you, we save you time, we save you clicks, we help you through your day, and either you or your employer pays us. Whether or not that is through a subscription model or a usage-based or some sort of token system is something we’re still figuring out, but we’re really excited about the ambition to say, “Hey, can you truly save that much time for someone that either them or their boss would fork over money for it?”

What are the pros and cons of the different choices?

A very long conversation, but I think subscription is easier in many ways. It’s more familiar. What I really like about something closer to usage-based pricing is that I really want a direct exchange of value. I want it to feel as much like the more you use it, the more you pay us because the more value we’re delivering to you. 

There’s some tricky things to think about in terms of you also want people to really develop a habit with your product because they have all this inertia from Chrome and Safari, and you don’t want to push people away from using it more and more. But I’m confident or at least hopeful that we can get around that. We’re always going to have a free plan. We hope to put as much in the free plan as possible, but it’s a tricky one.

There’s nothing in the product today that we are going to charge people for. So we’re really excited about this next evolution. How can we take the idea behind this automatic picture-in-picture player automatically cleaning up and managing your tabs for you? Can we take that to the extreme and do more and more busy work for you, such that that additional time savings, that additional work we take off your plate, that additional tedious, monotonous stuff that you have to do and you no longer have to do, you can imagine some of that being stuff that we charge for.

Also, this is a danger of doing this in person because I was not supposed to talk about this, but you loosened me up a little bit, so I’m going to get in trouble for talking about this later.

That’s why we bring people to the office. I just want to stick on it a little bit longer. So, you’ve got products today. You’ve got Arc Search and the Arc Browser. Will Arc Search be paid on the phone?

That is not currently the plan. And it’s worth noting we really think of Arc Search as the companion app to the desktop product. So, we definitely have a challenge with words and branding as a theme I’m taking from this conversation, but the intention of Arc Search: it is the mobile browser to the desktop browser.

Sure. Arc Search is an AI product. I want to talk about that a little bit, but the economics of AI products are pretty simple. Someone does a search in Arc Search. You have to go talk to a cloud provider, do some inference and come back — that costs you money. If you intend to keep it free, how much money can you spend before you have to change your mind?

So, our intention is that the paid offering — which, again, we’ll apply on mobile, too, not the Arc Search that you see today, but the additional functionality on top of it — is what will subsidize the free version for folks.

So, then the goal is you make useful free versions and people convert to the paid?

Yeah. What people do in Arc today doesn’t actually cost us all that much money, and our ambition is to make this free for as many people as possible. As we get into more AI inference-intensive tasks for people that take off more and more busy work, that’s where… I think we want to be a sustainable business that exists for a long time — it’s about time — but also I think the costs get more prohibitive.

You’re obviously competing with Google. Google loves to give things away for free. That search ad revenue is a cash machine basically. That search ad revenue is a cash machine for them. How do you think about competing against a competitor that will undercut you on price in the most ruthless way possible, which is giving it away for free?

In some sense, it’s terrifying. We have, on paper, absolutely no advantage. They have more money. They have more people. They have more all of the things. I think over time, as we’ve built more and more features and gotten this question more and more, I think what we’re realizing is if we’re truly going to build the successor to the browser, what comes after it — I’m going to avoid branding it since I’ve branded too many things — that is really a holistic rethinking of our interface to the internet. I think that, and the care and the detail that goes into that, is not as simple as popping on an AI sidebar chat onto Chrome. There are examples of other browser vendors that have clearly taken ideas from us and done their own versions of it, and it hasn’t gotten in the way of our growth or success so far.

So, I think if you look at it from a top-down perspective, how are we going to beat Google or Apple or Microsoft? It’s tricky to give you an answer that is convincing. I think the lived experience so far is that we keep our heads down, we optimize for building something that people love and truly helps them in their day-to-day, and we think about this from a blank-page perspective of not “what did browsers do yesterday?” but “how can we build a cohesive day on the internet that saves you time and does your busy work for you?”

I think it’ll be difficult for the other vendors to just bolt that onto their existing products. Now at some scale, might they do what happened to Slack with Teams? Of course, we’re in a capitalistic society — that will happen. I think there is the room for us to run if we are focused and we are fast and we really do what we’re best at, but time will tell.

