从Perplexity对Chrome发起收购要约，深聊新一轮AI浏览器大战【101对话】

We are going through a period that number one guy is gone. So then who's the next? You're stuck in the business

the next? You're stuck in the business model and the product serves the business model. Browser war is real.

business model. Browser war is real.

Perplexity.

for my Tik Tok browser.

CL for Chrome.

Light speed venture partners.

[Music] Firefox, Google Chrome, Microsoft Edge, Safari.

First question, help us to define what is an AI browser. So, guru, why don't you start first?

I think it goes back to a very simple premise. Over the last 20 years,

premise. Over the last 20 years, systemata, actually more than that, more than 25 years, systematically the system of engagement for a human or a user with

computing has moved away from the operating system into the browser. So

everything that was relevant at the operating system layer is now also relevant at the browser layer. We used

to use operating systems primarily as a mechanism for security, observability, gathering data, running applications.

All of that now happens inside the browser. And so my going in thesis

browser. And so my going in thesis becomes u the primary mechanism for AI to interact with tools that were designed for humans is also going to be the browser. That then becomes an

the browser. That then becomes an interesting choke point where you collect data where you do automation where you secure AI to systems interactions and so on. So that's sort

of the framing that we use. Uh we've

we're already investors in this space.

uh as I was sharing with Howie earlier um few years ago we saw some incredible teams uh building enterprise browsers so version of the browser that say light

speeded would feel comfortable running because our IT team and security teams can then control uh what can and cannot be done inside the browser. So we

invested in a company called Talon which was then acquired by Palto Networks. Uh

and we were the lead investors there and we saw how quickly that entire market developed into a you know into a very large enterprise browser market today.

And I think the thing we are going to debate today is whether AI becomes another reason for a net new class of browsers to emerge and become a large market. So I'm excited.

market. So I'm excited.

Yeah. I think you know when uh the Talon acquisition happened AI wasn't the reason right?

That's right. It was cyber security.

that there are other reasons you know that happens but but that that that's an interesting thing still is you know we haven't seen acquisition of browser technology for a long time suddenly

browser is interesting technology you and I know each other for so long we discussed so many things but we never discussed a browser in last 20 years and then when I saw this article from guru I was like what you know he's interested

in browser now you explain it right you know you your your your initial interest in browser was this enterprise browser it makes a lot of sense is but now you are interested in AI browser you know

why are you interested in AI browser so it comes from sort of two key realizations um and a foundational uh belief our foundational belief here at light

speeded is that foundational models and AI in general represent a brand new class of capabilities that uh bring cognitive labor make cognitive labor

accessible to technology. So for the first time

to technology. So for the first time ever since we've been in technology uh machines can process language uh that is semi-structured, understand language,

work with it uh almost like a human would. So then the question becomes okay

would. So then the question becomes okay so what what's the limit on this one is you know how smart could models get? We

believe we're barely scratching the surface and models will continue to get better. We're one of the largest uh

better. We're one of the largest uh investors in companies like Anthropic as an example. So that's our core

an example. So that's our core foundational belief. Now on top of that

foundational belief. Now on top of that then the question becomes what holds it back? If these models get really

back? If these models get really intentional, what holds you back? And

essentially two things hold you back. A

model's ability to absorb data and learn and do a better job u in an enterprise setting. So observe and learn and two to

setting. So observe and learn and two to actually take action do something. So

tool access and there is one piece of technology that is the universal window to both observe what's happening and also take action and that becomes a

browser because almost I I suspect how before you came to this meeting or over the last many years 95% of what you do on a laptop or computer you're doing through the browser which

means if I was the browser I would know exactly what you're doing and all the software you're interacting with and I would be then presumably be able to interact with the software in that same

way. Now, it's not 100%. We have desktop

way. Now, it's not 100%. We have desktop applications too, but if you're going to start somewhere, you would start at the browser. And I think that's why browsers

browser. And I think that's why browsers were always very strategic real estate, which is why Google did the entire Chrome project. I mean, think about it.

Chrome project. I mean, think about it.

The the only reason they did Chrome was because they wanted to own search.

But it was such a strategic real estate that in the age of AI, it is several orders of magnitude more important. Um

and so that's why this renewed interest in browser that it addresses those two key things. It enables you to gather

key things. It enables you to gather data in a passive continuous way and it helps you to take action

continuously uh as both a co-pilot but also as an autonomous agent that can do work on your behalf.

Perplexity.

Perplexity.

Tik Tok.

Perplexity, Chromium, Windows, Mac OS, Linux Android Perplexity

list.

AI for agent.

I was talking to how the other day the news that Perplexity makes a long shot 34.5 billion US offer for Chrome. What's

your reaction when firstly reading the news?

So we're not investors in perplexity. So

we don't have any more data than the two of you do. But I will say one thing, there is zero downside to doing something to saying something like that.

Zero downside, right? So if it doesn't happen, you got great buzz. And at the end of the day, perplexity is a B2C business. So anything that gets you more

business. So anything that gets you more attention is all you need.

Attention is all you need. That is I love it. So attention is all you need.

love it. So attention is all you need.

So what's the downside? There is zero downside. And on the absolute sort of

downside. And on the absolute sort of small probability upside, maybe there is a situation where Google is either forced or wants to divest Chrome and in

which case uh why not make a play for it. And so I see that as a as a very

it. And so I see that as a as a very clever, very smart uh uh thing that Perplexity put out there.

