Google is testing a new kind of browser that treats every messy cluster of tabs as raw material for a tailored mini app, rather than a pile of half-finished tasks. Instead of forcing users to juggle dozens of pages, the experimental project, called Disco, uses artificial intelligence to pull information together into focused, interactive views that look less like a traditional browser and more like a purpose-built tool.
The idea is simple but ambitious: let the browser understand what a user is trying to do, then automatically assemble the right controls, summaries, and actions from the tabs and chats already open. If it works at scale, the familiar tab strip could start to feel like a legacy interface, replaced by dynamic workspaces that adapt to each task.
Disco and GenTabs, Google’s experimental rethink of the browser
Google is positioning Disco as an experiment rather than a finished product, but the concept is unusually aggressive in how it reimagines the browser’s core behavior. Instead of treating each tab as an isolated webpage, Disco introduces a feature called GenTabs that compiles information a user has already provided into what Google describes as a custom, interactive experience. In practice, that means the browser is not just a window onto sites, it is an active participant that reorganizes content into a single interface tailored to the task at hand, a shift that moves browsing closer to app-like workflows than static pages, as early descriptions of Disco and GenTabs make clear.
The company is explicit that Disco is an AI browser, not just a cosmetic refresh of Chrome or a new skin on existing engines. GenTabs is designed to sit at the center of that experience, scanning what a person is doing across their open tabs and chat history, then turning that context into a focused mini app that can be manipulated directly. That framing, which appears across Google’s own descriptions and early coverage of Google’s Disco experiment, signals that the company sees Disco less as a side project and more as a test bed for how AI could reshape the browser itself.
How GenTabs turns chaotic tab piles into mini apps
The core promise of GenTabs is to transform the familiar sprawl of tabs into a coherent, interactive application that understands what a user is trying to accomplish. Instead of forcing someone to click back and forth between a spreadsheet, a product page, a review site, and a note-taking app, GenTabs analyzes the content of those tabs and the related chat, then builds a single interface that surfaces the relevant details and tools. Reporting on the feature describes GenTabs as processing everything a person does through their open tabs and chat history in order to automatically create interactive apps for that task, a description that underscores how deeply the system is meant to integrate with ongoing activity inside the browser, according to early technical breakdowns of GenTabs behavior.
That design is not just about convenience, it is a direct response to the way modern browsing has evolved into a kind of ad hoc operating system. People already treat their tab bars as to-do lists, research archives, and temporary dashboards, but the browser itself offers almost no help in organizing that chaos. GenTabs is explicitly described as a feature that turns tabs into interactive web apps, with Google explaining that it is designed to gather the pieces of a task that are scattered across multiple sites and bring them together in one clear application, a goal that is central to the way Google presents GenTabs inside Disco.
What Disco actually looks like in use
On the surface, Disco still resembles a browser, but the way it behaves once GenTabs is active is closer to a dynamic workspace that morphs around each project. When a user is researching a vacation, for instance, the system can pull hotel options, flight searches, maps, and notes into a single view that behaves like a travel planning app, rather than a loose collection of unrelated pages. That shift from passive pages to active interfaces is at the heart of how Google describes Disco as a new type of browser that lets people turn their tabs into custom apps to complete the task they are working on across tabs, a framing that runs through early hands-on descriptions of Google’s experiment.
In practice, that means the browser is constantly scanning for patterns in what a person is doing, then offering to consolidate those patterns into a GenTab that behaves like a mini app. If someone is comparing 2025 Toyota RAV4 and Honda CR-V models across several car sites while chatting with a friend about financing, Disco can theoretically assemble a comparison interface that lists prices, trims, and key specs side by side, with quick links back to the underlying pages. The company’s own positioning of Disco as an experimental new browser based on AI GenTabs, which it describes as a way to rethink how tabs function, reinforces the idea that the interface is meant to feel like a layer on top of the web rather than a simple window into it, as early coverage of Google Disco makes clear.
Gemini 3 and the AI engine behind Disco
Under the hood, Disco is powered by the same AI push that is reshaping Google’s broader product line, and GenTabs is explicitly tied to the company’s Gemini models. The star feature is described as GenTabs, powered by Gemini 3 AI, which scans a user’s open tabs and chat history to understand what they are doing and then builds interactive experiences on top of that context. That connection to Gemini 3 matters because it signals that Disco is not a one-off experiment with a bespoke engine, but rather a showcase for the company’s flagship AI stack, a point that is spelled out in early coverage of Gemini-powered GenTabs.
Gemini 3’s role is not limited to summarizing pages or answering questions in a sidebar. The model is used to infer tasks, group related content, and generate the interface elements that turn a set of tabs into something that feels like a standalone app. That could mean building a shopping list from grocery sites, creating an interactive solar system model from a cluster of astronomy resources, or assembling a study guide from multiple academic articles. The way Google ties Disco to Gemini 3, and the way it highlights those concrete examples of interactive outputs, underscores that the browser is meant to be a proving ground for how large models can orchestrate complex workflows directly inside the browsing experience, rather than just offering chat-style assistance.