There’s the Chrome of it. There’s also the Safari of it. Apple really wants people to use its integrated applications, particularly on mobile. Do you find that trying to ship a new browser on an iPhone is a lost cause? Do you think that that is a market you can actually get into, or is that just closed off to you?

I think the fascinating thing about Safari in general is that Safari — and we have this on good sources — is the most used application in the Apple ecosystem. More time is spent in Safari than any other application. But if you go look at the size of the team and the things they’re working on, there’s a mismatch there because Apple doesn’t want the center of gravity to move toward the web on desktop.

On mobile, it’s more difficult because the browser plays a different role. On desktop, it is increasingly the application environment, and on mobile, it’s a place where you go to quickly look something up, get some information really quickly, quickly read an article. And there’s some things that Apple does or doesn’t do that makes it more difficult.

They don’t let you bring keychain passwords over. It’s more difficult to check out. And so there are some structural challenges created by Apple on iPhones that make it more difficult. But I’d say the bigger thing is the role of the browser on your phone is that it’s almost a different product than what it is on desktop, and that’s the thing that we think about the most. But I think as we’ve seen with Arc Search, there is a desire if you build something truly new for people to change, and it’s just a question of what is the ceiling there on mobile versus desktop?

This brings me to the other Decoder question. You have a lot of challenges. You’ve got huge browser competitor that gives away its product for free. You’ve got operating systems that will and will not let you do certain things. You’ve got the changing nature of the browser itself. You’ve got pricing to figure out. How do you make decisions? What’s your framework?

I knew you’re going to ask this question because you always ask this question. I wish I had a framework. We think of our work as optimizing for feelings and instinct. I don’t know if this is a response to the technology industry that I was brought up in where you’re supposed to be neutral and unopinionated and have frameworks, but our approach is: What are we trying to express here? What feels right to us? What do we want to do for ourselves and our parents and our siblings and people that we care deeply about?

So, generally, of course, we have a data science team. We look at the data, we reason in all the ways that we should, but I think at the end of the day, [you have a] big decision to make, I’d say it’s more of a personal expression and a personal reflection of our hopes, wishes, and desires for our work than it is anything else.

One of the comparisons you made was to Google. You said it’s not just as easy as bolting on an AI chat box to the side of the browser. I could be pretty reductive, and I could say, “You’ve just described Google shipping its org chart. There’s a Chrome product manager. There’s a Gemini product manager. Just be next to each other. Don’t integrate the product.”

That sounds like you’re betting on Google not figuring it out, to some extent. The Google product culture will ship and kill things in the way the Google product culture does, and it will never make the turn toward integrating the AI products. You can feel however you want about that bet. I’m sure the people at Google feel some way about that bet, but is that what you’re thinking, that they’re big and slow and you can actually just be more nimble?

It’s worth noting I think the people at Google are very smart, and I’m not just saying that as what I’m supposed to say. I truly believe that. We hired Darin Fisher, who started Chrome and ran Chrome for 16 years. He worked at The Browser Company. It’s more about the incentive structure. I like to think a lot about incentives. It’s one of the things I wish I thought about more earlier in my career.

There’s a story that Darin told me that really stuck with me, which is Chrome had this idea that, when you go to the “new tab” page (one of the most popular surfaces in any piece of software you use), if they show you an icon for the webpage that you go to a lot, you might be able to notice it much more quickly — “Oh, it’s the Twitter icon. I’ll click on Twitter — versus just a screenshot of the webpage. And they ship that, and overnight, Google search ad revenue dropped by 5 percent, and they weren’t sure why. It was this big freak out. Now, that resolved in the way that it did, but that is the sort of thing that you have to contend with if you—

Because people were no longer doing navigational searches for Twitter?

Yeah, because they don’t want you to go to Twitter; they want you to go to search. Now, the Chrome team doesn’t — the Chrome team wants you to get to Twitter as fast as you can, but at a company like Google, in this moment, in the public markets, in this moment of AI even more, there are these incentives with the search ad model and the way that Chrome and the search ecosystem works so far that are just a huge… it’s inertia.

