Are you ready to see a revolution in web browsing? Google is unveiling its innovative "auto browse" mode for Chrome, which utilizes artificial intelligence to automate tasks that users frequently perform online, such as making purchases or setting appointments.
In the competitive landscape of AI browsers, other major players like Microsoft and OpenAI have already introduced their own versions of agentic AI capabilities, allowing users to automate browser functions with ease. Google has been integrating its AI assistant, Gemini, into Chrome for some time now, even adding a dedicated button for quick access to the chatbot. With this new feature, Google aims for the AI to manage more intricate workflows that involve multiple steps.
According to a blog post from Google, testers have reported using auto browse mode for a variety of practical applications. These include scheduling appointments, filling out laborious online forms, gathering tax documents, getting quotes from service providers like plumbers and electricians, checking bill payments, filing expense reports, managing subscriptions, and even speeding up the process of renewing driving licenses—saving users a significant amount of time in the process.
In one demonstration, a user activates the auto browse feature by instructing Gemini to purchase decorations from Etsy that match an image found in their Gmail inbox. The AI adeptly navigates to Etsy and successfully adds the selected items to the user’s shopping cart. Notably, it even personalizes one of the decorations by inscribing the phrase "Y2K Party" before summarizing the items in the cart and guiding the user through the checkout process. In another showcase, the auto browse feature efficiently extracts data from a document to complete an online form.
However, the auto browse function is not without its limitations. During a demo conducted for journalists, the feature was asked to visit a website and reorder a previously purchased jacket in the same size while also searching for discounts. The AI first sought permission to access the user's saved login credentials in Chrome, but it struggled to process the entire request promptly. This delay mirrors challenges encountered with other AI browser integrations.
For critical actions such as making a purchase or posting on social media, Google designed the auto browse feature to require user confirmation before proceeding to finalize these steps. While this measure can help prevent mistakes, it also somewhat diminishes the convenience that users might expect from an automated system.
We are eager to test this feature ourselves, hoping it represents a significant improvement over Google's earlier Project Mariner, which was an experimental AI browser that often had difficulties understanding commands and interacting effectively with third-party websites, particularly when faced with CAPTCHA challenges.
At this moment, Google is merely offering a preview of the auto browse mode for Chrome. Initially, it will be accessible to paid subscribers of Google AI Pro and Ultra in the United States, with monthly fees set at $20 and $250, respectively. Additionally, there is a limitation on the number of requests users can make each day: Pro users can execute 20 requests, while Ultra users can complete up to 200. It is possible for users to handle multiple tasks simultaneously.
The company appears to be deliberately delaying a broader rollout to refine the feature and address potential issues. A significant concern surrounding AI browsers is their propensity to generate inaccurate information and fall victim to "prompt injection attacks," which can manipulate the AI into performing unintended actions. Nonetheless, Google is actively developing "layered defenses" to mitigate these risks.
While it remains uncertain when—if ever—free users will gain access to auto browse mode, Google is laying the groundwork for this feature, ensuring that Gemini will eventually interact with third-party platforms. The company has announced that Chrome will support the Universal Commerce Protocol (UCP), a collaborative effort with industry leaders like Etsy, Shopify, Target, and Wayfair. This new open standard is designed to enable AI agents to seamlessly understand and engage with various websites.
What do you think about the future of automated browsing? Is the convenience worth the potential risks associated with AI technology? We invite you to share your thoughts and opinions in the comments!
