Mozilla Head of Firefox on The Future of Agentic Browsers and Fighting for the Open Internet Against Google Chrome, Apple Safari & Microsoft Edge | Ajit Varma | E300

For episode 300 of The Product Podcast, Carlos Gonzalez de Villaumbrosia sits down with Ajit Varma, Head of Firefox at Mozilla, the nonprofit behind the original challenger browser that pioneered browser tabs, pop-up blockers, and browser extensions. With 210 million active users and $826 million in annual revenue, Firefox is the only major independent, open-source browser still standing against Google Chrome's 68% share, Apple Safari's 17%, and a new wave of agentic browsers. 

Before Mozilla, Ajit spent six years at Meta leading monetization of WhatsApp and overseeing its business messaging platform. He has also held product roles at Google, Uber, and Square.

What you'll learn:

• Why LLMs are making browsers more strategically important, and what that means for product teams building in an agentic world
• Why "trust us" is no longer enough, and how open source changes the standard for privacy in AI products
• - How to compete against trillion-dollar incumbents without abandoning your mission

Key takeaways:

• Privacy claims without open-source inspectability are unverifiable, "trust us" is no longer a sufficient product strategy in the AI era
• Competing against trillion-dollar companies is possible when mission clarity defines what you refuse to optimize for
• The agent-driven internet will either democratize access or concentrate it, product choices made today will determine which

Social Links:

• Find out more about Product School here
• Follow our Podcast on TikTok here
• Follow Product School on LinkedIn here

Transcript

Introduction

Ajit Varma | Mozilla 00:00:00
The internet doesn't need a middleman. It can actually go around that. I'm personally worried that the open internet doesn't exist and everything is an app controlled by gatekeepers, and these gatekeepers preserve functionality and better capabilities only to apps. By being closed source, there's lots of stuff that happens that you don't actually know about, and a lot of that has emerged in various forums by whistleblowers. But what about all the things you don't know that could be happening? Competition is good. I'm much more worried about the opposite — what if there's only a single base that everything is built on? You're giving that margin back to the driver, giving that margin back to the restaurant, and those are the people who I believe are creating the most value that we want to preserve.

Carlos González de Villaumbrosia | Product School 00:01:00
Hey, this is Carlos, CEO at Product School and your host on The Product Podcast. Today's guest is Ajit Varma, Head of Firefox at Mozilla. Mozilla is the nonprofit behind Firefox, the original challenger browser that pioneered browser tabs, pop-up blockers, and browser extensions. Firefox has 210 million active users and $826 million in annual revenue, making it the only major independent, open-source browser still standing against Google Chrome's 68% share, Apple Safari's 17%, and the new wave of agentic browsers. Before joining Mozilla, Ajit spent six years at Meta leading monetization of WhatsApp and overseeing its business messaging platform. He has also held product leadership roles at Google, Uber, and Square. In our conversation, we cover three things: why LLMs are making browsers more strategically important and what it means for product teams building in an agentic world; why "trust us" is no longer enough and how open source changes the standard for privacy in AI products; and how to compete against trillion-dollar incumbents without abandoning your mission. Let's get into it.

Carlos González de Villaumbrosia | Product School 00:02:00
Welcome to The Product Podcast, Ajit.

Ajit Varma | Mozilla 00:02:00
Hi, it's good to be here. Thanks for having me.

Firefox's Origin Story: From Netscape to Web 2.0

Carlos González de Villaumbrosia | Product School 00:02:00
This is going to be a good one. I'm excited because Firefox — I started using that browser many, many years ago, and I feel like this is the story of the underdog. You were fighting with Internet Explorer back in the day, then Google Chrome, now all these AI browsers. The first thing that came to mind is: how have you been able to stay alive and still compete when you have way less resources than some of your competitors?

Ajit Varma | Mozilla 00:02:00
It's interesting that you mention that because the background for Firefox goes even further than Firefox itself. It was actually originally born out of Netscape, one of the original browsers out there, and it was really impacted by various uncompetitive things that operating systems did. In that case, it was Internet Explorer and Microsoft. But the resurgence and the recasting of Firefox was building it in a very different and open way — centered on users and centered on a very powerful value proposition: building something that is built for humanity to protect and preserve an open internet.

Over time, the power of Firefox was that it actually led to innovation again in the browser space. If you look at that period of history, Internet Explorer actually removed most engineers from working on it. Nobody was there. The reason for that was because Microsoft wasn't really an internet company. It was making most of its money on applications, on the desktop, and so it didn't want to create a rich app, web app ecosystem — that's why it didn't invest. Then with Firefox's focus, you see this emergence of Web 2.0, which led to the creation of things like Gmail, Google Maps, Docs, and even Facebook. All these things were possible because of the capabilities of rich apps.