So, it’s not just shipping the org chart; having worked at Facebook, there are real challenges there. But I think on top of that, there is the incentive structure of how the company makes money and has for a long time. And then there’s also the risk. If you think about it, if we start with a blank page, if you give me the most generous reading of everything I said, it may not work, and if it does, we don’t only need it to work for a 100 million people.

If we do something radically different and we find a hundred million people that love what we do, that is a raging success. For Google that’s an utter failure, and that’s if it goes right. So, I think there’s also the risk aversion to the scale they need to hit the number of people it needs to work for to be worthy, putting aside all of the product risk that comes with doing something truly new.

Google’s in a state of what I would call regulatory scrutiny. They just lost the antitrust case against the United States Department of Justice that said there was an illegal monopoly in search and in certain part of its ad business. The ad tech part of its business is going to an antitrust trial very shortly here. As part of the search trial, we found out that Google’s paying Apple $20 billion a year to make Google the default search engine. This stuff feels like it’s coming apart.

It’s a big moment.

There are opportunities here. Which of those opportunities is most right for The Browser Company, and how are you going to attack them?

Candidly, the way I think about it is there’s more pressure on them not to do anticompetitive practices or things that can be perceived that way. So, I think there are a lot of subtle things that these players do that make it harder for an upstart like us to compete. So, I would say it’s less a specific decision, though these are all big in their own right, and more generally that there are eyes on these companies not to do things that are monopolistic or perceived to be monopolistic, and that culture and climate, I think, is advantageous to people like us.

Do you think the Department of Justice should break up Google?

Yes.

How would you break up Google?

Come on, Nilay. You’re a lawyer. That is way above…

I’m just asking.

[Laughs] That is way above…

Well, there’s an obvious answer here, which is split out Chrome, which has been floated. Do you think you would have a better chance against the independent Chrome company?

I’m not a lawyer. I have no idea. But what I—

I’m asking you competitively. If Chrome did not have the pressure of Google search — you can put in the Twitter icon or whatever application icon without hurting the search revenue — do you think you’d have a better shot at competing with an independent Chrome?

Honestly, hard to say. I’m not trying to be evasive. I honestly don’t know.

Do you think that the deals Google has been making to make its search engine the default in different places, if they came to you and said, “We’ll pay you $20 billion a year to set Google search as the default in Arc,” would you take the money?

$20 billion was an unfair number to pick.

$5. We’re just going to keep going by fives. $10. Would you say yes to $10?

No.

$15?

Maybe this comes back. Maybe I should—

$20, Josh.

Maybe I should have a framework for optimizing for this stuff, but at the end of the day, I just want my day on the internet.

I’ll go to $100.

Just at the end of the day, Nilay, I want my quality of life on the internet to be much, much better.

Do you take money to set a default in search on Arc?

No.

Is there a default?

The default currently is Google.

You got to make a phone call, man. The money’s on the table.

That may or may not change soon.

The default, or the money?

The default.

Okay.

No, we are not going to... If we take money for the default search engine, then ultimately our customers, our search engines and advertisers, and that is conflicting to why we started the company, what we set out to do.

However, I do think one of the things that is very exciting about this moment in AI, alongside all the challenging things, is AI has this ability to route us to different places more intelligently and take us more directly to places we want to go that are not always Google, and oftentimes, it’s never Google. So, we’re going to replace the default search engine, but not with another search engine that’s...

One example I like to think of is I just moved to a new place in Brooklyn, and I was trying to decide if we should buy a HomePod. Valerie and I love to dance around the house and we didn’t have a speaker. I want to type in “The Verge HomePod review.” If I hit enter, that takes me to Google. In our 2.0 product, if you hit enter, that’ll just take me to The Verge’s HomePod review.

So, there are things that we can do in this moment that weren’t possible before that I think make Google vulnerable both in search and browsers. That means this question of default search engine is no longer just going to be Google vs. Bing and who’s going to pay you. It can be, “Let’s take you to the exact right place based on what you’re looking for.”

So, you’re building a search-like functionality. 

Again, it may sound tired, but the way we think about this is what are the things you need to do every day? There are these new technologies that make it more possible to blur the lines between what is a browser, a search engine, into something that more holistically end-to-end helps someone do something.