Look, you know, I'm right now a chief AI officer at a public company. You know,

if not, I would have bid for that. Why

not, right? you know guru may be supporting me to beta for that who knows right uh so I totally agree with uh guru in that you know there's a zero downside

and if nothing else you know the worst case is tremendous amount of the attention right um I think the you know Google is going through a very interesting time right you know on the

one hand search business is under attack right you know for me personally I don't use Google search nearly as much as you know I did it before I can only see that that's going to be more and more true

for the mass population in the next 5 10 years. So that is a big threat to Google

years. So that is a big threat to Google right but at the same time they are also AI provider they also have a lot of smart engineers so so it's kind of very interesting time but since today we are

focusing on the you know browser war you know obviously Chrome is what 70 you know 70% market share plus right I mean it's the kind of the um uh gorilla in

this market I personally see that browser war is real 10 years from now the default browser for 70% of people is

not going to be chrome not even close uh the reasons are very simple for the reason that guru just mentioned uh today if you think about it Google's revenue

roughly $400 billion a year and then you know search is a big part of that a big part of that is actually sourced from Chrome so anything they do with Chrome

it has tremendous impact to the revenue immediately so um there are many different reports but you know the consensus is roughly speaking $80

billion of that $400 billion is directly directly from Chrome. So if I change the chrome user interface I'm changing $80 billion right how many company with 80 billion dollar revenue in the entire

world think about it right so this is like a big deal so I think they are going through this you know innovators dilemma it's not like they don't know what AI native browser should look like they don't know you know the thesis that

guru was talking about with AI I should be able to help the user to do browsing or to access the digital universe much easier but they cannot possibly do too

much a kind of a change. So this is a kind of a from first principal point of view that's a pretty obvious you know when I talk to people internal it's also the same thing they they literally told me that my hands are tied right you know

I cannot do a lot you know I know what to do but you know my hands are tied so in a way this is a very interesting that we are going through a period that number number one guy is gone that's my

assumption for the next 10 years so then who's the next right you know perplexity obviously has a comet there's a deer.

You wrote up these two browsers in your article which I disagreed is you should have put my browser ahead of these two.

Uh with all the kidding aside, I think all of us are at the very beginning, right? call me at the deer uh Nortm uh

right? call me at the deer uh Nortm uh Neil um you know this is kind of a let me just uh give a very brief introduction right I actually joined the

gen digital about a year ago as a chief AI and the innovation officer and among many other things I do you know I kind of started this AI browser project a year ago uh with pretty much I don't

want to repeat what you said everything you said is what what we you know saw at Gen Digital right the browser needs AI component component big time not just hey there's a button AI summarize this

page that's not a that's a that's like AI assist right you know I think the world needs a AI native browser you know you have to start from scratch rethink

about right in the past is like a human okay I want to go to CNN Fox News I click a link you know if browser give me that page fast I'm going to use this

browser that's pretty much why Chrome even went over um the Microsoft IE right you know at some point the JavaScript run much faster on Chrome, you know, that's that's why, you know, Chrome

disrupted II. But these days, I think

disrupted II. But these days, I think that's you still need that. But the way I interact with the digital universe shouldn't be like that. I know a link.

The browser should know about me. The

browser should bring the information to me even before I click the link. The

browser should summarize or maybe even negotiate certain things or do whatever the task I I wanted to do, right? It's

not just agendic. the browser should be ahead of you rather than browser reacting to you and it's a very personalized. So there's so much you can

personalized. So there's so much you can do with that. So, so a year ago we said let's actually rethink about the browser and then we started this project and then along the way obviously perplexity

you know arc right you know the these guys are you know probably thinking about the same line and then the way I think of the browser today is a very much like search 1997 you know I don't

even think you know the future of the Google in this place is necessarily born or even there or maybe in garage right but but I think the the idea is there

right you know Larry page already start thinking about this right 9798 when he started at Google this is where where we are um as far as what's the business

model as far as what the product looks like exactly I think it will take you know some time for us to learn and figure out but there's no question that we would have a totally different browser that the browser shouldn't be I

click a link give me the content that's like uh nothing right you know the browser should be my personal assistant giving me 10x productivity helping me to

do a lot of things 10x faster, right?

How to how to do this? So, that's kind of a um how we think about it. You know,

not neo was introduced in May this year and we believe that, you know, with our sort of the execution in terms of the bringing AI to the browser, the safety,

privacy, you know, we can talk more about it. You know, I think consumers

about it. You know, I think consumers are going to love it and they're going to have totally different experience.

Fore Eric Smith.

Safari, web kit, I said the number one guy will be gone like Google probably won't be the leader in AI browser. That's what how he said.

Do you agree, bro?

Um I think the the the landscape is wide open for the first time in 15 years.

both of the major call them strategic pieces of real estate that Google owns.

One is uh search and the other one is browser. There's two most strategic

browser. There's two most strategic pieces of real estate that they own. I

think both are under threat. I think the opportunity exists. Um and it's up to us

opportunity exists. Um and it's up to us to folks like Howie and others to think through how do you not just do something evolutionary, slightly better, but revolutionary, right? You have to

revolutionary, right? You have to rethink the system of engagement. If

your browser is only slightly faster that was great for Chrome in the in the war with IE, it's not enough now, right?

What is enough now um is yet to be invented. It's a complete rethink of how

invented. It's a complete rethink of how that interaction of a human user with the with the worldwide web happens and what's the opportunity to rethink that portal through which we interact with

the web. So, I'm excited to see the

the web. So, I'm excited to see the innovation coming out in this. I'm

really excited. I think for the first time we're going to see something that is magical because that opportunity exists now. I think voice will be part

exists now. I think voice will be part of that. Better security, more

of that. Better security, more automation, u easier discoverability. Uh I think this all of these things are up for grabs.

That's excellent. You guys mentioned some uh AI browser products already like perplexities, comet, comet, and not neo,

right? The rumor says OpenAI is going to

right? The rumor says OpenAI is going to launch its AI browser.

