EP 565: OpenAI’s Agentic Browser: Is this the future of content consumption?

This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life.

It's almost as if some of the big tech companies and startups got together and they said, you know what? This is going to be the week of the agentic browser. Between some startups that have released some updated features and versions of their agentic browsers to Perplexity releasing their Comet browser to more users. And now a recent report talking about that OpenAI will be releasing its own agentic browser. And it's something we haven't talked about a lot on the show, but I think it's worth really diving into and investigating a little bit more, not just about the bigger picture of what is agentic browsing, but also the biggest AI company in the world is going down that route. And so what does that mean, not just for all of us, but how we consume content and on the services side as well? So I'm excited for today's episode. I hope you are too. So welcome. What's going on, y'all? My name is Jordan Wilson and welcome to Everyday AI. This is your daily livestream podcast and free daily newsletter helping everyday business leaders like you and me not just learn what's happening in the world of AI, but how we can actually leverage this information to grow our companies and careers. If that's you, that's what you're trying to do, you're in the right place. It starts here. But if you really wanna be the smartest person in AI in your department or at your company, you gotta go to our website, youreverydayai.com. Sign up for the free daily newsletter there. We're going to be recapping the key, uh, points and insights from today's show, but also keeping you up to date with everything else you need to know in the world of AI. Uh, so I'm excited for that. And if you want the daily news, uh, sometimes we do it here on the livestream podcast, but, uh, just go check that out in the newsletter today. All right, this one's gonna be a fun one because as a former journalist, I always take a lot of pride in, number one, making sure anything that we talk about is well-researched and factual. But I love talking to other journalists and people who used to be journalists, used to be in journalism. And that's why today's one is, I think, going to be a great show because we actually have one of the reporters who just broke this recent story about OpenAI going down this agentic browser route. So this is going to be a good one. So livestream audience, please help me welcome to the show as we bring her on. Crystal, who is the reporter at Reuters for AI and startups. Crystal, thank you so much for joining the Everyday AI Show.

Thanks for having me. Excited to be here.

All right, let's just get straight to it. If people read our newsletter, they saw your story in there. But tell us what happened with OpenAI all of a sudden going down this agentic browser route.

Thank you so much for featuring our story in your newsletter. It's definitely an interesting development for OpenAI. We have been following the company for a while, and obviously everything they are planning on doing, especially on the consumer side, because, you know, ChatGPT, since it burst onto the scene, it has become the fastest-growing consumer product in history, and now it has 500 million weekly weekly active users. So whenever OpenAI releases the next consumer products, if it gets adopted by one of the users of ChatGPT, it will have an impact on the existing market. That actually led us to the reporting about browser, and we are hearing they have been working on this project actually for almost a year now and are looking for, for a final launch this summer. Some of the features we have heard, which is pretty cool, is they're adding more agent systems into this browser. So think Chrome, but also with the interface of ChatGPT, that you can get a lot more information and actions done within the chat interface. Maybe you'll be doing a little bit less like tab switching, and you can make comments through the chat interface. So it's all very exciting, and we're trying to get a glimpse and help you understand what might come and how it might impact the current way people use and browse the internet.

And, and that one is, I think, very important. People, like how people use and browse the internet because it's become a cornerstone, right? Like if you're a knowledge worker, for the most part, your day is in a browser, right? So I, I don't think not everyone has fully been able to grasp what this means because You know, very few people have been using an agentic browser because it is a little newer, right? But we've seen other companies, the DIA browser, GenSpark, Fellow AI, Perplexities, one that they just launched, Comet. So it seems like it's one of these categories, right, that seems to be exploding. What does that mean for the everyday business professional who's working all day, obviously, in a browser? Are things gonna change overnight? How does this change how we consume information?