And yes, as part of that, when you type in the most popular text box on your computer, we can now take you and route you to lots of different places that oftentimes are much more direct and on the nose for what you want and don’t just funnel you into the Google ecosystem because that’s how it’s always worked, because that’s what their business model is.

One of the things we’ve seen a lot with AI in general, and you’re certainly talking about it now, is the idea that that text box, Command+T, is actually the user interface of your computer. You’re going to just tell the computer what you want, and the computer is going to go off and do it. And if you have the entire web behind you, you can do a lot of things, especially if you can take actions on web applications.

Yes.

Are you trying to build that kind of automation layer where you say, “Hey, just go to my calendar and bring all the dates out and put them over here?”

Yes. Again, you’re getting me in a mode where I’m sharing more than I should. But we have this internal prototype I tried last week where my son had his first day of preschool today. They sent us a PDF, which I opened in my browser with all the different dates for holidays and whatnot, and I could say, in one gesture, add all of these to my calendar, and it would do that.

And so what we’re doing is building the layer underneath all the applications to understand what is going on in your life, what are you looking at right now, what have you been working on previously, and the connective tissue between all of the applications and tabs that you use and rely on, and on top of that, we can take a lot of busy work like that off of your plate much more easily.

And sometimes, that’ll come through Command+T, and I’ll ask it. And other times, if I’m on Apple looking at a HomePod, we might say, “Hey, you really like The Verge. You read The Verge a lot. Here’s the HomePod review.” So, I’m using the text box as, yes, the most popular interface, but I think it should feel like your entire experience on the web is more personalized and more proactive to you, not just when you explicitly ask for something.

This idea that a robot’s going to go click around the web for you is very popular. We’ve seen a number of startups say they can do it. I don’t think they’re actually doing it, but they say they’re going to take AI and do it. Then, there’s just a set of follow-on problems to this. 

The browser has to see everything in all of the websites. It has to see my data, it has to read that data, it has to interpret it presumably using an AI system in a cloud somewhere. It has to click on things for me without getting anything wrong, and then it has to not hallucinate. That’s a lot of steps. How do you protect people’s data and actually hit the level of, essentially, 100 percent reliability that people are going to demand from products like this?

The first thing is we really think about right-sizing AI. There’s a lot of discourse about AI right now, and it tends to be of the martini-sipping version where we’re going to replace teachers and doctors and there’s going to be the superintelligence being, and that’s, in our opinion, not the right way to think about this stuff.

I think the equivalent there as it relates to clicking is you’re going to tell the computer what you want to do and it’s just going to do a bajillion things for you with 100 percent accuracy. Today, that’s not possible. That’s not how it’s going to work. But what is possible is in these small ways, again, saying “add these to my calendar,”  we can do that, and we can do that with close to 100 percent reliability. 

Our approach is — as much as possible, which is increasingly very possible, especially on high-end MacBooks — doing that on-device. Data does not leave your device — it’s all done locally and, when it can’t be done locally, making sure that the person says, “Hey, I’m okay with that tradeoff of sending the contents of this PDF to an LLM provider in order to add it to my calendar” and let them make that decision. 

But I think the large point here is what we are not saying is the robots are going to do all of your work for you. That is not our belief, but what it can do is it can save people from a lot of the mundanity that relates to futzing around with boxes on the internet all day.

Do you think that that is a separate set of use cases from what Arc Search is doing?

Absolutely. In fact, Arc Search was really a first prototype. There’s so many things that I wish we’d done differently and we’ve now since learned, but really, that was the first experiment of this larger idea of us playing with this new Play-Doh, which is, “Okay, we can click on things for you. We can read things for you. Wow. We definitely can’t... the writing’s really bad. Oh, but interestingly, we can transform one type of data format into another type of data format.” Just feeling out the edges of what it can do today.

As part of that, one small thing that you do is you want to find out a quick answer to... I got a skirt steak the other day, and the guy at the butcher was like, “You should make chimichurri sauce.” I don’t know how to make chimichurri sauce, and sometimes I want to know that. A lot more frequently, there’s something for my job or my livelihood where I have to go click a bunch of buttons in the same order every single time. I think we’re much more excited about doing that sort of busy work for you because, candidly, that’s what people complain about the most when we interview them about their jobs.

