Imagine your browser doing the heavy lifting for you! Google is taking a significant leap forward today by rolling out a powerful new feature in its Chrome browser: an autonomous browsing agent powered by Gemini, aptly named Auto Browse. This isn't just a minor update; it's a fundamental shift in how you might interact with the web, designed to automate those tedious, repetitive tasks that eat up your time.
For months, Google has been integrating its Gemini AI into Chrome, and this latest expansion is particularly noteworthy. While the chatbot is becoming more accessible and connected to a wider array of Google services, the real game-changer is this Auto Browse agent. Think of it as your personal digital assistant, capable of navigating the web and performing actions on your behalf, much like other experimental tools we've seen emerge. The core promise? Freedom from mundane online chores.
But here's where it gets interesting... The new Gemini features are accessed through the familiar AI button that has appeared in Chrome. Previously, this button would simply open Gemini in a pop-up. Now, Google is defaulting to a split-screen or "Sidepanel" view. This provides Gemini with more space to operate while it's actively manipulating web pages, and it’s a welcome change for those who want a more integrated AI experience. This update has already begun rolling out, so you might already have it at your fingertips!
While you can still opt for a floating window, the split-view is particularly useful when you want to leverage Gemini's integration with other Google applications. The AI can now directly access and interact with Gmail, Calendar, YouTube, Maps, Google Shopping, and Google Flights, all within your Chrome window. Google technically introduced this connectivity in mid-January but is only now highlighting its capabilities.
And this is the part most people miss... Gemini in Chrome can now also edit images directly using a tool called Nano Banana. This means you no longer need to download an image, edit it elsewhere, and then re-upload it. Simply open an image from the web, and in the Sidepanel, describe the edits you want. Similar to the standalone Gemini app, you can choose between the slower, higher-quality Pro model or the faster, standard model for your image manipulation tasks.
Can’t someone else do it? The star of today's announcement is undoubtedly Chrome's new browsing agent, which is currently in preview. The allure of agentic AI is the prospect of being able to delegate tasks and reclaim your time. Instead of manually filling out forms or painstakingly copying information from emails, you can simply instruct an AI agent to handle it. However, the reality of many current AI agents has been that they are often sluggish and unreliable, requiring constant human oversight – which rather defeats the purpose. Now, Google is stepping up to demonstrate the potential of its own browsing agent.
Google states that Auto Browse is built upon its advanced Gemini 3 models, benefiting from insights gained from the experimental Project Mariner agent. Essentially, if a task can be performed using a keyboard and mouse within your browser, Auto Browse is theoretically capable of taking over. It can access your browser's content and open tabs, and importantly, it will request permission before accessing sensitive data like your passwords, which is a crucial security measure.
Apartment hunting with Auto Browse: When you initiate an Auto Browse task, Chrome will open new tabs as needed, clearly marked with a distinct AI icon to indicate where the agent is active. You don't need to keep these tabs in the foreground, and you can even run multiple AI tasks concurrently. The AI will notify you when a task is complete or if it requires your input. However, there are usage limits for Auto Browse, mirroring those for some of Google's more demanding chatbot functions. AI Pro subscribers are allowed 20 AI browsing tasks per day, while AI Ultra subscribers get a generous 200 tasks daily.
The idea of an AI churning away in the background might raise some eyebrows. After all, generative AI is known to make mistakes, and granting it control over your browsing experience is a significant trust exercise. Google assures users that it has implemented a comprehensive suite of security and safety protocols to prevent misuse. For instance, if you instruct Auto Browse to research and purchase an item, it will not complete the transaction on its own. Instead, it will locate the item and advance to the checkout screen, leaving the final purchase decision to you.
Filling a form with Auto Browse: It's also important to understand that Auto Browse does not operate locally. All content from the tabs managed by the AI is streamed to a cloud-based Gemini model. This implies that your page content will be shared with Google, though the precise extent of this sharing isn't entirely transparent. Google states that Auto Browse adheres to the existing Gemini in Chrome policy, which indicates that website information is stored in your Gemini Apps Activity (if Activity tracking is enabled). Page content is also temporarily logged to your Google Account. When asked whether Google would use page content processed by Auto Browse for further AI model training, a spokesperson declined to offer specific details.
Chrome's Auto Browse is launching today in preview, but it's not entirely without its prerequisites. While it doesn't incur an additional cost, the feature is exclusively available to AI Pro and AI Ultra subscribers. Google's wording suggests that the feature might eventually be extended to free users after the preview period, but it's reasonable to expect stricter usage limitations if that happens.
Now, let's talk about what this means for you. Is this the dawn of truly effortless web navigation, or are the current limitations and privacy concerns too significant to ignore? What are your thoughts on giving an AI agent control over your browsing? Let us know in the comments below – we'd love to hear your agreement or disagreement!