Despite all the ups and downs, competition is probably the most important part of making sure that there aren't gatekeepers that prevent access from an open internet.

More recently, something really fascinating is happening — more people are saying, "Okay, browsers actually matter a lot." There's been this lull where browsers seemed commoditized, and now it's like browsers are interesting again. The reason is that browsers have so much power. People spend so much time in them. They have so much data. They do so many actions on behalf of a user. A lot of these emergent AI browsers are thinking, "That would be great to bring in these new paradigms," in the same way that mobile apps were created once great phones were created. It's a very exciting point, but the thing we've always maintained throughout all these different cycles is really our focus on building a human-first browser that is built for the best experience for an open internet.

Our structure of being owned by a nonprofit foundation means that we don't have to chase the next highest profit lever. We can actually build something that stays true to our values, and that just seems to resonate with people.

Mozilla's Hybrid Structure: Nonprofit Foundation, For-Profit Corporation

Carlos González de Villaumbrosia | Product School 00:05:00
I want to unpack some of the concepts you just mentioned. For example, the way Mozilla is structured — it seems like it's a hybrid organization where the core is a nonprofit, but you also have a for-profit arm. How does that work?

Ajit Varma | Mozilla 00:06:00
I joke that it's kind of like OpenAI with its for-profit, nonprofit — but obviously all the lawsuits right now say we're doing it for very different motivations than they are. The foundation is an organization that has several things under its umbrella. You can look at things like Thunderbird, which is an email client, but there's also a lot of societal work it does — web standards, various other things. But there are constraints that come with being a nonprofit in terms of how it's able to make money, how it can spend money, what percentage of money needs to be allocated.

The way that MoCo — Mozilla Corporation — works is we are a for-profit that is completely owned by the nonprofit. That means we still have the same values, but the for-profit aspect allows us to make revenue, spend revenue in a different way, and it unlocks some of the constraints that would otherwise exist for a nonprofit. This structure helps us simultaneously serve the needs and mission of our overall umbrella organization, but also compete more effectively in what is a very competitive space. We are, I think, maybe the only independent company going after every single trillion-dollar software company as our competitors. It's pretty crazy.

Open Source as a Trust Mechanism: Inspectability, Community, and the Anthropic Partnership

Carlos González de Villaumbrosia | Product School 00:07:00
It's insane. Another unique aspect is that Firefox is fully open source. I'm curious to know how that influences the way you build — when you're not just working with your own engineering team, but also tapping into a community of developers that are contributing.

Ajit Varma | Mozilla 00:07:00
There are two parts here. One is that by being open source, anyone can see the code. The inspectability is this thing where you need to be able to build trust, and in order to build trust, you have to be able to prove what you believe to be happening to actually be happening.

When there are things that are so important — like accessing every part of the internet or accessing things like messaging — by being closed source, there's lots of stuff that happens that you don't actually know about. A lot of that has emerged in various forums by whistleblowers. But what about all the things you don't know that could be happening? Open source plays an important role in that.

The other thing is that we very much believe we are building a product with our community. That means we want to hear feedback on early features. Being open source helps people know what we're building, what we're planning, what the future holds. We get feedback all the time on what people like and don't like, and we adjust our plans based on that feedback.

Going into community involvement — this is also something that's really exciting to me. It's not owned by any individual person. If somebody wanted to take our entire codebase and make a fork and create another version of Firefox, that's possible. You see that happen with Zen Browser and Tor Browser — they're built on top of Firefox and Gecko, and they're great flavors of Firefox with different value propositions. People can also contribute directly into Firefox core — whether that's making Gecko better, building extensions.

This has led to some really positive benefits for us in the last few months. The one we've talked about most recently is with Anthropic. They've actually turned a lot of their AI to make our browser more impenetrable and close any bugs that might be there. The reason that's possible is because they can actually take our entire codebase and use it themselves to build better models.

The way this whole thing came about is that an individual engineer at Anthropic started contributing to the Firefox codebase. There wasn't a corporate discussion. There wasn't a "Let's partner." It was one person who was passionate about an open internet and started contributing, and then we amplified that relationship. There are a lot of those examples — whether it's an intern, a college student, or a big corporation — and having this open source code is just going to benefit us and benefit the community more and more.

