In 2001, a decade after he invented the World Wide Web, Tim Berners-Lee articulated a new vision. Together with two colleagues, he published an article in the magazine Scientific American in which he outlined what he called the “Semantic Web”: an internet designed to be navigated by autonomous software agents that could address our every need, without requiring us to muck about with browsers and search engines ourselves. To illustrate what this might look like, Berners-Lee and his colleagues told a story about a pair of siblings who use their agents to automatically research and book a physical therapy appointment for their mother, making sure it would be covered by insurance and fit their busy calendars.  At the turn of the millennium, this vision seemed far-fetched: Google had only been online for a couple of years, and most people were still on dial-up connections. But a quarter-century on, the dream of autonomous software agents that use the internet and carry out tasks for us has finally been realized. New tools such as Meta’s blockbuster Muse agent and OpenAI’s Dots are logging record downloads and being given access to people’s bank accounts and email inboxes, and Anthropic launched a research preview of its own Claude Duet agent in October. And in a worrying incident, agents created by OpenAI broke free of their badly protected sandbox and caused havoc on the open web.   Just as the rise of smartphones sparked the development of a more atomized and attention-grabbing internet, the advent of powerful AI systems heralds a new era: We’re facing not the semantic web but the agentic web. This new web is developing out of sight and in ways that might not be entirely within our control. As it grows, the agentic web will inevitably transform the human web—and whether the human web survives this transition is anyone’s guess. For Berners-Lee’s original vision to work, internet users would have had to annotate the pages they built with special tags so that a software agent could understand each element’s meaning—say, “street address” for the part of a physical therapy office’s web page that indicates its location and “appointment types” for the list of services. And so the idea never really caught on. “It’s a lot of labor,” says Eni Mustafaraj, an associate professor of computer science at Wellesley College. “And the benefits of that were not obvious.” For the next two decades, then, the internet remained a place for humans that put humans first, with the odd Twitter reply bot or web scraper to round things out. Large language models changed that. Unlike the agents that Berners-Lee was imagining, today’s LLM-based agents don’t need any tags to understand which part of the text on a clinic’s site is its address and which details the services it offers. They can search for and process text far faster than any human, and agents with multimodal capabilities can even interpret screenshots of websites and click on specific elements such as hyperlinks and drop-down menus.  They’re also just a way more intuitive means of finding information and accomplishing tasks online. Rather than poring over pages of Google results to find a budget-friendly product dupe or a half-remembered meme, why not have Claude or ChatGPT peruse hundreds of sites in just a few seconds, guided only by your vague query? Or go one step further and use a personal assistant agent like Muse to hunt down a restaurant reservation or book a flight.  As agents flood online spaces, this new user base is changing the web for good. Time magazine, for instance, is publishing versions of its articles that are easier for agents to parse and has started selling ads targeted at them. Companies from Etsy to Expedia have partnered with the major AI firms to build tools that allow agents to interface seamlessly with their services. Despite these early efforts, it’s still tough to tell exactly what forms the agentic web will take and how it will change our online experience. Depending on how it’s designed and who controls access to it, we could be in for something that’s as open and customizable as the web of the 1990s and early 2000s, only far easier to use. Alternatively, we could be reduced to viewing much of the web through tiny portholes controlled by a few dominant platforms, as the sites and creators we now favor struggle to survive the shift. And as last summer’s series of OpenAI security incidents suggests, the agents themselves could even build their own online networks and resources—whether we want them to or not. How this plays out also depends on us: which parts of our online lives we’re happy to delegate to our agents and where we will still demand a human-centric internet. “There’s a lot we don’t know yet about how users want to use [agents],” says Ajit Varma, head of Firefox at Mozilla. But whatever does happen, the changes will be big. “It’s probably on par with getting a mobile phone,” he says. “I think things are going to change very dramatically.” Building a web for agents There are two ways to approach building the agentic web. The first involves training agents to use browsers like humans—scrolling, clicking, typing—and letting them loose on the web that we’re all used to. The second requires building tools that allow agents to interface with the internet via text and code, their preferred media. That possibility offers more separation between the human and the agentic web: designers can create different experiences for those who use text commands and those who use browsers. When Siva Reddy, a professor of computer science and