Google has introduced its Gemini in Chrome artificial intelligence features to Android users in India, marking an expansion of AI-powered assistance directly within the Chrome browser. This update allows the integration of AI capabilities to enhance web browsing experiences by summarizing webpages, connecting information across Google services, and performing various online tasks.
Gemini in Chrome functions as an AI assistant designed for the browser environment. On Android devices, users can benefit from features such as webpage summaries, integration with services like Gmail and Calendar, and the ability to generate images. For Google AI Pro and Google AI Ultra subscribers in India, a more advanced auto-browse feature is also being rolled out for both desktop and Android platforms.
The auto-browse feature enables Gemini to navigate websites and execute certain tasks on behalf of users. It is capable of managing complex, multi-step activities that typically require users to interact with multiple webpages and complete various actions. This includes tasks such as comparing products, finding deals, booking event tickets, making restaurant reservations, and scheduling appointments.
Given the potential for AI agents to interact directly with websites, Google has implemented safeguards to mitigate risks. Gemini is equipped to detect certain threats, such as prompt-injection attempts, and users might be required to confirm sensitive actions before they are executed. These measures are crucial when the AI system is capable of making changes, submitting forms, or conducting transactions on behalf of the user.
The number of multi-step tasks that users can perform through auto-browse is contingent on their Google AI subscription plan. Google AI Pro subscribers are allowed up to 20 multi-step requests per day, whereas Google AI Ultra subscribers have a limit of 200 requests per day. This expansion signifies Google’s commitment to enhancing agentic AI capabilities within Chrome, enabling users to go beyond merely asking AI questions and instead utilize it for performing tasks across the web.