Yeah, I think it's really interesting to see this becomes a summer of browser wars. Like, you're seeing so many launches at the same time now, adding OpenAI to the list. And if you look at the background a little bit, browser has been this uninteresting corner of internet for a while because Chrome was just so good and so dominant. With over 70% of the marketplace. And I think what OpenAI seen this opportunity is since already have so many people using ChatGPT as the starting point of their everyday journey online, and also they're starting to use ChatGPT to replace some of their search behavior. So what if we provide a quote unquote, according to OpenAI sources that we have talked to, the more optimized experience experience, which is more in the chat interface information and action instead of opening 50 browser tabs like I am right now. So it is really interesting to see that vision that they have. And I think it might have a real impact on the ecosystem of the internet. Because if you think about the fundamental economics of the internet, at least for a lot of content owners like publishers and social medias, It's about, you know, getting the clicks and people going into the website and spend some time there. You know, you will see the ads and you may even click on the ads and make some purchases. And that's how content creators on the internet get rewarded. And now if a lot of things will be happening within this context window you have with ChatGPT on the front page of the browser instead of all the tabs, I think some of the websites will get much fewer clicks and traffic. And actually, this is already happening. I'm hearing so many publishers are complaining about what they're seeing in the drop of traffic. So everyone is trying to figure out like what we do there. Do we partner? Do we fight against this? It's very, all very active conversations.

Yeah, I think we covered that earlier this week. It was something like 35 or no, 37 of the top 50 50 publishers, according to SimilarWeb, I think saw huge drops in traffic, mainly because of AI. So, okay, I have to ask because the former journalist in me is wondering, what does this mean for the companies, the publishers, the journalists that are actually creating this content that large language models are, you know, using to serve these answers, right? Because you talked about clicks. For a lot of publishing companies that maybe don't have subscription services or, you know, monthly, uh, you know, premium opt-ins to, you know, have more unlimited use of a certain website or a certain publishing company. What does that mean for these other, other publishers as these AI agentic browsers become the norm?

As a journalist/content creator on the internet, I definitely spend a lot of time thinking about that, uh, and covering that as well, because from the industry perspectives, uh, I think it's very interesting to see I see the approaches that AI companies like OpenAI and Perplexity, as well as Google, Meta of the world are taking. I think it's a mix of carrot and stick approach when it comes to working with the publishers and content creators on the internet. I think one playbook that we can use as a reference is how people are gathering their training data when they train all the large language models. So you have seen in that age of era, players like New York Times sued OpenAI for using their content illegally, and that's still actively playing out in courts. But on the other side, you actually have publishers, you know, like AP, like Reuters, and actually striking some deals with some of the players. Reddit is an example as well. If you remember that they signed this content licensing deal with Google that will pay them $60 million a year. So the comments from you and me and the users will be the training data for Google's AI. But at the same time, they also sued Anthropic for scraping their websites heavily. So last week I talked to folks at Cloudflare who oversees a lot of internet traffic. So they had this tool to help publishers to either block, or you can use a pretty novel monetization way, which is pay-per-crawl. So it's not pay-per-article, but every time I crawl your website, I need to pay you. So I think people are experimenting a lot. And to your point, in this age with the new agentive browser and the search function, I would expect some of the AI companies are also behind the scenes having those conversations right now. Who should we partner to get those content and actions needed in front of our, you know, 500 million ChatGPT users, or who we should, you know, just like trying to partner behind the scenes. And if some people do not want to partner with us, can we go to others and use that kind of pegging against each other? See, this company has the visibility and distribution on ChatGPT and OpenAI browser, but you don't. So I feel like that will be the approach they are taking here.

Yeah, that's an interesting one, especially, you you know, we covered that earlier this week talking about Cloudflare's, uh, new, uh, pay-per-crawl, uh, and also blocking AI, uh, bots and AI crawlers by default, right? So shutting off those, uh, you know, big tech to like, uh, I think 20 or 25% of the entire internet, uh, I think could really change the future of training data and potentially, uh, bring future model quality potentially. I mean, is that something that came up in your reporting at all, or is that maybe looking too far down the future on what happens if these big tech companies are just able to get a little less or fewer pieces of high-quality journalism into their training models?

