AI Browsers and the Future of Web Interaction: A Glimpse into Agentic Browsing

I switched from Safari to an AI browser and got a glimpse of the future

Agentic browsers can eliminate some tedious tasks, but you're still better off searching on your own... for now. Generative AI is the transformative technology of the moment. Tools like ChatGPT are helping people realize the power of AI by automating tasks such as writing and proofreading text, creating images and videos, and even solving complex problems-or, at least, providing insight toward solving a problem. AI is also changing another routine task: browsing the web. Companies like Opera and Perplexity have been working on AI-based tools called agentic browsers. Agentic browsers are capable of understanding the content of a web page in real time and also using AI to interact with the page and run workflows for you. Agentic browsers promise to change the way people interact with the web. Instead of having to worry about finding things on your own, agentic browsers take control of everything so you can focus on other tasks. The promise sounds too good to be true, but do these browsers work in real life? I'm a long-time Safari user. It's not the best web browser in the world, but it's undeniably the one that has the nicest integration with Apple platforms. While Apple has been struggling to deliver AI capabilities, I dove into testing two different agentic browsers to find out if they're really game changers and if they can convince me to switch. I was invited to try out two different agentic browsers: Opera Neon and Perplexity Comet. Both are currently available in a very limited beta. When you think of a web browser, there's usually a common interface that comes to mind. At first glance, both Opera Neon and Perplexity Comet look like something you'd expect from a browser, but with a touch of an AI assistant. Both have a ChatGPT-like interface where users can ask anything. Here's where they differ. Instead of searching on a conventional search engine like Google or DuckDuckGo, these browsers use their own AI chatbot to answer your questions, so you don't have to go to websites to find something. Instead of navigating to Google or a website to start your browsing session, you ask the AI agent a question. It's not a button like Copilot in Microsoft Edge -- the browser is completely built around interacting with the AI engine to conduct searches, fill out forms, book trips, and create web apps. To see what each browser responds to a query, I asked both browsers for the "latest iPhone 17 rumors." Opera Neon generated a text about the main rumors, while Perplexity Comet went a bit further and combined both AI results with some links to articles on the web about what I asked. However, there's a lot more you can do with these browsers than just ask questions. Imagine, for example, that you need to buy flight tickets, but you're not in the mood to open the airline's website and find the flights on your own. An agentic browser can take care of that. I asked both Perplexity Comet and Opera Neon to book tickets for a specific route in September. I made it clear that I wanted the cheapest tickets possible. They both opened the airline's website and selected the cheapest fare. Impressive, although Perplexity Comet didn't proceed to checkout until I gave it another command; Opera Neon did. Opera Neon can also generate webpages and mini web apps for you. I asked it to make a page comparing all the rumored iPhone 17 models, and it built one in less than five minutes. One thing I've learned from these AI-based browsers is that you can get better results by being incredibly specific. When want them to buy something with Safari, for example, you can type exactly what product or service you want (iPhone 16 Pro Max) to closer to what you want,rather than typing, "I want to buy a new iPhone." It's a little different when using AI agents. When I asked the browser to find me a hotel in São Paulo, Brazil, that's what it did. But when I asked it to "book me a good hotel in São Paulo for September 17-18 and make sure you find hotels around the Jardins neighbourhood", it took me straight to the checkout page based on what I requested. Is booking a ticket on your own still faster? Probably. Both AI browsers took several minutes to complete the most complex tasks, during which time I could have searched on my own. But again, the main point is to let the computer do the hard work for you while you do something else. This is probably one of the best showcases of the usefulness of AI in our lives. Compared to companies like Opera and Perplexity, Apple is noticeably behind when it comes to fully agentic AI experiences. While macOS 26 and iOS 26 bring improvements to Apple Intelligence, the features announced thus far are more limited than what agentic browsers are already capable of. Unlike Perplexity Comet or Opera Neon, Safari hasn't integrated any agentic capabilities (at least not yet). Apple Intelligence focuses on personal context with things like rewriting emails and summarizing notifications. It's an impressive first step, especially given Apple's emphasis on privacy and on-device processing. But Apple Intelligence, for now, isn't really trying to change how you use the web. Right now, I still feel more comfortable using the web the traditional way, but it's undeniable that having a browser that does things for me sounds quite appealing in certain situations, and I'd love to see Apple working on something similar. That might happen. Reports suggest that Apple has internally discussed the idea of acquiring Perplexity, the startup behind the Comet browser. Valued at around $14 billion, Perplexity could help Apple catch up in the AI race, especially in areas like web search and agentic browsing, where Safari still lags. It's unclear whether Apple will go ahead with its acquisition plans, but with the billion-dollar deal between Apple and Google under threat due to antitrust laws, the company would have a good reason to invest in its own AI-powered browser and search engine. Agentic browsers like Opera Neon and Perplexity Comet are not perfect and often painfully slow, but they offer a glimpse of what the future of web browsing looks like. While I'm not ready to give up Safari just yet, I can see the appeal of letting an AI assistant do the boring stuff. As the technology evolves and gets quicker, I can see it replacing Safari for some tasks, especially if Apple doesn't make an effort to catch up. Whether Apple builds something like this itself or ends up acquiring a company like Perplexity, it's clear that the web as we know it is going through a major transformation. And this time, AI isn't just enhancing the browser. It's becoming the browser.

