Google launches Disco, a Gemini 3-powered experiment that suggests and creates GenTabs, web apps based on the user's open browser tabs and Gemini chat history
Context & Ripple Effects
Disco extends Google’s Gemini work from answering prompts toward producing interactive interfaces: Google had already described generative UI that Gemini can build for a prompt, while Gemini’s arrival in Chrome across platforms creates a closer browser-level entry point for such features.
The experiment also sits beside Google’s effort to make user-created AI tools reusable through mini apps built with Opal and saved as Gems. Disco applies that creation model to the immediate context of tabs and prior chats rather than a blank prompt.
First-order effects
- Users in the experiment can turn their active browsing and Gemini conversation context into suggested or generated GenTabs, making the browser session an input to app creation rather than just research material.
- Google gains a test bed for Gemini 3-powered generative interfaces in a browser-adjacent workflow, including whether tab and chat context yields useful apps without extensive manual setup.
Second-order effects
- Browser-based productivity and research tools face pressure to make their own context actionable, not merely searchable or summarizable, as Google tests AI-generated task interfaces from an existing session.
- The move increases the value of Google’s connected Gemini and Chrome surfaces: the more relevant context users keep within them, the more tailored the generated interface can become.
Third-order effects
- If these experiments prove reliable, software creation may shift further from selecting fixed applications toward assembling temporary, task-specific work surfaces—a broader Gemini Intelligence direction that includes automation across apps and widget creation.
- That shift would make context access, permissions, and the trustworthiness of generated actions central competitive boundaries for assistants, rather than model capability alone.
The trend: AI assistants are evolving into agentic work surfaces that use live user context to generate the interface and workflow for a task.