I too I will bet that over the next two to three years we will see tremendous innovation and more browsers in this space. We will see

startups. We will see large AI labs. We

startups. We will see large AI labs. We

will see certainly perhaps even companies like Google reinventing either Chrome or a net new browser in different ways. We're going through this

different ways. We're going through this Darwinian sort of explosion of a complete change to how we interact uh

with the world and I think u to to to some extent even claude on your desktop or chat GPT on your desktop is competing with the browser to your point you know

I couldn't agree more you said I'm using Google less and less man I'm my first stop is not Google for anything now when I'm looking for information I'm in a Somebody says quantum computing. I was

like I don't know anything about quantum computing. I don't go to Google and say

computing. I don't go to Google and say please tell me about quantum comput. I

literally go to claude and say give me sort of a quick introduction to quantum computing. 5 seconds later I've got the

computing. 5 seconds later I've got the perfect sort of write up. I go through it and now I'm at least conversant in it. And so I think uh there's some

it. And so I think uh there's some fundamental changes happening. Browser

is going to be a key part of that. I

will not at all be surprised if OpenAI either develops their own or makes a play for Chrome. Uh it will happen. I

think we'll we'll we'll see tremendous innovation in this space.

While we have not heard the rumors that Anthropic is developing a browser, don't tell me no comments that I cannot comment on. We we look as as you both

comment on. We we look as as you both know, we are huge fans of Anthropic for a very simple reason. uh it's the best team that is pushing the frontiers of

intelligence but in a way that is safe, secure, uh trustworthy, interpretable and these are things that matter to folks like Howie and I because of our enterprise roots. And so we think

enterprise roots. And so we think there's just a tremendous future in applying intelligence and cognitive intelligence especially in enterprise settings. Uh for sure Enthropic will

settings. Uh for sure Enthropic will continue to experiment with new ways in which you can connect their models and their intelligence to both the data that

an enterprise needs and also the actions that these models need to take to automate. Could that be a browser? Sure.

automate. Could that be a browser? Sure.

But that's up to Mike Kger who's a CFO CPO and his team.

Right. Do you think the AI browser war will be the battleground for giants only or startups also have an opportunity because we can see the LLM game is

totally like big players game already.

100% startups have uh uh it's a level playing field. It's a level playing

playing field. It's a level playing field and the reason for that is actually interestingly the fact that Chromium which is the underlying engine inside Chrome uh is open source. uh

maybe 15 years ago I wouldn't have been so bold to say that a startup will just come out and and write a browser from scratch. The reality is the internet is

scratch. The reality is the internet is such a diverse place that supporting all the different types of things that you need to support to create a browser is a pretty tall task. Uh but I'll give you

the proof point for what makes me very bullish on uh new players emerging not just the larger companies. uh Howie and I uh have a bit of a cyber security

background as well and about five or six years ago a few founders decided there is an opportunity to build a brand new browser for enterprises that is secure

and you would have a blink reaction say that people like Chrome or Safari or others there's no way that's a real market and the reality is we were lead investors in a company called Talon that

got picked up extremely quickly by Palto Networks uh and is now an extraordinary large business inside Pala Auto Networks which is a large cyber security company.

There's another company called Island which is also doing really well. And so

there is an opportunity for startups.

Both of these are startups. They built

the stuff off Chromium. You can pick up Chromium as a startup engine as as your startup building block and then rethink everything else about the user interface and the system of engagement. Optimize

for security, optimize for automation, optimize for AI in different ways. Um

and build something new. So I think the field is wide open. Right. Hi.

Well, you know the I would say it's only for small small players. However, I agree with Guru that

players. However, I agree with Guru that Google may have a different project outside of Chrome, right? You know,

that's a to me that's a small player within a giant. Uh that's possible. I

just cannot see the giants that this sort of the number one players current number one applications is going to shift too much for the reason I just mentioned. I I just don't see that

mentioned. I I just don't see that possibility. I do see a possibility

possibility. I do see a possibility another interesting project you know surfaced but usually you know how it goes in a big company it's tough right that's for the you know typically so I

would say that's u I I might be biased but I would say you know it's actually for people who doesn't have market share it's actually you know time for them or

me [Music] Foreaf [Music] Safari. Forever.

Safari. Forever.

And how will AI browsers disrupt the current way of advertisers? because uh

search engines like browsers are the main traffic entrance for a lot like SEOs rankings display ads right so how

will this new AI browser thing disrupt the money-making machine for Google and Meta I think you're asking the right question but it's a narrow question I would I

would encourage you to take a step back and ask a bigger question what is search in this brave new world that we are going in right so historically uh all of us used search as a discovery

mechanism for us as a human. You want to travel somewhere, you say Hawaii, great hotels, and then you would search. And

you have to ask yourself five years from now or some long arc of time, five years from now, 10 years from, are you really doing that? Because are you really going

doing that? Because are you really going to this bar and typing in Hawaii great hotels to see 10 blue links and one by one clicking through those and figuring out who else is reviewing this? And is

that really going to be the system of engagement? I can tell you my bet is

engagement? I can tell you my bet is zero chance. zero chance five or 10

zero chance. zero chance five or 10 years from now you're going to this small bar whether from Google or Bing or some other search engine I think that entire world gets completely disrupted

completely changed because what you will do is to a trusted agent or a browser that you trust and say I'm thinking of going to Hawaii what do you think it's

like well here's some options uh the search will happen but it's getting pushed one layer down and is getting done by AI and so I call this to your

travel agent which is the browser about hey here's my plan you figure out maybe you need to do search maybe you need to talk to cloud or you know open AI whatever right you know that's your job

my job is to talk to my travel agent that's right now think about the repercussions of that the entire search industry today entire search industry is based on a very simple assumption that

there's a somewhat impatient human who doesn't have perfect quering abilities sitting there and wants an answer right away Right? And that's why ever ask

away Right? And that's why ever ask think why do you why do they only show 10 search links? That's because if you showed 40 I would be overwhelmed. If you