Well, besides worry about training models, I also just in general worry about as an internet users on the internet browser experience, it's like everything, like internet was known as this like very diverse, like different corners you can explore. But now if the incentive is not quite there for people to create the most original content and they got rewarded for their content, like, do people still have that incentive to, you know, put in there? And I have heard this thoughts about, you know, for if you talk to like those website builders, they're having talks about, oh, in the internet, there will be specialized websites built for AI. Instead of human because you want it to be more optimized for AI scrollers rather than human users. And that could, you know, require different UX design and that could require totally different experience. I think that is just really fascinating to me. And also I want to add, we have spent a lot of time, you know, talking about the impact of a more agentic internet on publishers, content creators. But potentially if you talk to all the AI people, The optimized internet experience is just not— it's not just on a content side. They also want it to optimize the experience, the clicking, like think about booking and ordering. So I've heard like similar conversation happen. You may have seen the operator from OpenAI. This new browser will incorporate some of the capabilities that they will be able to go on other websites like order food delivery or maybe book a flight to you. And when I talk to, um, those kind of service providers, I think there is a real debate going on, much mirror the ones that content creators are having, is do we want to partner with them or do we want— just want to say no, like fend off our APIs, like not working with them because we want to own that direct relationship with our customers.

So I do want to dig a little bit more into the details on the service side, but one thing I just completely forgot to ask— I kind of assumed everyone knows what agentic browser is, but we didn't really go into the details. It's in your story, but can you just describe a little bit what's the difference between a normal browser and an agentic browser, at least in the lens of what you reported for OpenAI's agentic browser? How does it work? You mentioned operator.

Walk us through it.

Yeah, for the record, I have not had the pleasure to try OpenAI browser yet. But so I got information from my reporting with people who had direct knowledge. But from what we know so far, and obviously we'll get a better picture in a few weeks when they finally launch it, but from what our reporting so far, the browser will have like a default ChatGPT entry point where you can do search, you can use it as the way you're using that right now, but also you can make some commands and give it a bit more flexibility and power to potentially connect with other parts of the internet. And also it can maybe do some actions of switching back and forth of the tabs for you and following your more complex command. So to your point, agentic in general is an AI system that can take actions. There are smaller ones on the basis level, for example, to browse an internet page that you ask the AI to do. Browsing itself is an agentic action, but it's relatively a basic one. And there's more active and complicated ones where it's taking step-by-step action and they know exactly what to do to perform a complicated task. For example, order a pizza, but specifically from this brand and deliver to this place. An agent has been a kind of buzzword in AI for a while. And I think the reason people have been talking about it is like, it's such a great vision, right? Like you just see, oh, you don't have to do anything, you just need to give a command. But in reality, when I talk to people who work on those products, it's actually very demanding on the system because the room for error is very low. Remember, they have to follow step-by-step commands. So if you make a really big hallucination on one step, you may not be able to just like follow the whole command. And so that's so far, I think the agent experience is not like mind-blowing, it's not as cool as the demos show so far.

Yeah. And I think maybe it's important to kind of dig into the differences, right? Because as an example, you've brought up Operator, right? And we've covered Operator plenty on this show. We've done the live demos. I haven't done anything yet with Project Mariner, with kind of Google's version, but you know, walk us through what's the difference between maybe using a computer using agents like Operator, like Project Mariner, like all these others versus an agentic browser, right? Like, because there the AI lives right in the browser. You can't really get around it. What's kind of the difference between at least, you know, what you found out in your reporting? Is it just going to be performance-wise so much better? Right. But I think that's where a lot of people might get confused.

Yeah, I think it's more of a full-service experience. So think browser as kind of the interface or base where you can use amazing tools like ChatGPT, like Operator, or other agents maybe come online in the future. Like, we need a base to start. And the other interesting, I think, most kind of important aspect of browser being known as an unprofitable business is it does give, you know, the browser maker a lot of information of your online presence, and that's enormously helpful for them to either want to develop a business model in the future if it wants to be ad-targeted, or it, you know, trying to decide on its next roadmap, what is the core use cases they need to focus on using agents. Oh, is it really restaurant booking? Is it shopping? Or is it actually something more? Repetitive or simpler to some extent. So I think that really the browser is like the foundation that lets you perform different activities and interact with different agents.