Early access, waiting lists, and what “experimental” really means
For now, Disco is not a mainstream product, and Google is careful to label it as experimental in its own descriptions. The company is opening access through Google Labs, which is setting up a waiting list for people who want to try the browser and its GenTabs feature. That approach, where Google Labs opens a waiting list for Disco as an experiment that users can sign up for, signals that the company wants to gather feedback and iterate before deciding whether to fold the ideas into its primary products, a process that is spelled out in early notes on how Google Labs is handling Disco.
Access is not just limited, it is framed as a test bed for ideas that may or may not make it into Chrome or other mainstream offerings. Google labels Disco as experimental and explains that, initially, only a number of users will be able to try the browser while the company evaluates how its concepts perform in the real world. That positioning, which describes early access and limitations for Google’s Disco Browser and notes that experiments like this are often refined before they are advanced into mainstream products, makes clear that Disco is as much a research project as it is a consumer tool, according to the description of early access and limitations.
Why Google is chasing AI-native browsing now
Disco does not exist in a vacuum, it arrives at a moment when the browser itself is becoming a front line in what some observers describe as AI browser wars. Google faces intensifying pressure from competitors that are trying to rethink the basic browsing experience with AI at the center, rather than as an add-on. That context helps explain why the company is willing to experiment with a project that promises to end what it has described as the tyranny of the tab, where dozens of open pages represent half-finished tasks that never quite resolve into a coherent workflow, a dynamic that is highlighted in coverage of the AI browser wars.
At the same time, Disco is a direct answer to AI-native search and browsing tools that have emerged outside Google’s ecosystem. The company is described as answering rivals like Perplexity and OpenAI with an AI browser of its own, one that uses Gemini 3 to build interactive experiences from the same web content those tools summarize. By turning that content into mini apps rather than static answers, Google is betting that users will prefer a browser that can bundle a set of product pages into a shopping list or convert a cluster of recipe tabs into a structured meal plan. The way Disco is framed as a response to those competitors, and the way it leans on Gemini 3 to differentiate itself, shows how central AI-native browsing has become to Google’s strategy, as early analysis of Google’s answer to Perplexity and OpenAI makes clear.
From tab collectors to task-centric workspaces
For everyday users, the most immediate impact of Disco’s design is the way it reframes the browser from a collection of pages into a set of task-centric workspaces. Instead of leaving a dozen tabs open as a reminder to finish planning a birthday party, a user could let GenTabs assemble those pages into a single interface that tracks RSVPs, compares cake options, and lists decorations from different retailers. That shift aligns with Google’s description of GenTabs as a feature that turns tabs into interactive web apps, designed to gather the pieces of a task that are scattered across multiple sites and bring them together in one clear application, a goal that is central to the way GenTabs is positioned inside Disco.
For power users who already rely on tab managers, vertical tab bars, and extensions like OneTab or Workona, Disco’s approach could feel like a natural evolution. Instead of manually grouping tabs and saving sessions, the browser itself would infer which pages belong together and offer a mini app that sits on top of them. That is a more opinionated vision of browsing, one that assumes the software can correctly guess what a person is trying to do, but it is also a logical extension of how people already use browsers as makeshift project managers. By embedding that logic directly into the browser and tying it to Gemini 3, Google is effectively arguing that the era of the passive tab strip is ending, a point that is reinforced by its broader push to end the tyranny of the tab and replace it with AI-organized workspaces, as described in early analysis of Google’s new browser vision.
How Disco fits into Google’s broader AI strategy
Disco is also a strategic signal about where Google wants AI to live inside its ecosystem. Rather than confining Gemini 3 to chatbots or search results, the company is embedding it directly into the browsing layer, where it can see the full context of what a user is doing and act on that information in real time. That approach mirrors how Google has been weaving AI into Gmail, Docs, and Android, but Disco goes further by making AI the organizing principle of the browser itself, a direction that is underscored by the way the company describes Disco as an AI browser that turns tabs into mini apps through GenTabs, as detailed in early looks at Google’s AI browser experiment.
At the same time, keeping Disco inside Google Labs and labeling it experimental gives the company room to test how far users are willing to go with AI-driven interfaces that watch and reorganize their browsing. If people embrace GenTabs as a way to tame their tab overload, the underlying concepts could migrate into Chrome or other mainstream products. If they push back on the idea of a browser that constantly interprets their activity, Google can adjust the balance between automation and control before committing to a broader rollout. That measured approach, where experimental projects are refined and only the most successful ideas are advanced into mainstream products, is consistent with how the company describes early access and limitations for Google’s Disco Browser, and it suggests that Disco is as much a barometer of user comfort with AI-native browsing as it is a product in its own right.
More from MorningOverview