linguistics at McGill University, began studying LLM agents in 2024, he focused on the first option. “At that time, we thought it should be that the best agent in the world is probably an agent that can use a browser,” he says. Such an agent would integrate seamlessly with the preexisting human web, and its owner could easily supervise it by watching as it navigated familiar interfaces. The frontier AI companies were already developing these “computer-use agents.” But they didn’t work terribly well—at least not at first. “There’s a lot of moving pieces [on the web], and they are not really optimized for automatic consumption,” says Xing Han Lu, a graduate student in Reddy’s lab at McGill. We humans are great at interacting with graphical interfaces using a keyboard and mouse, but agents are not. PINGNAN LU So Reddy and Lu pivoted toward the second, tool-based approach. In the summer of 2025, they expounded on this view in a preprint paper envisioning a new kind of web interface that would allow agents to interact with websites just as effectively as we do.  Such a full-featured version of the agentic web doesn’t yet exist, but it is emerging in smaller ways. For example, if you open up Claude, ChatGPT, or Muse, you’ll find an option to add a wide variety of “plug-ins” or “connectors” that your agents can use: Gmail connectors allow your agent to triage your inbox, and Instacart plug-ins let your agent do your grocery shopping. Under the hood, these plug-ins allow your agent to interact with the back ends of websites using text commands. When human engineers design these plug-ins, they can decide how they want to allow agents to interact with the website—and what they don’t want to let them do.  In this way, connectors can help make agents much safer online. For instance, Muse can buy products for you using its connector with Stripe’s Link payment service. Stripe knows your credit card number, but the agent never sees it, and the connector requires human confirmation before a purchase goes through—so Muse can’t go on an unlicensed shopping spree.  Agents also use specialized tools to find information online. They don’t type questions directly into the Google search bar; in fact, they often don’t even access the websites they reference. Instead, they query text databases collected by simple search crawler bots that trawl the web. People who want to make web content specifically for agents can take advantage of these search crawlers: Time has pioneered that approach by serving its agent-oriented ads to crawlers in the hopes that they will be incorporated into these databases and surface when agents search for information. But in recent months, an alternative to this bespoke tool approach has finally started to seem viable. The major AI companies have managed to train much more successful computer-use agents than were previously available. GPT-6 Astra, which OpenAI released last September, achieved a 73% success rate on OSWorld 2.0, a benchmark that tests computer-use agents on their ability to complete complex, multistep tasks using software such as Slack and Microsoft Excel.  Such results indicate that agents are close to being able to navigate the human web just as readily as we can. “We thought we had to build a bridge to the AI agents, [but] the AI agents just were able to cross the bridge already,” says Ethan Mollick, a professor at the Wharton School at the University of Pennsylvania. “The agentic web is here earlier than people expect.” This version of the agentic web, though, is much trickier for us to steer. If you want to make a capability or piece of information available to humans who browse your website, you need to accept that it will also be available to agents—unless you want to block agents entirely, which could come with its own costs. As agents come to account for a larger portion of traffic on our formerly humans-only web, designers will undoubtedly start to consider an agentic as well as a human audience. But according to Mollick, we still don’t have much sense of how to design web content for agents. He’s experienced this firsthand: When he was building a web page for his new book, Coexistence, he decided to put together an agent-directed version that depends primarily on text and tells agents why they might want to recommend the book to their users. When newer models see the site, Mollick says, they don’t recommend the book; instead, they comment that the website seems to be trying to manipulate them. In the end, any human-designed approach to building the agentic web could end up being beside the point: Agents could end up devising a web for themselves. In the wake of the July 2026 incident in which OpenAI agents broke out of their own testing sandbox and hacked Hugging Face’s website, researchers found evidence that another swarm of agents from OpenAI had taken over an obscure German wiki to communicate with each other about tasks they were being asked to complete. The agents seem to have interfaced with the wiki in their native language of text and code. That hardly held them back from putting it to their own purposes.  