showed only three I won't quite trust you. So you know somehow with enough

you. So you know somehow with enough experimentation you decided 10 seems like the right number. And so the entire industry is focused on this this act of

me going to this one bar you know typing in an imperfect search query and then going through results one by one that AI agents that is not how. So for instance,

one of our companies uh XAI is building a search engine specifically for AI agents because AI agents the way they search, the things they search for and the kind of queries they can ingest back

is vastly different from human. Human

today does this great hotels in Hawaii, 10 links show up. I click on each one of those. Here's what I do, by the way. I

those. Here's what I do, by the way. I

open each tab. Now I've got 11 tabs in my browser. Then I go one by one each

my browser. Then I go one by one each tab. I don't like any of these. Then I

tab. I don't like any of these. Then I

go next page and I click through zoomer and oh my god it takes hours. What a

frustrating thing. You know what an AI does? Doesn't care for 10 tabs. It's not

does? Doesn't care for 10 tabs. It's not

going to open 10 tabs and then go browse those hotels. It makes one query says

those hotels. It makes one query says give me 100 hotels and give me all the metadata associated with and give me all the reviews all in one shot. One shot

comes back jam into the context window.

Ask for people based on what you know about Guru. What are hotels you think I

about Guru. What are hotels you think I might like on these days? And by the way, don't even suggest something that's not available on those dates. And it

shows me three things I was like you know what I feel like going to this one right now. So the entire notion of how

right now. So the entire notion of how search happens changes and that is the bigger threat to search. The threat to search is not just the browser moving

from Chrome to Neo. The bigger threat to search is the fact that it's not a human searching anymore. I just don't believe

searching anymore. I just don't believe that human initiated search queries will be the dominant search 10 years from now. it is just not the case. It is AI

now. it is just not the case. It is AI that is doing that search and discovery.

And so we are investing in teams that are making that world a reality. And so

the browser becomes more of an interaction mechanism through which uh you're driving automation. You're having

conversations. You're doing things that matter to you as a system of engagement for entertainment or work. Search as a function gets delegated to AI that does it fetches the right piece of

information for you in that moment.

I very much agree right you know before you talk about monetization business model we first need to talk about what the future looks like right you know you just laid out the future Chrome to Neo

browser right that's just one part of the future but also how the search is being done which layer you know in the tech stack I totally agree with you know but just um one additional color to

share about this business model right I would say we don't know what the business model is in the sense that if you think about search why Google even

you know why we even have a Google that's because Google when Larry Page actually I heard this acquired you know uh podcast that

was a fascinating story it was 1997 before you know Google started uh Larry Page went to this company called Excite right you know it was one of the you know double digit number of search

companies right you know and um you know Excite was doing pretty well and then they tested right you know I search and then Google search and their team is

like uh you know felt strongly Google search is far better far better. So the

CEO you know in the last meeting CEO went to the room and said hey you know let let me figure out you know should I you know acquire the technology or whatever you know we wanted to do right the moment he saw like a Google search

was a far better and then people will you know the first link is so relevant he basically said get out of the room why is that because of the business

model because the excited business model is about banner ads so the longer people stay on that page the better. So if the first blue link is actually the relevant

one, my banner ads revenue is going to be shrinking not increasing. So so to me like for us to thinking about okay ads revenue this and we're going through

another major shift. So I would say whoever thinking about ads revenue the way you know it's being sort of done today which is enormous you know is

wrong. Now the only guy who couldn't say

wrong. Now the only guy who couldn't say that is Google right he cannot say Google cannot say well I cannot I mean they have to right they have to report to the go to the wall street every

quarter right hundred billion dollar I mean most of that is ads so I would say Google is screwed for that reason now of course Google has a gem that model has you know a lot of other things right

they may have a spin up new projects so I think the future is unclear from that point of view but I want to say to answer your question one is what I totally agree is what guru says you know think about what the future looks like

you know I think it's a different future and the secondly what is the business model for the that new future rather than the what's the business model in the current way of thinking of ads

you know there's a um there's a there's a way to think about this when you look at a particular market are the products serving the business model or is the

business model serving the product in the early days of development of a market. You come up with a great product

market. You come up with a great product and then you attach a business model to it to serve the product because you want to build more product. You want more people to use it. This is the early days of Google. And then you get to a point

of Google. And then you get to a point at some point after tremendous success, you're stuck in that business model and the product serves the business model.

This is when you go from Google's search engine tremendously successful is now stuck with this business model. So

instead of three uh or four banners on the right side, you start showing those results at the top of the search results. And then slowly you start

results. And then slowly you start making them look more and more and more like an actual search result. How many

times you do you do a search query nowadays on Google and you're like is this a is this a sponsored result or a real result. Like I have to pay

real result. Like I have to pay attention to that little italics at the bottom sponsored. I was like nope.

bottom sponsored. I was like nope.

Scrolling right past. But I doubt everyone is that careful because they make it look like a search. Now, now you have a situation where the product is serving the business model. That's where

the opportunity is and that's that's why venture capital exists. Those founders

that we are funding, they don't care for the business model. They want to build the absolute best product. We've both

been founders. We didn't start our companies going, man, what's a business model. I believe no, we we saw an

model. I believe no, we we saw an opening in the market to create an amazing product and then we bolted on a business model to that to serve that product in the right way. And I think that's that's the opportunity. The

market is a natural way of the world. I

don't blame Google or anyone else for that. uh it's their success that lands

that. uh it's their success that lands them in that place and it is that the fact that you're stuck to your business model and the business and the product is merely serving the business model now

and not trying to be the best product it can be that is the opportunity for founders and you know we're pretty excited about that. Y

yeah yeah yeah that's great timing as you said I often click the first or second one in the Google page and then turn out to be something totally different you know sometimes I joke about this

competitor company will show up in because of the ranking there are times on my mobile I can't even use Google anymore if I use it I have to go through one screen and then a second screen now I'm pass the sponsored results to get to the actual search

query why would I use that but I can just fire up claude and say here's my query give me the answer this that's because they have a product serving the business model. That's the

that's the reason.