Thinking about this from a zoomed-out perspective, I can't help but to think a little bit about Google's previous partnership that came under a lot of scrutiny with the DOJ with their paying $20+ billion to Apple to become the default search engine, right? And how that really helped Google, you know, not that they needed it, right? But that helped them maintain their hold on that 90-95% share of the search market. Is the search market, is the search engine going to become a thing of the past? And is it just going to be all about who has the bigger share of the agentic Browser Pi?

AI moves too fast to follow, but you're expected to keep up. Otherwise, your career or company might lag behind while AI-native competitors leap ahead. But you don't have 10 hours a day to understand it all. That's what I do for you. But after 700+ episodes of Everyday AI, the most common questions I get is, Where do I start? That's why we created the Start Here series, an ongoing podcast series of more than a dozen episodes you can listen to in order. It covers the AI basics for beginners and sharpens the skills of AI champions pushing their companies forward. In the ongoing series, we explain complex trends in simple language that you can turn into action. There's 3 ways to jump in. Number 1, go scroll back to the first one in episode 691. Number 2, tap the link in your show notes at any time for the Start Here series, or you can just go to starthereseries.com, which also gives you free access to our inner circle community where you can connect with other business leaders doing the same. The Start Here series will slow down the pace of AI so you can get ahead.

I think it totally depends on how soon the agent browser can get really good. And so that is a question both on technology, user adoption, and how soon you get partnerships from other participants on this internet, like web publishers, like service providers. But to your point, Google is in a very interesting position even in this browser world, right? And they are actually facing an active antitrust trial regarding whether their Chrome browser will be sold off or not. And guess who said they wanted to buy it? It's OpenAI. So they said publicly that in April that, oh, if Chrome is forced to sell, like, we are in the market, we're interested in buying it. So it will be interesting to see what the competition dynamics look like once OpenAI gets out their browser. And, you know, that could may have an impact on the active trial on if this is still a monopoly in the market.

And let's, let's go into the competition piece a lot more because I'll say before the fourth quarter of 2024, it wasn't a competition. From my viewpoint, I can say that. I'm not the reporter on this story, right? But OpenAI was so far ahead of everyone else. And in December, Google came seemingly out of nowhere. And since then, it has been such a heated competition between OpenAI and Google in terms of benchmarks on large language models, fighting for active users. But at the same time as the competition is heating up, they're also getting friendlier, right? So in your story here, obviously OpenAI choosing Chrome to build on top of Chromium for their Agentic browser, but also one of your earlier stories about the partnership between Google Cloud and OpenAI. So how have you seen this kind of fierce rivalry still be very fierce, yet they're also getting friendlier and friendlier in some instances?

I'm so glad you asked about this because I think the frenemy dynamics between the big players in AI is one of my favorite topics following this because you got the best business story, personnel stories, and also money stories. So the dynamics between Google and OpenAI, as you pointed out, is definitely a very fast-evolving one. And the back— part of the background is OpenAI was once known with Brahman's best relationship with Microsoft, who basically bankrolled its initial development and helped with this distribution. But since last year, they have actually been in a breakup process, like trying to figure out what's the new stage of their relationship look like. So if you are OpenAI, like, you are knowing, like, you don't— you are not gonna be exclusively tied to Microsoft anymore. That actually opens more door for you to get new partnerships that potentially means more distributions for your technology. And potentially you can use other relationships to get more compute, which is absolute number one cost at an AGI lab like OpenAI. So what we have reported is on the compute side, they actually went to Google Cloud and signed a partnership there. They're telling them, oh, now we're not going to just use Microsoft. We're not going to just use our own Stargate because is very— a lot of that are still being built right now. We also want to get some allocation from you guys. So if you are giving me the kind of power of compute to develop and also run inference for a lot of our products, that's actually a good deal for us. And guess what? Google actually said yes, despite the competition that it has on search. And now, as we reported on browser, because I think they see an opportunity there as well to potentially expanded pie for their cloud business, which they have always been, you know, the third behind AWS and Microsoft. And they see an opening there. And that didn't affect— they are still being very actively, you know, competitive on other aspects, especially on consumer.

