Google Tests Remy: Gemini's AI for Enhanced Control

Google is experimenting with a new artificial intelligence agent named Remy, intended to integrate with Gemini, its digital assistant application. According to a report from Business Insider, this tool is designed to assist users with their professional and personal tasks by proactively acting on their behalf.

Remy is described in an internal document as a "24/7 personal agent," aiming to transform Gemini into an assistant capable of acting on behalf of the user. Currently, Remy is in an internal testing phase, reserved for Google employees. The report notes that it reviewed an internal document and spoke with two individuals familiar with the matter. However, it does not specify when, or if, Google plans to make Remy public.

Remy stands out for its ability to integrate with Google services such as Google Workspace, as well as other platforms like GitHub, Spotify, YouTube Music, Google Photos, WhatsApp, Google Home, and Android utilities. It is designed to handle complex tasks and learn user preferences while ensuring increased user control through the Gemini Privacy Hub.

This Privacy Hub allows users to monitor and manage the data collected by Remy, modify automatic deletion settings, and control access to connected applications. Google emphasizes transparency and the limitation of the agent's powers, in accordance with recommendations from Google Research and Google Cloud. Google Cloud stated that agent activities should be transparent and audited through logs, using the principle of least privilege to limit the powers of AI agents.

The report compared the concept of Remy to OpenClaw, an AI agent that garnered attention earlier this year for its ability to autonomously respond to messages, conduct searches on behalf of users, and take autonomous actions. The project is currently in a "dogfooding" phase, where employees test the product before a potential wider deployment.

Although the report does not provide technical details about Remy's architecture, it highlights the importance of human oversight and memory controls to ensure the safe and ethical use of AI. Users can manage the information they have asked Gemini to save, and the Privacy Hub covers customization controls based on past discussions and Personal Intelligence.