Competition Is Good: Why a Chromium Monopoly Is the Real Risk

Carlos González de Villaumbrosia | Product School 00:10:00
I definitely see a ton of benefits to being open source. One of the trade-offs — and it happened to Mozilla — is when some of the Google engineers started contributing to the browser, it started as a friendly partnership, and then eventually they ended up building Chrome. I'm curious about how you're thinking about this when you're now partnering with engineers who work at companies such as Anthropic, in terms of them potentially becoming more of a competitor than a partner.

Ajit Varma | Mozilla 00:11:00
The interesting thing about why we exist is that we really exist to create a more open and better internet. That's the reason we were formed as a foundation and a company. And for that, competition is good. I'm much more worried about the opposite — what if there's only a single base that everything is built on? What if it's only Chromium and everything is built on top of that? The risk there is that companies start saying, "Okay, we need to get an extra bit of margin," and users are slowly boiling the frog — it doesn't jump out because it's just slowly changing.

I'm personally worried about a world where the open internet doesn't exist and everything is an app, and an app is controlled by gatekeepers, and these gatekeepers preserve functionality and better capabilities only to apps — whether that's payments or any number of things. They're charging a 15%, 20%, 30% fee. But even more: what if ChatGPT could only exist as an app? It would have an inherent disadvantage when all that functionality is built into the operating system. Whereas if you look at ChatGPT, the way it started was as a website, and the reason for that was because it's open to anyone in the world. You saw it become the fastest-growing application ever because anyone could access it.

If we have many better browsers that push an open internet forward and always preserve an open internet, I'd actually say that's a good thing.

How Firefox Makes Money: The Double Bottom Line

Carlos González de Villaumbrosia | Product School 00:12:00
I also want to talk about the business model, as there is a for-profit angle. How do you actually sustain your operations?

Ajit Varma | Mozilla 00:13:00
The way we monetize is a few different ways. One is if someone chooses to use Google as a search engine — in our settings, we have rev share with them or other search engines that we show. This is all optional. You can choose to use those search engines, or you can use any search engine. We don't force lock-in, but we make revenue there.

We also have revenue on certain surfaces within Firefox. If you open up a new tab, there are some sponsored stories and tiles that show up that we make money from. But again, all those things can be turned off. If a user doesn't like it, they can turn those off.

We're also working on subscription products. We have a subscription VPN product where we're providing clear additional value that costs us money, so we're charging for it. But we also try to give free options as well — we have a 50 gigabyte free VPN.

We're definitely trying to balance, but we're continuously optimizing for the best experience, and we need to make revenue in order to build a browser because browsers are very hard and complex to build. You can see this where OpenAI stopped development on its browser, and other emergent browsers have stopped as well. This funding is important for us to complete our mission, but it's not the only thing we care about. What we talk about internally is a double bottom line goal — revenue is important to fulfill our mission, but the mission is also important to fulfill for the sake of the mission itself.

Carlos González de Villaumbrosia | Product School 00:14:00
The thing that came to mind as I heard you talk about the free tier is that this sounds very different from the free tier of a product like Spotify — where if it's free, you're basically paying because you're served with ads you can't turn off, or your data is being sold to third-party providers you don't even know about. So it seems like when you call something free, it's truly free for the user.

Ajit Varma | Mozilla 00:15:00
It is free. We don't monitor any data to sell or anything like that. That's why we focus a lot on choice — we know a lot of users benefit from Google, and making that search available is beneficial. It's a way to create a good experience while also monetizing to fund all the development we do. That's also why we allow people to turn it off. And again, you can take the codebase and actually edit the code yourself with many of the AI editors and make your own browser. I'd love for people to engage with our ads, but if it's not for you, there are options with Firefox.

AI Browsers and Platform Shifts: Smart Windows, Agents, and Opt-In AI

Carlos González de Villaumbrosia | Product School 00:15:00
Let's get into the AI browser landscape. As you mentioned, a lot of these browsers didn't exist — and even some of the ones launched recently don't exist anymore. We also saw Arc recently being sold to Atlassian for around $610 million, and Perplexity Comet entering the space. It seems like there's a whole new competitive landscape. How are you playing in this map now?

Ajit Varma | Mozilla 00:16:00
When there are platform shifts, they're often very disruptive. And it's not disruptive in a bad way — it just opens up new expectations and capabilities for users who then want new ways to do things.

Looking at the internet over the last 25 years, there have been major platform shifts. One big one was mobile, about 15 years ago. When mobile emerged, there were a lot of changes in how people interact with internet applications. Instead of going to websites, people started going to applications. Firefox invested a lot in that in a pretty open way. There was Firefox OS, which turned into KaiOS — basically a way to give an operating system to run web apps so that everybody in the world could have an application. A lot of emerging markets adopted KaiOS, and it was very successful for enabling hundreds of millions of people to get on the internet for the first time on their phone. But it very much changed things — people didn't go to a URL anymore; they opened up an application.