“I want to make chimichurri sauce” is a great example because what Arc Search will do is it’ll go read a bunch of webpages, it’ll summarize them, it’ll show you the answer with some links. That is a very controversial move across the web right now. When I say there’s a lot of pressure on the web as a document or consumption medium, that’s the pressure.

In particular, a bunch of AI companies are scraping the hell out of the web, remixing the web, and the people who actually made the information are getting nothing for it. Arc Search is right in the middle of that. That is the thing you are doing. Do you think that that is a sustainable thing to do?

No. And I think this is a really complicated one, so I want to try to share both sides, and let’s take it head on. That’s part of the reason I’m here. From the perspective of an individual, I want the chimichurri recipe, I show up to the website, I got 17 trackers tracking me all of a sudden. I get a newsletter pop-up saying, “Do you want to subscribe to our newsletter?” I wade through five paragraphs about the author’s grandmother and the history of her chimichurri recipe, and all the way at the bottom is the recipe. 

That doesn’t feel good to the individual. It feels like we can do better, and it feels like for pretty much everyone that uses the web, a much better thing would be, “I want to know the ingredients and the recipe steps. Get it to me as quickly as possible.” And on the other side, it breaks the model of the web historically. 

Now I think we are not going around any paywalls. We are not training our own models. A lot of the stuff that I think is more problematic is not anything that we do, but I do think it’s fair to say that those trackers, as much as I feel like they’re unfair to me as an individual, are part of how that recipe site makes money. The fact that they show ads — which, if we are reading the sites on your behalf, you’re not seeing — it breaks that model in some way.

So, this is a moment where I’m an optimist. I think it’s a very exciting moment for publishers and media companies because for the first time… so much of this is dictated by Google and the way that Chrome and Google Search has worked for so long. So, I think something’s got to change. I think publishers have to get paid. I wish I had an easy answer for you, but I definitely don’t think it’s sustainable. Even if I also think for the individual, we got to do better as well.

In February, my friend Casey Newton wrote about Arc Search. He said he felt a rare emotion and “a kind of revulsion at the app’s mere existence and what it portends because it’s taking the value from the people who write the recipe website.” And I could do a full hour on why there’s a story at the top of every recipe website. That is the way that the money is made.

It’s the incentives of the system, absolutely.

You can’t sell the recipes for a variety of reasons, so you got to sell something else. You can sell ad inventory around the recipes. Do you understand why Casey felt the revulsion? I know he talked to you for that piece.

Yes.

And he talked to you, and the quote is, “Miller had not put much thought into the second order implications of a world where search queries no longer result in outbound clicks.” That was February. It’s September. Have you thought about it since?

Yes. Actually as recently as last week, I had a conversation with David [Pierce] at The Verge. I thought we were doing a good job of citations. He read me the riot act on the fact that we weren’t, and in the app today, we have citations even more prominently than I thought was the most prominent app out there that shows what we read, put them at the top, you can click them easily.

We’re also having a bunch of conversations with media companies right now. At the end of the day, I think media companies need to get paid and publishers need to get paid. And I think the truth is, as you know, the scale of that will not mean that it works for everybody, but we are trying our best behind the scenes and out front to be better here.

Candidly, one of the challenges we have is we don’t have the scale of other players in the space. So, if we show up at a media company’s website and say, “Hey, let’s figure something out here. Let’s figure out how we can pay you,” we don’t always get the same receptivity as what I assume other companies do. But I’m curious what you think about the OpenAI model for this, because we’re kind of seeing this all from afar. But I think what I come back to is I’ve been on the board of Patreon for five years, and I think you know better than anyone I don’t think the old model was working for anyone, even before all this AI stuff.

I think you make a great point that AI accelerates it and it hurts it, but I think the old model wasn’t working. What I do think this new technology provides is a way for all of us to rethink everything from the products themselves — the media products, the software products — all the way to the business models. And I’m curious, for Vox, how you’ve thought about that and how you think about it in the context of OpenAI and these publishers that are doing that.

