Google has announced a significant update to Gemini in Chrome for macOS, Windows, and Chromebook Plus, introducing a new side panel experience, deeper integrations with Google apps, and an agentic browsing feature called auto browse. The update is built on Gemini 3 and represents the company's most comprehensive push yet to embed AI assistance directly into the browsing experience.

A Persistent Assistant in the Side Panel

The centerpiece of the update is a redesigned side panel that keeps Gemini accessible regardless of which tab the user is on. The feature is designed to enable multitasking without interrupting the primary workflow — users can keep a document or webpage open on one tab while delegating a separate task to the assistant on the side. Reported use cases from Google's internal testers include comparing products across multiple tabs, summarizing reviews, and managing calendar events.

Nano Banana Brings Image Transformation to the Browser

Google is also integrating Nano Banana into Chrome, enabling users to transform images directly within the browser without downloading, editing, or re-uploading files. The feature is accessible through the side panel and supports both creative and practical applications, such as redesigning interior spaces or converting raw data into visual infographics.

Connected Apps and Cross-Platform Workflows

Gemini in Chrome now supports integrations with Gmail, Google Calendar, YouTube, Maps, Google Shopping, and Google Flights under a Connected Apps framework. The integrations are designed to handle multi-step workflows across platforms — for example, retrieving event details from an email, cross-referencing flight options, and drafting follow-up communications, all within a single session. These features can be enabled or disabled through Gemini Settings.

Personal Intelligence: Context-Aware Browsing

Google has confirmed that Personal Intelligence — a feature already present in the Gemini mobile app — will be coming to Chrome in the coming months. The feature allows Chrome to retain context from past conversations and deliver more tailored responses over time. Users will be able to opt in, connect or disconnect apps at will, and provide specific instructions to further personalize the assistant's behavior.

Fonte: Generated by ABN – Copyright free

Auto Browse: Agentic Action for Complex Tasks

For AI Pro and Ultra subscribers in the United States, Google is introducing auto browse, an agentic feature capable of executing multi-step tasks autonomously. The system can research travel options across multiple dates, fill out online forms, manage subscriptions, collect tax documents, and handle appointment scheduling, among other tasks. In more advanced scenarios, auto browse can use Gemini 3's multimodal capabilities to identify objects in images, search for similar products, and add them to a cart while applying discount codes and respecting a set budget.

For tasks that require authentication, auto browse can leverage Google Password Manager with user permission. Actions deemed sensitive — such as purchases or social media posts — are designed to pause and request explicit user confirmation before proceeding.

Universal Commerce Protocol: A New Open Standard

Chrome will also support Google's Universal Commerce Protocol (UCP), a newly introduced open standard for agentic commerce developed in partnership with Shopify, Etsy, Wayfair, and Target. UCP is designed to ensure that AI agents can execute commercial actions on a user's behalf seamlessly across participating platforms.

Analysis

The update positions Gemini in Chrome as Google's most direct response to the growing competition in AI-assisted browsing, placing it alongside offerings from Microsoft Copilot in Edge and emerging standalone AI browsers. What distinguishes Google's approach is the depth of its ecosystem integration — by connecting Gemini directly to Gmail, Calendar, Flights, and Shopping, the company is leveraging its existing infrastructure in a way that standalone competitors cannot easily replicate.

The introduction of auto browse, however, marks the most consequential shift. Moving from passive assistance to autonomous, multi-step action represents a fundamental change in how browsers interact with the web — and with users. Google's decision to limit auto browse to paid subscribers in the U.S. initially suggests a cautious rollout strategy, likely driven by the need to monitor agentic behavior in real-world conditions before broader deployment.

The Universal Commerce Protocol is also worth noting as a strategic move. By co-developing an open standard with major e-commerce platforms, Google is attempting to establish the infrastructure layer of agentic commerce before the category fully matures — a pattern consistent with the company's historical approach to web standards.

Whether users will embrace autonomous browsing at scale remains to be seen, but the architecture Google is building suggests a long-term bet that the browser will evolve from a navigation tool into an active participant in digital tasks.