Google Labs has introduced Disco, an experimental browsing experience designed to test what the web could look like if AI didn’t just “assist” your browsing—but actively built tools for you on the fly.
At the center of Disco is GenTabs, a feature Google says is powered by Gemini 3. Instead of giving you another chatbot panel bolted onto a familiar browser layout, GenTabs scans what you have open—your current tabs (and related session context like chat history)—then generates an interactive mini-app tailored to what you’re trying to do.
It’s a noticeably different bet than the fast-growing crop of “AI browsers,” many of which are still fundamentally traditional browsers with AI layered on top.
GenTabs turns your tabs into task-specific apps
Disco’s pitch is simple: too much modern browsing is really “tab management.” You open 20 pages while planning a trip, researching a topic, or comparing products—then spend half your time juggling sources instead of making decisions.
GenTabs aims to compress that chaos into something usable. In Google’s examples, vacation research can become a trip planner with maps and itineraries. A dense subject can turn into a visualization or study aid. Meal planning can become a recipe organizer with a shopping list. The point isn’t just summarization—it’s tool creation, grounded in the sources you already have open.
You can then refine the generated app in plain language, and Disco keeps the work tied back to the underlying pages so you can trace where information came from.
Why this is different from “chat-in-the-browser”
A lot of the current AI-browser wave still looks like “normal browsing + AI.”
OpenAI’s ChatGPT Atlas, for example, is positioned as a Chrome challenger, but it leans heavily on the same modern browser foundation: Chromium, the open-source project led by Google with contributions from across the industry. Atlas adds workflow features—like right-click actions and an agent mode meant to help complete tasks while you browse—but it’s still recognizable as a browser first.
Microsoft’s Edge has also pushed beyond a simple chatbot sidebar with deeper integration like opt-in Copilot Mode, which can work across your open tabs.
And Perplexity’s Comet has been framed as an “AI-native” browser with agent-style behavior—not merely a chat window glued to the side.
Disco’s approach is more radical: it treats AI-generated applications as first-class browser elements, not an add-on.
A strategic rollout: macOS first, waitlist only
For now, Disco is limited. Google is taking waitlist signups for macOS users and describing the project as a “discovery vehicle”—a place to test ideas that might later influence Chrome or other Google products.
That wording matters. Google Labs experiments don’t always graduate to mainstream releases. But the company clearly wants to explore a future where browsing isn’t just retrieving information—it’s generating the tools you need to use it.
The real browser war is about the interface, not the engine
One of the more interesting dynamics here is that many would-be Chrome challengers are building on Chromium. That speeds development and compatibility, but it can also steer innovation toward features that fit comfortably within today’s browser shape.
Google, by contrast, has the leverage to rethink the experience from the inside: not just adding AI, but redefining what “a tab” or a “page” is supposed to do when a model can dynamically assemble interfaces, workflows, and summaries in real time.
Whether GenTabs becomes a breakout idea—or stays a clever Labs demo—will depend on how well it handles the messy reality of the web: conflicting sources, incomplete context, and users who want control as much as convenience.
But Disco makes one thing clear: the next phase of browser competition won’t be won by who has the best chatbot. It’ll be won by who redesigns the browser around how people actually use the internet now.