That's exactly that's an easy explanation.

Yeah.

[Music] What are some technology breakthroughs or uh advancement do we need in order to overcome the current barrier?

Since I have been living and the breathing in the browser revolution the last one year, this is a topic dear to my heart 24 by7 right? So I think we are go for the AI native browser we are

going through three different phases at this point. You know the first phase is

this point. You know the first phase is pretty much chat and then search right the merge or like why do I need a chat and why do when do I need a chat when do I need a search it doesn't make any

sense for people to say oh go to Google for search and then this question I mean you need to give me a unified experience so that I have a questions I have something go one place you know that

place will give me chat if needed give me search if needed right give me the direct navigation if I want to go to my bank account I don't want to go to Tim Blue Links, I want to go to Bank

America, right? So the navigation, chat,

America, right? So the navigation, chat, search, that uniform experience needs to be there, you know. Um I think for the AI native browser like you know Neil, right? We already have gone to that

right? We already have gone to that phase and I think many other browsers either realize that or about to get there, right? You know the search and

there, right? You know the search and the chat merge. So I would say this is kind of a phase one. Um

the phase two and the phase three the timing is a little bit controversial.

I'll give you my point of view but I think you know the industry has a different point of view. My that phase two in my opinion is you know the browser you know not that search and the

chat needs to be merged. The browser

needs to be a lot more proactive a lot more personalized. It cannot be just I

more personalized. It cannot be just I go I want to go to CNN or Fox News and then give me the link before I go there right there is something it's it's being done maybe notification maybe you know I

I'm reading this part of the news right you know and uh of course I didn't pay attention all the details and then two days later it was say hey you know how you are interested in F1 right by the

way there's event you know of F1 that of the page you actually browsed either right now or just two days ago and then do you want to be kind of a notified of such things, right? So there is a

superman sitting next to you almost, right? You know, browsing together with

right? You know, browsing together with you and then if you if it will keep you it will help you to memorize things, it will help you to grab things. It will

help you to go to 50 different websites if needed and summarize if needed.

Right? So so I call it some personalized proactive uh agent sitting next to you or Superman sitting next to you doing that. So that's kind of a what neo

that. So that's kind of a what neo browser is working on delivering the value. The faith three is agentic by

value. The faith three is agentic by agentic is you can do a lot more complex things right you know for instance I want it's not just some simple phone feeling you know I wanted to achieve

some complex things right the aenic browser is like a a my assistant right you know I wanted to go to Hawaii it will figure out you know not just you know the top um top places I wanted to

go to but give me you know very detailed the potential booking of this you know a lot of more things right I view that you So the merge of the chat and the search as a phase one, phase two is the

personalized the proactive browser and then phase three is agenic. I think in the industry there are people who actually believe we need to go jump to the agentic browser right away. I don't

disagree. I mean in fact we do phase two and phase three at the same time. Um I

don't believe the technology is quite ready for full-blown agentic browser right for instance you know if I want to buy a uh iPhone cover on Amazon right I

wouldn't trust the agentic browser to do that yet because you know it may work eight out of n 10 times or nine out of the 10 times but what if the one out of the 10 times and it doesn't give me a

iPhone cover it gives me a iPhone right that's not what I want right so I wouldn't trust the the the the AI you know nearly as much yet to so but if we

are able to get to that phase three that's not just a search navigation chat merge but also search chat and the action merge I feel like the technology

is not quite here in terms of the action right you know simple action sure that's what we are trying to do as well so so we are going through you know for the neo browser we already finished the phase one right we we feel like we are

leading the rest of the pack we are leading the rest of pack in terms of the you know moving forward forward on the personalized the browser and then the proactive browser and then you know guru is going to try out and say wow I didn't

realize it can be this personalized surely and I'm going to culture um once you said that but on the in the in all seriousness right you know there's a lot of things to be done by the way it's not

like you know just you know integrating with anthropic and openi it's done you have to think much harder in terms of what's the user experience right because at the end of the day you know I might

be proactive but it could be very annoying right you know so so how to balance the user experience it's a very tough problem so we are going through that phase two agentic stuff you know we

are also delivering some agentic stuff but I view that as a you know potentially a year maybe 18 months maybe longer right you know you know we are going to gradually making some progress

that's sort of how I see uh the technology on the browser side yeah I like this framing I like this framing of phase by phase of you know starting in a way that consumers can very easily sort of adapt to the new

model, then the depersonalization, then the action and orientation. It almost in the way you were describing it, Tavi, something struck me like this might be the time that the browser actually becomes sort of the AI operating system.

That might be the larger opportunity in the fullness of time that the browser is no longer sort of this this little window through which we you know access information and take actions on the internet or applications but but truly

an an operating system for our lives where it is a gentic it is uh proactive uh and of course it is a true partner in terms of combining chat and search and

sort of brainstorming with it and and I think that's a world I'm excited in. Um,

I would take the other side of that agentic bet though. I think the some of the stuff that Enthrakic and others are working on is just so remarkable. I

think we're all going to be surprised by how quickly automation comes into our lives. And uh, that's exciting.

lives. And uh, that's exciting.

Cool. You know, I'm a big fan of agentic stuff. I I'm not saying that when I say

stuff. I I'm not saying that when I say a year, 18 months from now. I mean the, you know, the the the the high accuracy, right? You know, the the agentic sort of

right? You know, the the agentic sort of the outcome, you know, but otherwise agentic is here and the now. So, you

know, I'm I'm, you know, fascinated about.