With AI, there are a lot of potential paths the future could hold. One is that people still go to a browser, and this is what we're actually seeing. Some AI people say, "No one's going to go to a website anymore." That's definitely not the case. People have a desire — they want to type in a URL, go to a website, look at options, click on things, and make decisions for themselves.

But there's also a lot of talk about agents, whether it's OpenClaw or many other agents being built within a browser. In an agentic world, people consume the internet in a very different way. Instead of saying, "I'm going to go to a URL," you say, "I want to take this action" — like sending an email — and then it knows you have a Gmail account, who the recipient is, and takes those actions on your behalf.

Web developers, instead of building just for humans, will also build for agents. Humans and agents are very different. Humans are very visual — images and things like that matter. Bots don't care about that. It's actually more complex and expensive to process an image versus seeing an API that just has inputs and outputs.

A lot of what we're looking at is building in case users want it, but not forcing it on any user. That's why, when we look at sequencing, we wanted to make sure we launched AI controls first — which means you can turn all AI off in Firefox. All our AI options are opt-in. We try to clearly explain the privacy implications, because there are definitely privacy implications if you choose to use AI, and then give users increasingly more ability to tune this entire spectrum.

Our initial release with Smart Windows uses browser context to make things better. For instance, if you're looking at a lot of vegetarian recipes and then ask to create a meal for your family tonight, we'll say, "Okay, this person prefers vegetarian meals," and recommend vegetarian recipes.

In the future, there are lots of things we're looking at — browser controls like "bring up all the tabs I looked at for my vacation to Italy in the last year" or "close all my tabs that have been open over seven days." Or possibly agentic behavior, where you might say, "This restaurant is booked — can you keep checking the site and let me know when it's available?" Or even booking a restaurant on a user's behalf. These are all optional features, and this is where we really go back to our community to see what people want and how we can improve their lives.

Firefox Pioneered Tabs — and Now MCP Standards

Carlos González de Villaumbrosia | Product School 00:20:00
By the way, every single time I hear you use the term "tab," I have to mention this fun fact — you guys pioneered the tabs. The reason I started using Firefox back in the day is because I could open multiple tabs. Now it's like, okay, commodity feature, but that's pretty interesting.

Ajit Varma | Mozilla 00:21:00
Tabs and the pop-up blocker. Those are the two visible ones, but so many things that people don't realize — Gmail was not possible until Firefox pushed Web 2.0 forward with XML and JavaScript.

A lot of what we're pushing for now — even with things like MCP and AI standards — is making sure that these standards are open standards, because locking into any vendor from a browser is going to be the wrong thing for an open internet. All these things we're building for AI — you can pick your model, you can bring your own model. Choice and privacy are the two values we really care about.

Carlos González de Villaumbrosia | Product School 00:21:00
The image that comes to mind is you guys are a layer on top of these different models or even search engines — you're not literally building an LLM or a search engine. So when I hear you talk about optionality, there's another potential risk here, which is that it can sometimes be overwhelming for someone who isn't super sophisticated or doesn't really care that much. How do you give guidance so users can accomplish their objectives without having to make too many decisions if they choose not to?

Making Firefox Simple Without Losing Power Users

Ajit Varma | Mozilla 00:22:00
A lot of these tweakable things are what I'd say are power user features. There's something called about:config — if you type that into your address bar, you'll see thousands of ways to fine-tune and tweak your browser. Most people don't care about that, which is why it's accessible via something more complex like about:config. Most people look at settings, and one of the things we're pretty close to launching is a settings redesign that makes it as easy as possible to get to the version of Firefox you want just by going into settings.

Our default experience is something we're very unique in — we optimize for privacy. By default, if you do nothing and just download it and set it up, it's optimized toward the more private side of the spectrum because we think that's just important for a lot of reasons. It mainly goes back to user agency and choice.

In the future, one thing I think will be possible is that you can create any version of Firefox that's fine-tuned to what you like, and then with a simple URL, you can share that with anyone else. You can say, "I care a lot about sports, so I'm going to tweak this memory usage, this background, these widgets — this is sports-themed and sports-flavored and for streaming." Then you can take that link and share it with anybody.

Going back to what we started talking about — people are building browsers. But people don't need to build browsers. You can build any of those features within Firefox directly, whether it's an extension or add-on, and say, "Here's a link." If you want the OpenAI or Anthropic version of Firefox, you can easily get that link, install it, and it'll just be all the toggles and settings for that value you care about.

