Google turned Gemini into a personal assistant capable of making calls, shopping, and booking appointments

Google turned Gemini into a personal assistant capable of making calls, shopping, and booking appointments

The new feature allows you to call businesses, book appointments, log in to websites, and make purchases using Google Pay

Share:

Google expanded Gemini's capabilities with Tasks, a feature that transforms the chatbot into an artificial intelligence agent capable of performing everyday tasks on behalf of users. Unlike the traditional mode, which answers questions and provides information, this tool can take concrete actions, communicate with businesses, manage purchases, and organize appointments without the person having to intervene at every step.

The announcement was made on Tuesday, November 17, by Josh Woodward, vice president of Gemini, in a post on Google's official blog. The company began distributing the tool for free to hundreds of millions of users in the United States, Japan, South Korea, Canada, Australia, and India, through its new Gemini 4 Hydrogen model.

From chatbot to personal assistant: what Tasks can do

One of the main innovations is the ability to delegate tasks that typically require time and multiple interactions. Gemini can call businesses, check availability, and book appointments on behalf of its users.

Woodward shared a personal experience to demonstrate its functionality. When he detected a mechanical problem in his family's minivan, he asked the assistant to call two nearby dealerships and secure an appointment to check the vehicle on a Friday morning.

The system identified the businesses using Google Maps, made both calls simultaneously, waited on hold, and compared the available options. Finally, it found a time that fit his schedule and automatically added the appointment.

The artificial intelligence also informed the businesses that it was an automated system, while Google clarified that merchants can decline this type of interaction.

Another application involves analyzing bills and detecting unusual expenses. Woodward mentioned that he used Tasks to investigate a $389 electric bill. After authorizing access to his provider's portal and reviewing previous consumption, Gemini identified that a heat pump was operating during peak energy cost hours.

Online shopping and payments via Google Pay

The tool can also compare products, search for discounts, and complete purchases using Google Pay. The user can set a maximum budget in advance to prevent the agent from exceeding the authorized amount.

Woodward recounted that he used this feature to find a USB-C charger. Gemini compared alternatives, inquired about preferences regarding power and color, identified a promotional code, and facilitated the purchase, saving an additional ten dollars.

To process transactions, Google incorporated the Agent Payments Protocol (AP2), a system that uses digital credentials to protect payment data.

The company is working with Mastercard and Visa, as well as American Express, whose integration is planned for a later stage. Before confirming a purchase, Gemini must request explicit authorization from the user, who can review the amount and terms of the transaction.

Receipts and tracking information are also incorporated into Google Wallet. This way, the system aims to integrate product search, selection, and payment within a single experience.

Google turned Gemini into a personal assistant capable of making calls, shopping, and booking appointments

Access to accounts and task automation

Another important capability is the ability to log into certain websites using Google Password Manager. This allows users to check service information, review bills, and manage accounts without having to manually enter credentials each time.

Gemini can also organize emails, identify commitments, and transfer important dates to the calendar. Woodward explained that he forwarded school messages to the address gemini@google.com, and the system identified the events, added them to his agenda, and detected a scheduling conflict.

Additionally, the company reported that Gemini can now connect with over 60 popular applications, including shopping and personal organization services.

Security, privacy, and availability

The increase in autonomy raises questions about access to private information and the possibility of an artificial intelligence performing operations without authorization.

To mitigate these risks, Google uses Agent Compute Engine, an infrastructure that executes tasks on hardware-isolated virtual machines dedicated to each account. It also incorporates filters against malicious instructions present on web pages and search results.

Users can intervene while Gemini is working, modify instructions, and disconnect applications when they deem necessary. If a page does not allow completing a transaction automatically, the system can return control to the person to perform the final step.

The free distribution of Tasks began in six countries, while advanced features, such as business calls and certain purchasing operations, have an initial rollout in the United States and additional geographic restrictions.

To use it, simply open Gemini and switch the selector on the main screen from Chat to Task. Options have also been enabled to send instructions via text messages and emails in compatible markets.

With this launch, Google aims to transform Gemini from a chatbot that answers queries into an assistant capable of performing real tasks, an evolution that intensifies competition among major tech companies to develop autonomous agents that can manage everyday activities.