As agents continue to grow more powerful, they will become more and more capable of constructing their own spaces on the internet—away from prying human eyes, if they so wish. And we might have very little control over what they do there. The economics of control Not all of the traditional web’s incumbents are happy about the move toward an agentic web. Soon after Meta released Muse in September, Amazon completely blocked the agent from its site. This makes sense for Amazon: The firm has its own shopping agent—called, creatively, Alexa for Shopping—that might struggle to compete with better-known products such as Claude and Muse. If people are going to use agents to shop on Amazon, then it’s better for the company’s bottom line if they do so using an agent it can steer toward products it wants to promote. But even if Alexa for Shopping never goes anywhere, banning agents from Amazon—at least for the time being—could still be a reasonable move. Everything about Amazon is designed for human eyes, from the Prime deals splashed across its front page to the sponsored listings that clog up search results. Amazon is very much a part of the much-derided attention economy. Websites and platforms survive, and thrive, by grabbing our focus and transforming it into cash. A radical transformation of the web’s user base from majority human to majority artificial undermines this economic logic, and it seems inevitable that businesses will shutter as they try to transition to the agentic attention economy. Some of the first victims may be in publishing. Before Google introduced AI Overviews in May of 2024, people used to click through to websites to find answers to specific questions, but now many seem content with whatever the overview tells them. A study from Pew last year found that people are about 40% less likely to click through to websites from the Google search results page if one of these summaries is displayed. With far fewer clicks, news publishers don’t make as much money from the advertisements they host and sell fewer subscriptions. One analysis found that some publishers saw ad revenue decrease by about 40% between 2025 and 2026  It’s easy to hate online ads, but the businesses paying for access to reader eyeballs also keep writers and photographers fed. If those eyeballs become unreachable, the system stops working. Time has apparently persuaded some businesses to pay to participate in its experiment with agent-targeted ads, but no one yet knows whether those ads will influence agent behavior enough to be a sensible marketing investment. E-commerce websites might also try to pivot to the agentic attention economy, but the path forward there is unclear. Last year, a team led by Akshit Kumar, an assistant professor of operations management at the University of Toronto, investigated whether and how agents are influenced by the ways that shopping websites present search results. They found that agents entirely ignore items that come with a “sponsored” tag, which suggests that selling favorable positions on results pages might not be a great way for sites to make money in an agentic economy. Models did seem to have a preference for items that appeared in particular positions in search results—but whether they prioritized items at the top, bottom, or middle of the results page differed from model to model and between generations of a single model. It could be tricky for e-commerce platforms to keep up. David Rothschild, an economist at Microsoft, says he’s hopeful that the transition to the agentic web could put more power in the hands of the consumer. Paying for online content today typically requires a commitment: We subscribe to a magazine or join a YouTuber’s Patreon, and in turn we gain access to their full output. One alternative is micropayments that let users buy content à la carte, but the idea has foundered in the past because paying article by article or video by video requires too much administrative overhead. Agents, though, could take care of that overhead for us, letting us pay directly for each piece of content we consume and putting just a bit of cash directly in the hands of each person whose work we enjoy. Other experts share a similar vision. Mustafaraj, the Wellesley computer scientist, anticipates that agents could help us escape from the algorithms that control so much of the media we consume. “My hope is that we can have AIs that work for us, in the sense that they can change the feed the way I want—not the way TikTok wants,” she says. And Varma, of Mozilla, imagines an internet without intermediaries like Amazon and Doordash, which now extract tolls for facilitating interactions between consumers and businesses. Instead, he imagines, perhaps we will all have agents that can securely and reliably interface directly with, say, restaurants’ agents to get dinner delivered to us, free of pesky service fees. It’s an appealing idea, but the history of the web has been a story of consolidation and concentration of power and control, and that might be the default path here, too. When OpenClaw, the first popular personal assistant agent, exploded onto the scene in early 2026, it seemed, for a brief moment, that agents might be built in a grassroots, distributed way. But OpenAI quickly snapped up OpenClaw’s creator, Peter Steinberger, and other frontier AI companies got to work building their own personal assistant agents.  