Happily, my role in the newsroom is to spend money. I don’t make any money. It’s a real problem for this whole company. We’ve had Nick Thompson talk about his deal from The Atlantic on the show. His view is we need to get this money, and OpenAI is offering us a bunch of stuff in exchange for this money, including tokens and credits to use their systems to build new products.

What I see, and maybe it’ll work out, but what I see is we are absolutely hastening the demise of the web as a publishing platform because we’re making it easier and easier and easier to extract value without any payment or compensation going in the other direction. And eventually, all those people are just going to say, “Well, at least there’s a creator fund on TikTok. At least there’s YouTube payments. At least there’s other platforms with some built-in way to compensate me for my work.”

Whereas on the web, everyone just takes everything away. Big publishers left and right are saying, “Well, at least Apple News exists. We’ll just take that money.” I don’t know if that’s good or bad. But the theme of this conversation is the web is increasingly an application platform. We can tailor the browser to it being an application platform.

And over here, the part where people browse the web for information, maybe we can extract value from that and that will go away. Or maybe it’ll just be a handful of preferred providers that OpenAI pays or Perplexity pays or you pay. But that open web, the part where there’s just information on the web for people to click around and look at, that seems like there’s nothing here that indicates it can make a resurgence.

The other thing, too, is we talk about the web or publishing like it’s one big category. But for example, if you go to a local restaurant in my new neighborhood and they have a reservation booking tool, I’m sure they’re totally fine with the idea that an AI system might come around and make a reservation more easily for people. So, that one’s easy.

I really believe what you or Ezra Klein said on that podcast about this idea of a flight to quality. I’ve never listened or engaged with The Verge more, and I predict that across mediums — TikTok, podcasts — I think that will only continue. And my hunch is that things like “browse for me” or OpenAI or Perplexity, that’s not going to replace the HomePod review that I rely on before making a purchase.

I’m very bullish on that. I’m curious if you’re not, but I am very bullish on that. There is this middle tier of content and content providers that we might call quick facts or more commodity type content, where, candidly as you know, most of those or some large percentage of those are content farms, or they’re contractors that are just churning stuff out or copying stuff or AI-generated. I think it’s that middle layer, that middle layer of, “I want to know what Sauvignon Blanc tastes like because I don’t know anything about wine, but I’m at the wine store.” That, to me, is the tricky one. I think The Verge is good and going to be better. I truly, truly believe that. 

So, I think, in many ways, Casey’s revulsion comment, obviously that hurts and it hits, especially after speaking with him. I think it is fair in many ways, but I think it really hits on one percentage of the content. I’m optimistic for what will happen to the media at that end of the spectrum, but maybe that’s ignorance. But again, I’m curious from The Verge, my assumption is this Decoder podcast, I would bet that the ad slots are sold out for the rest of the year.

I think so.

That’s great.

But I look at the platforms, and I have the extraordinary privilege of getting to say that I’m a precious journalist and I have no idea what’s happening with the ads and I won’t read them and we still get to sit in a fancy studio because I have a whole company, and the economics of social platforms are not great for that. 

You have individual creators who cannot support a giant company, who are in bed with the companies they cover. I’m not even naming names — just broadly, they do the brand deals, they read the ads, they mix the commerce and the content in a way that journalists do not do or should not do. And I say, “Well, the web supported the other model for a minute, and now maybe the flight to quality is a bunch of paywalls.” And what we’re going to be left with is a bunch of free content on platforms that is corrupted in some way by the commercialization of the work because the rates aren’t high enough.

And somewhere in there is, “Well, we’re just going to let it happen because the web is an application platform and not a document platform, and we never figured out how to actually sustainably distribute this information in a way that works for everyone.” It feels like there’s a lot of opportunity to make the web a better application platform, but it feels like if you turn that all the way, you do end up with a bunch of weird ads on TikTok and a bunch of paywalls on the web.

Again, maybe I’m just too much of an optimist, but I think that it’s going to take creativity and dreaming on both sides. I think from a media standpoint, tell me if you think this is wrong, I think a lot of media organizations made the mistake, maybe a decade ago, of trusting the platforms and, in many ways, outsourcing their product development.