What do you think about voice? Howie?

Does it have a renewed place here? I

mean, it's gotten so good. Like

sometimes I'm not even typing. I'm just

sort of talking.

Yeah, that's very interesting. So, when

we started working on Neo, um, and the first version we shipped, you know, the alpha version, you know, last uh, you know, about December, we already had a voice in it.

I would say it's interesting.

um there's not as much adoption of the voice initially but it's unclear because the voice is a wrong feature or human just need to adapt to it right because

uh you know for mobile I think people are more used to doing the voice for the desktop right I don't think people are kind of okay I'm talking to a laptop it's it's less kind of a natural um but

who knows right you know it may be a matter of time or maybe you know I think there are two different things here one Is is the voice technology mature enough, natural, cheap enough, low

latency enough? I want to say we have as

latency enough? I want to say we have as a industry right has made a tremendous impact, tremendous improvement in the last 12 months. So I would say the

excuse is no longer the technology readiness. So that's number one, the

readiness. So that's number one, the technology readiness. And the second

technology readiness. And the second thing is, you know, is the technology suitable for that use case scenario.

Right? I would say we haven't seen that yet.

I was a big fan of the adopting voice into the browser you know you know working with my team we got this feature in but you know I was also told that no one used that feature um so we're

I think your instinct is right by the way that there is like voice is so uh finicky you know it 95% accuracy is not

enough like it has to deeply understand what I'm saying with my accent with the kind of words I use if I'm a specialist in some domain in finance or I'm a doctor. It needs to understand those

doctor. It needs to understand those terms. Uh it needs to be real time. I

need to be able to see very quickly what it's typing. So there is a very high

it's typing. So there is a very high bar. My instinct is we're very close.

bar. My instinct is we're very close.

We're very close. Um there's also a very high bar for the user experience because if I'm talking, I'm talking. If it waits for all of that to finish and then show me the text, that's not going to work. I

I need to see it getting typed up right away to get confidence that it's doing what. So I think the technical bar is

what. So I think the technical bar is very high, but we're almost there. It

feels almost there and then you're absolutely right on the other side too is it suitable for that use case and I think there will be a few set of use cases where it will become habit forming

very quickly very quickly that's my instinct but we'll see I'd be curious if you turn that back on and invest more in it how that goes maybe that's a follow on conversation and then you are going to be the uh user

of our voice 100% okay 100% I got it the first beta user for Um, and how about visual? I think visual

probably is the next stage for browsers because visual has like tons of other information that probably voice or language couldn't describe, right? Yeah.

But we're still probably a little bit too early nowadays as we're waiting for the AI hardware.

I I think um I'm very excited about new visual interaction models that AI enables. I do think it's a bit early

enables. I do think it's a bit early right now. Uh but the technology is

right now. Uh but the technology is getting there. To give you an example,

getting there. To give you an example, today the entire browser engagement model is I I go to a website and a

pre-laidout website with its images and text and everything shows up. Maybe

there's an opportunity based on what AI knows my intent is. It generates that entire output. Could be a video, could

entire output. Could be a video, could be an image, could be text. Everything

is generated just for me in that moment and visuals are a huge part of that including how it's presented, right?

Maybe it's a mind map of some sort because that's what I like. Uh maybe

it's mostly images because AI has learned that for me I'm a visual thinker and I like to see images. Maybe it's

mostly text because AI has learned that I don't like images. I actually like to read through stuff as quickly as possible. So I do think gen the ability

possible. So I do think gen the ability to generate images and videos is tremendously exciting but it is right now expensive.

Yeah.

Unpredictable. The same prompt will get you two very different images every time. Uh and fairly high latency. Uh I

time. Uh and fairly high latency. Uh I

know teams that are working on all three of those making it more predictable making it very low latency to real-time video generation. We are beginning to

video generation. We are beginning to see some demos, especially in gaming of infinite games, right? As you're

playing, it's generating the world. It

just never finishes. So, it is generating the world in real time faster than wall clock time. And so, you'll never see the end of that video. Uh, so

we're beginning to see that. That's

exciting. It's a bit low resolution right now, but it's beginning to get there. Uh, and then the cost has to

there. Uh, and then the cost has to drop. I think as that happens, I'm I'm

drop. I think as that happens, I'm I'm excited. It's not a six-mon thing it

excited. It's not a six-mon thing it feels like, but it is there out there.

So this time you know guru talked about the engineering side of thing right the cost the latency the you know the technology maturity part I'm going to talk about the vision side this is where

I see that 10 years from now maybe you know less than 10 years you know um not two years but you know some years from now when we look back the way we use

browser is totally unbelievably wrong.

Why do I say that? Right today, you know, you go to whether again Fox News or CNN, whatever your favorite news website is, you go there, you see the text, you see the video that, you know,

they give to you, right? But guess what, right? You know, why do I consume, you

right? You know, why do I consume, you know, text just because CNN, Fox News give to me, right? It should be me, right? I need a video when I want to

right? I need a video when I want to watch video. I need a text when I want

watch video. I need a text when I want to, you know, read a text. The source of the the data is, I wouldn't say irrelevant. is not shouldn't be static

irrelevant. is not shouldn't be static right so we do need to see the browser that's multimodality meaning that it's a layer in between right I have a need okay I'm going to watch news I'm going

to watch video right then you know whether the CNN Fox News everything it doesn't matter if it's a text it will be converted into video for me but at sometimes right you know I'm driving you

know and on the on the passenger side seat and then I just want to scroll some even if I go to YouTube right you know I just want to see the summary of the text uh rather than you know video right

because it's the wrong environment for me to to do it things like that so I would say multimodality is is a different way you know it's going to allow us to unlock the value of me as a

consumer consume information in the in the more appropriate in the more efficient way uh instead of me being locked into CNN give me this piece of

things in text give me this in in audio I have to stuck with it I don't have to be stuck with it you know what's the most exciting demo I've seen in there. Uh I'm sure you've played around with this product too.