It seems like there's been a lot of momentum and investment and time spent rightfully so, with enterprise companies trying to move a lot of their day-to-day operations in the front end of large language models, right? You see different reports from OpenAI saying that more than, I think it's 90% of Fortune 500 companies are using ChatGPT, not on the back end, on the front end. How might this whole agentic AI change that, right? Like, I know you can't see in the future, you know, you and I haven't been able to use this new OpenAI agentic browser, but, you know, has this come up in conversations or how do you think this might ultimately change how people are trying to bring in their work and their processes into this, you know, quote unquote small little box of an AI chatbot if it's actually just getting bigger and bigger and, oh, it's the actual browser now?

Yeah, that's a really good point because I think that leads to another kind of an unexplored question is what's a kind of an enterprise use cases. For a browser, and how is that going to be your personal use cases? Like, if history taught us anything, I feel like Chrome is a very much both consumer and enterprise product at the same time, which means if you get the best browser, you can potentially, you know, get like touch points from both market. And everyone is, you know, betting on like enterprise market is going to be the more profitable one because people are going to willing to pay for So I think that opens a lot of questions about if you were to automate a lot of workflow and if it's already done on a browser and how much of the workflow can be automated using a browser and adding AI agent services, which I think can be solved to some extent by the browser itself. But I think a lot of it, if you're talking about enterprise environment, it actually requires a lot of proprietary data that may not be available through the browser itself that needs to connect into more of your existing workflow and data resources. So without that, I think it would be really hard to make it like a truly different work experience with AI agent browser alone.

All right. So Crystal, we've gone over a lot in today's conversation. You gave us some of the inside scoop on this exclusive story on OpenAI's new browser, talked about how this might change both content consumption and services, how the enterprise might react to it, and the evolving relationship between OpenAI and Google. But as we wrap, what should our audience be keeping their eyes and their ears on in terms of, you know, not just this OpenAI agentic browser, but how it might impact, you know, business leaders on their day-to-day?

That's a really interesting question, and I like when you have an open question. I don't have a specific answer, but one thing that I find interesting is to pay attention to where AI is best at doing, because as time evolves and actually like being hearing different answers from people who are developing those systems, if you, you know, kind of zoom back a few years ago, you may remember all people talk about was AI replacing some of the blue-collar workers or manufacturing floors and all that and robotics. And then when generative AI wave comes in, everybody was like, oh, it's gonna replace the knowledge workers and any like white-collar jobs that requires writing, whether it's marketing writing or something else. And now the latest wave I have heard is actually AI. What AI has been really good at is to solve reasoning and math problems that has a very deterministic answer. So that's not gonna be writing, but a lot of it's like coding and it's like a high-value technical engineer type of jobs. That people believe AI would be best at. So I think there is generally a lot of sentiment and uncertainty about job replacement in the AI age right now. And we at Reuters are also spending time, you know, covering that impact on the society in general. But from like talking to the tech insiders, how perspectives around that evolve is very fascinating to me. I'd encourage your listeners pay attention to where AI is best at too.

We'll be continuing to pay attention both to your stories, but also just everything that's happening here because it might be happening under the surface, but I think it's extremely important. So Crystal Hu, startup and AI reporter at Reuters, thank you so much for taking time out of your day to join the Everyday AI Show. We really appreciate it.

Thanks again.

All right, y'all, there is a lot there. If you missed anything, don't worry, we're going to be recapping it all in our newsletter. And if you didn't miss it before, we're going to go ahead and throw it in there again. So make sure if you haven't already, please go to youreverydayai.com, sign up for the free daily newsletter. If you're listening on the podcast, thank you as always. Please make sure to subscribe to the show, leave us a rating. We'd appreciate that. We'd also appreciate if you come back and join us for more Everyday AI. Thanks, y'all.

And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going. For a little more AI magic, visit youreverydayai.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.