OpenAI Wants Google Chrome -- Here's Why AI Companies Are Circling The Browser

(Image credit: Pavlo Gonchar/SOPA Images/LightRocket via Getty Images) Chrome isn't being used to its full potential, the founder of Google's generative AI team has explained, days before a judge rules on whether Google should be broken up, potentially through the sale of its web browser. "Google built Chrome as a way to defend against potential competitors in the future. So by owning their browser, they can secure a spot for their search," AI pioneer Jad Tarifi has outlined in an interview. This strategy may have worked for the past 17 years, but the vultures are already circling the world's most popular internet browser. With a judge set to rule on the search giant's future by the end of month, the tech world anxiously waits for the gavel to fall to get more clarity on what will happen next. AI companies, including OpenAI and Perplexity, have already publicly expressed their interest in buying Google Chrome. However, tech entrepreneur and privacy expert Mark Weinstein dismissed the recent bids for Chrome as "lowball numbers," noting the enormous value of the user data that would come with it. "Sam [Altman] must be drooling at the possibility of monetizing with this kind of data, regardless of whatever altruistic statement he makes," Weinstein has explained. If you're looking for a breakdown of why Google Chrome's sale is being discussed and what it could mean for browsers in the age of AI, you've come to the right place. The first question you might be asking is, why would Google want to sell its own Chrome browser in the first place, given its massive success? Even describing it as massively successful feels like an understatement, given that the browser is estimated to capture over 60% of the global browser market share. However, what for Google would have been considered an inconceivable idea became a possible remedy suggested by the US Department of Justice's antitrust division, the one that deals with laws meant to prohibit companies from engaging in anticompetitive practices. A US district court later ruled that Google violated antitrust laws by monopolizing online digital ads. So while Google isn't voluntarily looking to sell Chrome, it has been specifically asked to divest from the browser to address these antitrust challenges. "Google Chrome isn't for sale," Info-Tech Research Group analyst Thomas Randall expressed in an interview. The software powering Chrome is open-source, but Google owns the brand and the default funnel into its services. From Randall's perspective, very few companies could afford Chrome or survive the regulatory scrutiny of such an acquisition. "Antitrust regulators in the U.S., EU, and beyond would almost certainly block a sale to another tech giant, particularly one already dominant in AI or digital advertising," Randall said. Similarly, Andrew Gamino-Cheong, the co-founder of AI compliance platform Trustible, said Google is unlikely to want to sell Chrome. "They'll fight pretty hard against antitrust regulators for a long time around that. And there's probably other things they would try and do first to appease them, long before they're forced to sell," Gamino-Cheong explained. From his end, senior VP at Pierpont Communications Chris Ferris said since monopolies are bad for consumers, a court forcing the breaking up of Google could be a win for regular users. AI company Perplexity, which runs the AI-driven search engine that shares the same name, made global headlines with its $34.5 billion bid to buy Google Chrome. Interestingly, the unexpected announcement came relatively close to the launch of Perplexity's own web browser, Comet. "I was talking to an early Googler a few days ago, and he said this is probably the first time I see that I'm worried for Chrome," tech VC Nick Davidov has told me. "You cannot really migrate people with all of their details unless you give them a very strong reason to switch, and this AI browser is a very strong reason to switch." Davidov, who co-founded Davidovs Venture Collective (DVC), was an early investor in Perplexity. He explained that Comet was based on an acquisition from DVC's own portfolio, a company called Sidekick that created a so-called headless browser, a kind of browser that scrapes information from a website without rendering it visually. So, if Perplexity released a browser to end all browser wars, why would they want to also purchase the almost 20-year-old Chrome? "Oh, because of distribution," Davidov explained. "Instead of having to get all of the users to switch, they will just have them." Some people were sceptical about Perplexity's bid since the three-year-old company is valued at 'just' $14 billion. However, Davidov said the company has several heavyweight investors that could pitch in. "If the judge says that Google has to sell Chrome, it doesn't mean that Google has to sell 100% of Chrome. It just has to stop being the majority shareholder, right?" Perplexity offered $34.5 billion to buy Google Chrome, a figure Davidov describes as "a bargain". For comparison, Microsoft acquired LinkedIn for $26.2 billion back in 2016. "It's probably worth a bit more than that," the founder of Google's Generative AI team, Jad Tarifi, told Tom's Guide in an interview. Tarifi, who now heads Integral AI, a startup working on AI agents smart enough to be useful in the real world, said that valuation would mean capturing the attention of 3 billion people at $10 a head. "The acquisition cost of the attention of someone is higher than $10, so... purely based on that, [buying Chrome at that price] makes sense," Tarifi explained. Although he said it's still a risky move since just purchasing that attention doesn't automatically mean Perplexity would be able to extract the whole value from it. Weinstein expressed that $340 billion would be a more accurate valuation of Google Chrome. He described Perplexity's offer as "a bankruptcy price." Tarifi has suggested that Perplexity's multi-billion-dollar bid was partly a publicity