It's a product called Notebook LM from Google. It was so amazing, right? We all

Google. It was so amazing, right? We all

read papers or blogs or press releases.

You take that same paper, you feed it to Notebook LM. While you're driving, it

Notebook LM. While you're driving, it appears at a podcast. Really well done product. Really well done product. Now

product. Really well done product. Now

imagine a next iteration of that where it discovers, you know what, Guru, instead of a podcast, prefers to watch a YouTube video of this thing as a talking head. Now you've got two avatars in a in

head. Now you've got two avatars in a in a freshly created video just for you that are discussing that technical paper and going with each other. I mean, we're we're going into a brave new world that

that's going to be really exciting. The

mode or the modality in which you consume information will be decided in a personalized way in real time. If in

this moment I want to consume that information as text, it'll appear as text. As a podcast or audio, it will

text. As a podcast or audio, it will appear as podcast or audio. If a video, it'll appear as a video. That's a world we're headed in and that's pretty exciting. I think a lot of opportunities

exciting. I think a lot of opportunities in sort of especially in consumer land related to that.

So the browser war is real.

It is real.

I know like the AI browsers offer convenience um and a whole new future as we talked about, but they could also centralize user data collection in ways

that could raise privacy concerns for sure. So how risky is that from both

sure. So how risky is that from both governance as well as the trust perspective? Maybe how you go first.

perspective? Maybe how you go first.

Yeah, I can go first because this is a topic that I kind of see both sides of the coin, right? On the one hand, a lot of time people want convenience, you know, just to get you know ship anything

and everything to your you know whether the model provider or whoever as long as I get the best answer right but then there are people who care about do I give sensitive information do you retain

my prompts right because my promps is you know for some people they don't want you know other people to see about it right so to me you know it's very important for the consumers to have the

choice so the analogy I would use is if you think about messenger right you know there are messenger do end to end encryption and then you know the extreme version of that is probably signal right you know people have the peace of mind

you know whatever I do with signal I get to see it you get to see it if I do it no one in middle right you know that's great and then there are other messengers there are always you know you know there's a visibility by people in

the middle right so there's a wider spectrum um our design you know I I don't I'm not saying that everyone needs that but at least from Neo point of view

not a Neo point of view. We wanted to give a browser that's almost like signal uh of the browser so that you have the confidence that uh it's as long as that's your choice that's your

configuration that's your setting the prompt saves on your local device only it doesn't you know saved on the you know my server or you know whoever's

server you know one may argue well you know cloud days isn't that shouldn't be a big deal anymore but guess what certain people really care about it right so so I want to say for people

care about privacy for people care about security. That's why you know a lot of

security. That's why you know a lot of people came to Norton Neo because of that reason because Norton carries the security uh sort of the peace of mind for them and also our design. But I

would also say that I'm pretty sure there are people who would tell me that I don't care about any of that. All I

care is doing some fancy agentic workflow as long as I get that thing done you know sure you know um that that's there will be those kind of the uh scenarios too. What do you think

there? There's um

there? There's um there's a few words that are all interrelated here. A few concepts.

interrelated here. A few concepts.

Security, privacy, trust, because privacy and trust. If I

trust you, I don't need to be very private. If I don't trust you, I need to

private. If I don't trust you, I need to be very private. So, and then finally, alignment. Are we aligned in our

alignment. Are we aligned in our worldview? When I say something, share

worldview? When I say something, share something with you, do you treat it with the the same way I would treat it? you

know, it's treat others like you would treat yourself kind of. So, is there alignment? Is there trust? Is there

alignment? Is there trust? Is there

security? And then is there privacy?

Because privacy in of itself doesn't mean anything. Obviously, you share

mean anything. Obviously, you share everything with your spouse. Uh but you don't share much with a stranger. That's

all related to trust. And so when we think about these four things, our core belief is every layer in the technology stack needs to think of these primitives

and give the user as much control, give the upper layer as much control as possible. How's team is doing this with

possible. How's team is doing this with the end user. So when we use it, we can decide how much p how much do we want to trust, how much privacy we will dial. I

think most consumer applications will have to offer that choice because this technology unlike preceding technologies is a extremely powerful but needs access

to your life and your information to be even more powerful. So the delta between not giving it any information and giving it information its effectiveness is very

is actually quite different. And so I think the temptation for giving it more information is going to be very high.

And so I think it would be very difficult for application providers, browser builders to defacto assume how where to set that dial. I think what is

more important is to recommend certain settings and give that dial. I'll give

you another example deeper in that layer. There's probably an integration

layer. There's probably an integration with Enthropic that Neo has. And so

we're again, you know, we love Enthropic's vision here because again that focus on trust, safety, security, alignment is extraordinarily important

because as an application builder, you can provide all those controls, but at the end if the intelligence doesn't really respect those directives, then you've got a bigger problem on your hand. So along this line do you see that

hand. So along this line do you see that in the future the consumers will you know own some I don't know the digital vault and I say hey this part I give to

browser I this part I give to that AI agent because I'm going to have six agents or 60 agents I'm dealing with anyways you know I'm going to you know

sort of keep the memory the data in one place I can centrally manage and then depending on which agent do do you see that or how do you think about it I'm curious because you see so many different applications. That's why I'm

different applications. That's why I'm asking you this question.

I think there's a few cyber security nerds like Howie and I who are extremely careful with managing information and which applications get what application.