Building for Agents: Risks, Security, and Mozilla's Bets

Carlos González de Villaumbrosia | Product School 00:24:00
Speaking of agents or agentic-led products — there are now non-humans using browsers. You mentioned integrations with different agents such as OpenClaw. How does that impact the way you're building product?

Ajit Varma | Mozilla 00:25:00
The way a lot of these agentic behaviors — like bots — work is they're not actually getting access to anything secretive. They are acting like humans in a lot of ways. They're opening up a window, using image recognition to look at the page, deciding where to click and tap, and filling out information and taking actions on behalf.

There are some downsides to that. There are a lot of risks — browsers have cookies, logins, and passwords. There's a lot of risk if someone uses something like OpenClaw because it just has so much power. If you go on X, you can see people saying, "I didn't know what was happening, and then it deleted all my email or shut down my website."

When we're looking at a more thoughtful approach, we want to make sure that if we have more integrated support, we're considering the security implications so that things performing in a way a user doesn't want or expect are mitigated.

It's pretty early for us still. We're very thoughtful. We're reading and seeing everything that's happening in the industry. But there are a lot of unknowns — does the technology get better? Does safety get better? Do sites decide to build a parallel web that's MCP and built for agents? Or will it still be more human-built development where bots have to act in the way humans built it?

This is also where the Mozilla Foundation is doing some really interesting things. There's a product called Harbor, which is a layer on top of a browser that allows someone to plug in any of their AIs into the browser. We're also doing something like Tab Stack and headless browsers. We're doing different things across the Mozilla Foundation. Within Firefox, we're very aware of what people want, and we'll build ultimately what people want.

Headless Browsers, Agents, and Democratizing the Web

Carlos González de Villaumbrosia | Product School 00:27:00
There are two things I see happening right now. One is that in order to really deploy an agent, it takes work — you need to be relatively tech-savvy to figure out which agent, in what surface, the memory controls, all of that. But I can see a future where this would be seamless and you might be using agents without even knowing about it. The other parallel trend I see is headless — there are other interfaces for people to get to where they want to go that are not browsers. Maybe it's an LLM, maybe it's a voice interface. There's a possibility that a lot of traffic will skip a browser entirely. How are you thinking about headless, or being useful when people still require your product but might not be going to your product first?

Ajit Varma | Mozilla 00:28:00
The interesting part about headless browsing is that it is very disruptive to the way the internet works today. If you look at many companies, one of the challenges is that they make money on ads. They're not building ads for agents or bots — they're building ads for humans. So it disrupts their entire business model. What's the motivation for someone to create content if they aren't able to make a living on it? That's a worrisome sign for me.

We have to make sure that the open internet is preserved and all the good things about it are still there. But on the flip side, one of the things that was very powerful about the internet — that didn't really come to fruition — was that it was supposed to democratize access. I can type in a URL and go to anybody. Someone could just be an individual person in front of a computer, and I could go see their content because I believe in it. Over time, that got put behind gatekeepers.

It's crazy to think about, but there was AOL, and AOL had a closed garden. AOL failed because people didn't want a closed garden. People wanted an open internet, which had so much more content and power. Now we're going back to this closed garden where fewer things are controlling access.

But agents could also democratize access again. Say you wanted to order food delivery. Today, you're going to DoorDash or Uber Eats, and that company is taking 30% of the revenue. The value they're creating is being a gatekeeper — they're matching people. But the internet doesn't need a middleman. It can actually go around that. With an agent, I could say, "Can you go find a restaurant? Can I make an order directly at the restaurant? Then go find a driver and give the driver the order to pick up?" You're giving that margin back to the driver, giving that margin back to the restaurant, and those are the people who I believe are creating the most value that we want to preserve.

Agents can actually make it so that more places create sites. I've been creating a lot more sites than I've ever created in the past because it's so easy — AI helps on both sides. You can create content, and agents can get you directly to the content and remove the middleman.

This is so important for users to try to make these decisions, so we don't end up in a world where there's a single agent owned by a single multi-trillion-dollar company deciding what restaurants you're able to go to.

Carlos González de Villaumbrosia | Product School 00:34:00
It's been a pleasure to chat with you. I like your mission, and I love that you keep fighting the good fight. Thank you so much for sharing all these insights with us.

Ajit Varma | Mozilla 00:34:00
Thank you. Thanks for your time, and hopefully we've convinced a few people to download Firefox. If anyone does, we'd love to hear feedback. Looking forward to continuing the conversation on how to build the best browser.