stunt. Still, from an entrepreneurial perspective, he explained, making the offer was a smart move -- even if the odds of actually acquiring Chrome are slim. "At least they put themselves on the map as a serious, ambitious company," Tarifi said. He added that for OpenAI, making a bid is less about pure publicity, as he imagines the company would be able to do more with the acquisition in a smoother and faster way. OpenAI's Sam Altman recently told The Verge over dinner that he would be interested in purchasing Chrome if Google is forced by courts to sell it. "One of the secrets to Google's success was that its search function was built into their browser. In fact, the address window has effectively served as a de facto search box for many years. So if an AI company purchased Chrome, they could make the address bar function as a de facto AI query box," Ferris said. Gamino-Cheong noted he sees three main reasons why AI companies could be interested in owning their own browser: Gamino-Cheong believes there could be new data privacy laws since the value of data will increase: "It's not just statistics about user behavior that's good for ads. It's the raw content itself that's now the important part. Companies are getting more aggressive in their kind of desire to collect and use that directly." That's aside from the potential intellectual property issues all-seeing browsers could create. "If I'm, let's say, creating my own art inside the browser, and it's observing that, you know, I want to make sure it's not taking a photo of that, using that to train potentially without consent," Gamino-Cheong said. AI models are also data hungry to sustain themselves, but not all data is healthy for them. In 2024, researchers published a study showing that when large language models (LLMs) are trained on too much AI-generated material, the LLM collapses in a phenomenon called model collapse. Gamino-Cheong said that it may be hard to figure out whether a piece of text was written by AI just by looking at its final version. However, if a browser observes the whole writing process, it could accurately determine which parts were drafted by a human user and which parts, if any, were edited or generated by AI. Weinstein highlighted that AI companies are incentivized to monetize data they gather through AI search. Algorithms have already been collecting data, AI just does it faster and instantaneously analyzes it, he said. Could everyday users benefit if an AI company created a fully-functioning AI browser? Gamino-Cheong believes so. For one, it could make using agentic AI smoother. He recalled testing ChatGPT's agent mode, which opened a new browser and said it could finish tasks if he logged into the necessary accounts. "That's a big scary part, because I now need to put in the raw text of my passwords into the system," he explained. An AI browser that keeps you logged into apps and services could offer more security, he added, especially if the AI ran locally on your own device. Davidov's advice to Perplexity would be to add their AI as a paid feature on top of Chrome. When it comes to data and privacy implications, users should expect more of the same if Chrome were to change ownership, Weinstein said, warning that AI could know more about you than any country or secret service. "Google's already using AI. I mean, what's different is that the data points to a different player," Weinstein said. It would be Google that has to figure out how to continue collecting all this data that it would suddenly be missing if it were forced to sell Chrome. According to Weinstein, this isn't the first time users have faced such dilemmas. "We've been here before," he said. "And it was called the rise of Facebook." With VC fundraising dropping 33.7% year-over-year, there's less liquidity for riskier deals. With chatbots like ChatGPT, Perplexity and Claude already reducing the need for Google searches, spell-checkers, and translation tools, could it be that AI chatbots will eliminate the need for a browser in the first place? In this case, is Perplexity's $34.5 billion bid a smart investment or a risky gamble? Davidov said he doesn't believe browsers will become obsolete. Rather, AI browsers will enable the personalization of the internet. They'd implement your site preferences throughout the web and filter your searches for products or services tailored perfectly to your needs. Tarifi, on the other hand, believes browsers will be dynamically generated in the future. "If you think of text-based chat, like ChatGPT now as MS-DOS, right, there's going to be a new move in the next few years, from MS-DOS to Windows. So, from a purely text-based chatbot to fully multimodal, immersive operating system interfaces generated on the fly." However, Tarifi predicts that the owner of Chrome, be it Google or another company, would be in the pole position to deliver this new era of browsers to users. Gamino-Cheong said eventually users would just need to log into one AI system from where they perform all their tasks. Browsers would play a more behind-the-scenes role, similar to operating systems, as they stop functioning as the user interface. Chrome is still one of the best free browsers, working reliably most of the time. But it hasn't evolved much, and with new AI browsers like Perplexity's Comet emerging, users now have real alternatives. Some might argue that while Google is reluctant to let go of its flagship, younger AI-first companies could reinvent Chrome. Tarifi said users should have some sympathy for Google, which he says is facing a classical business dilemma. He said generative AI that summarizes information for users instead of directing them to sponsored links puts their income in jeopardy. While Google may have the best talent and AI tech, it makes its money from a legacy business model, Tarifi said. "They're stuck between this rock and a hard place," he added. "So, are they using it to the best of their capability now? No," Tarifi said. "But is the giant waking up? And might they change? Quite possibly."