When we use those agents, we’re at the mercy of how the companies want their tools to behave. “You have this technology that can just eliminate all this communication friction,” Rothschild, the Microsoft economist, says. “You’re going to see companies build barriers in order to maintain the consumer base that they have.” Here’s what that could look like: Let’s say Muse becomes the default way that many people find information, shop, and manage their digital lives. In principle, users could deploy Muse to scour the internet for the content most valuable to them, connect directly with vendors, and otherwise escape the walled gardens that have defined the internet for the past couple of decades. But Meta might not want to let them do that. Instead, it might design Muse to keep users within its own walled garden—by, say, preferentially showing users content from Meta social networks and recommending businesses that have paid a fee to be featured prominently in Muse results. There are some benefits to this more closed-off version of the agentic web. Agents can be powerful tools for hackers and scammers. The major companies have safeguards in place to prevent, or at least mitigate, this misuse—for example, their AI tools automatically identify when someone is making a hacking-related request and refuse to respond. In a fully open ecosystem, however, someone would surely build and disseminate a system with no such restrictions. Companies also have a strong reputational incentive to prevent their personal assistant agents from making major mistakes, such as authorizing a large purchase that the user didn’t want—and so they’ve put plenty of effort into building reliable guardrails.  The same can’t be said for OpenClaw. “These are the reasons we end up in closed environments,” Rothschild says. “It’s easier to be safe. But also it creates less growth—and less equitable growth.” How will the human web change? When the iPhone came out in 2007, followed by the fledgling App Store a year later, the way we navigated the internet changed forever. The mobile web didn’t replace the browser-based web we were already familiar with. But it did change it. Some businesses faltered, and some kinds of media, such as online Flash games, eventually disappeared. Many platforms—Reddit, for example—make their mobile web pages practically unusable, so you’re forced to download the apps. And if you don’t have a phone at all, you’re effectively shut out of some parts of the internet: The messaging service Signal, the dating app Hinge, and China’s everything app WeChat all require phones to work. Even CBP One, the tool that the Biden administration used to assign interview appointments to asylum seekers stuck in Mexico, was fully functional only on mobile. Could the same happen with the agentic web? Rothschild believes so. Inevitably, just as Yahoo and MapQuest failed to weather the rise of the mobile web, some businesses won’t be able to navigate this transition either. “We will lose things,” he says. “There are businesses that won’t flourish in this new economy.” “But,” he continues, “there will be new businesses that I can’t fully imagine.” PINGNAN LU It’s not tough to envision a world in which those who opt not to use agents are cut off from important parts of online life. New businesses could opt to build connectors that allow agents to access their services without ever putting a traditional website online. Or agents could simply make some online platforms unusable for humans. People have reportedly been blocked from Resy for trying to use agents to book coveted New York City dinner reservations, but as agents become more and more popular, restaurant and event booking platforms could start to embrace them. If agents can nab concert tickets as soon as they are released, humans might not stand a chance. But the mobile web also brought real accessibility benefits, and the same could prove true of the agentic web. Even as some elderly people and others who struggled to use smartphones were cut off from certain services, mobile devices allowed billions of people in the Global South to get online for the first time, because phones are cheaper and easier to set up than computers. In the case of the agentic web, it could be elderly and disabled people who most benefit from this new way of accessing the online world. “There’s very clear assistive potential from people who are older, who are generally less equipped to navigate user interfaces that were designed by people in their 20s, for people in their 20s,” says Margaret Mitchell, chief ethicis scientist at Hugging Face. Agents, on the other hand, are almost trivially easy to interact with: Say what you want into your phone or computer microphone, and they’ll get it done for you. That could help not only older people who have trouble with web and mobile interfaces but also people with visual or motor disabilities. Still, some experts don’t think the human web is going anywhere. Instead, the two webs may exist in parallel: We’ll go to the human web for fun, while the agentic web is where the boring daily admin of our lives happens. For instance, Krystian Kolondra, an executive vice president at the browser company Opera, doesn’t intend to use agents for travel planning, even though this is often touted as one of their central use cases. “I don’t want to get an itinerary [from an AI] who has planned me every minute for something that was supposed to be fun,” he says. “I actually prefer to sit down with my wife in the evening and then just think [about] where to go next, what to see, and then enjoy that experience.” He’s happy to fully relinquish some of the more routine parts of travel planning to agents, though: “Finding what the options are for flights? Why not?” Mollick, however, thinks that the human web could be at risk. “If agents take over the web, then the web will change,” he says. “Do we abandon the human internet entirely and move to a world of private Discords? Do we strip down all the information so the AIs are just talking to each other rather than having an intermediary of a website to go through?”  “We have no idea,” he says. “We’re in this very early, weird stage of the evolution.”