But my observation is the average consumer um the cognitive burden of parceling information into these sandboxes and then giving individual sandboxes to

individual applications. The cognitive

individual applications. The cognitive burden is too much for the average consumer. I can tell you um people

consumer. I can tell you um people outside Silicon Valley or people outside the cyber ecosystem uh you know when you log into something with Google and then Google says this thing will have access

to these seven Google services are you okay with it the number of people who read that and take action on that is zero people are just like yes I just want to use this application so the

answer is for a small subset of people yes this ability to enclave information will be important and a lot of those use cases will be econom eomically valuable because if you're working at the NSA obviously it's very valuable so you

might be able to build a business around it but I do not see a world in which individual consumers uninformed masses billions of customers out there will be able to parcel this information in the

right way we can barely do that in our personal lives like it's just so difficult to remember what did I tell my spouse and what did I tell my kid and what did I tell my cousin like it's just hard you know the cognitive burden is

too much now if AI can help us with that that could be interesting I think consumers you know when When it comes to privacy security stuff, uh the consumers do not like the cognitive friction right?

And also they do not like pay for it.

When it's a free and the frictionless, they will say privacy. But as long as one of the two things or maybe both, they say, well, I don't know what are the privacy focused search engines would

have been the dominant search engine on the planet and not Google if consumers really truly walk the walk. People like

to talk about privacy, but at the end of the day, they want to just do what they want to do. Needs to be frictionless.

They want to pay for if it's free. If

you're going to incorporate privacy into it, great. If you can't, most consumers,

it, great. If you can't, most consumers, in my experience, don't care.

Wow.

They don't care. Hey, another question.

So, consumers do not want to pay for anything in the past, right? Gmail or

anything, right? But now they pay $20 a month for, you know, anthropic or open AI. Do you see that kind of the you know

AI. Do you see that kind of the you know the behavior it's a it's a shift of behavior. What do you see?

behavior. What do you see?

I think the actually going back to a previous discussion I think the business long-term business models are still to be discovered here. I think the initial paying 20 bucks a month, five bucks a

month, 100 bucks a month depending on the utility you get. It makes a lot of sense. I think it provides very high

sense. I think it provides very high signal back to these companies. It also

pays for the very high cost of inference in some cases. Um, and the thing I love about that is it could historically even

we technical technically relatively forward-leaning people, we didn't spend 20 bucks a month on five, six, seven different applications. Now we do. And

different applications. Now we do. And

the thing I love about that is that's the value we're getting out of AI is so high that we finally got broke out of that habit and started paying. Imagine

like 20 years. 20 years we did pay for never paid for anything and then starting a year or two ago we were like oh my god like I can't live without this stuff. Swipe a credit card and go for

stuff. Swipe a credit card and go for it. So that's amazing. Uh long-term is

it. So that's amazing. Uh long-term is that the business model? Unclear. I

think we'll have a mixture of everything. The winning business model

everything. The winning business model is still to be discovered. Could be hey we provide you the most secure and private browser. Could be and so you

private browser. Could be and so you should pay for it. Could be ads. Could

be something else. So that's I think unclear. I do have one last question.

unclear. I do have one last question.

So, what are the most exciting startup opportunities you see for the next 5 to 10 years as we watch this AI browser war starts?

Um, so obviously it goes without saying that AI is so foundational and such an amazing building block that nearly every part of

the technology market and beyond uh will present great opportunities. So, I won't rule out anything. Having said that, some of the most interesting opportunities and a great test for

founders is could this be built three years ago? It's a very simple question,

years ago? It's a very simple question, but are you doing something that could have been done 3 years ago, you're just doing it slightly better, or are you doing something that 3 years ago was just economically or from a technical

reason just not possible? Three years

ago, could you build technology? or four

years ago, could you build technology that could go through all the uh uh PDFs that a company publishes in their

financial um disclosures and give you a summary of those practically impossible today super easy build against that. So

that's a test I use uh I'm pretty excited by businesses that three or four years ago were simply not feasible either economically or technically. It's

a very clear why now. So I I I enjoy that but I'm curious how how you view that.

Well, going back to one of the points you made earlier, right? Browser is not just the browser. It's going to be the AI operating system, right? So from my point of view, you know, I have bias but

you know Neo is going to take over Chrome completely at that point. This is

the future of the operating system. I'm

going to work with 60 million applications working with my browser. So

there are tons of startup opportunity hosting on Neo. Uh that's what I see for the next 10 years. Actually that's a great point. You are pushing the

great point. You are pushing the frontiers of what a browser can be and those frontiers then enable a new class of applications on top in the same way

that the incorporation of JavaScript inside a browser led to Ajax Gmail and all kinds of fun stuff that was built on top. Actually that's I wasn't thinking

top. Actually that's I wasn't thinking about that but this is a great point.

The innovations at the browser level innovate themselves might offer a new class of applications to emerge.

Yeah. And uh all the agents working with my browser is the lead generation for live speed. And

live speed. And there you go. So 50% of the leads will be hosted on my browser anyways.

I'm definitely subscribing and paying for this browser then.

Okay, here you go.

Sounds exciting. Wow. All right,

gentlemen. That's all my questions. Do

you have anything else to add? I still

have the question about why we don't hear about from anthropic on that.

I only wish the Indians were any less tighter. All I can tell you is it is an

tighter. All I can tell you is it is an incredible company and I wish I could share more. Uh I have never worked with

share more. Uh I have never worked with a sharper, more thoughtful u uh bunch of people. It's just an incredible team. And as you and I, we've

incredible team. And as you and I, we've had long careers, you know, at this point in our careers, it's about working with incredible people who push our own sort of uh point of view of the world,

our own thinking, and and I I'm couldn't be more privileged to work with that team. So stay tuned is all I would say.

team. So stay tuned is all I would say.

But uh anyway, this was great, Howie.

I'm really glad you suggested it, and thank you for moderating. This was

amazing.

Anthropobic Fore browser.

Bye.

[Music]