Google is re-engineering how its Chrome-based AI Mode handles external data. The latest update introduces a side-panel view for linked sources, a design choice that keeps the conversational interface active while rendering webpage content. This architectural decision moves away from the disruptive tab-switching model, creating a persistent context window for the language model. The system can now process live page content for follow-up queries, a technical step toward more grounded, real-time inference.

Separately, Google has extended the feature's context retrieval capabilities to browser tabs. Users can now point the model at specific open tabs as source material, bypassing manual link pasting. This integration leverages the browser's own tab management APIs, suggesting a deeper OS-level hook than previous web-only implementations. The engineering trade-off is clear: increased utility and context specificity against the inherent complexity and potential latency of scanning live document object models.

These updates, currently rolling out in the US, represent a tangible move toward making the browser itself a primary orchestration layer for AI agents. The focus is shifting from standalone chat to a system designed for continuous, multi-source interaction.

Source: The Verge