I don’t think media companies are going to make that mistake again. And I think there are many, like The Verge, that are innovating on what their product is, and they’re innovating on what their product is in a moment where there’s actually leverage to go after these... I can’t overstate, not to you, but to your audience, how stuck the web has been. And all of these things have been for decades because Google controlled it all. For the first time in decades, there is this technology, this Play-Doh, that gives a window to mess that up at a moment where you all—

And you think that technology is AI, to be clear.

Yes. In a moment where you all have been burned as media companies by outsourcing your product to Facebook saying, “Hey, trust us. Just give us your content. We’ll pay you. It’ll be great.” You’re not going to make that mistake again. You have Play-Doh to play with. You are innovating on product. And I think on our side, I knew, coming on this podcast, you were going to ask these questions, and I knew I wasn’t going to have a perfect answer, but I think this is important for the same reason I think it’s important to pick up David’s call, hear him kindly yell at me, and make changes based on it. And it’s why we show up at media companies offices saying, “Hey, let’s collaborate on something here. Let’s figure out a way where we pay you.” 

It’s going to take experimentation. It’s going to take collaboration on both sides. And I think that collaboration bit is the hardest bit because there are bits of what Casey said that I found deeply unfair, and there are bits of it that I found fair, but I know where he’s coming from because it’s the same part of me that was burned as a 20-year-old by these promises of “tech’s going to change everything.”

We have the moment in history, which we should not take for granted. We have the Play-Doh, we have the lessons from the past, and now we just got to dream a bit and come together in some way. And maybe this is the part of me that makes decisions through feelings, and this is naive, but I truly, truly think something good is going to come out of this, but I think we’re going to mess some things up. Everyone’s going to mess some things up, and we got to be open about it and talk about it. 

And I think there is this generation of entrepreneurs both in the media space and in the product space or the technology space that has seen, again, the models that came before it and what went wrong there and is encouraged to come on a podcast like this, even if it’s not always going to be effortless.

That is a good and optimistic place to end it, so I’m going to ask one more question.

Okay, great.

The idea that the web will come into balance and the web will endure, I want to believe. I am a web person at heart. I continue to run a website in 2024. That is just a personal decision that I’ve made. What is the chance that the web actually turns all the way into an application platform, that that dominates the next generation of the web?

Oh, I think very low.

Yeah.

I think very low. And I will need media training from you after this. As someone that is full of ideas and prototypes and we have an experimental culture, there’s nothing I want to do more than blurt out all of these ideas for what might turn it back. I think I need to learn my lesson of the folks that came before me and say, “I don’t know the answer yet.”

It is hard to imagine looking at the state of things today as we’ve spoken about, but I think there is some innovation on the product side, both from the media side and the technology side, that can turn those tides. Because I think, again, from the Patreon perspective, everyone is burned. Everyone is overwhelmed. They are burnt out. It is just not sustainable. And I think out of that will come a generative creativity that can bring it back. And I think the truth of these other historical platforms is they have these taxes, and they have these anticompetitive behaviors, they have these things that I think will work against them, and the web has a lot going for it.

So, if it’s okay, can I ask you one question?

Sure.

A birdie told me that of your Vergecast hosts, David Pierce uses Arc, Alex Cranz uses Arc. Nilay Patel does not use Arc. Why don’t you use Arc, and what can we do better?

I started using Arc in preparation for this episode. I just got to use it more. I think, unlike my Vergecast cohosts, I am reticent to actually depend on software. I think there’s a danger in being dependent on software or a workflow, and maybe that’s because I’ve had a lot of software in my life go away. So, I’m a very manual brute force kind of person. And the idea that I’ll give up some part of my workflow or my process to a tool has always scared me, but I’ll keep trying.

Which browser do you use?

Obviously, I use Chrome and Safari, and now I’m using Arc.

Oh, you can’t use Chrome. We’re having the conversation about the future of the web, and you’re still on Chrome? Come on.

We are a Google Docs company. We are a Riverside company.

Okay.

I’ll do my best.

Well, more than that. I hope out of this, I hope there is some sort of collaboration we can do. Jim Bankoff, if you’re listening, let’s do something. It’s going to be great.

I promise you that’s the other side of the house. I’ll make the introduction for you.

Okay, awesome. Thanks for having me, Nilay.

Thanks for coming on, Josh. This was great.