Gemini is useful, but it mostly just answers things. You ask it a question, it gives you a response, and then you go do whatever needs doing yourself. Google has apparently been working on something with a lot more ambition. According to a Business Insider report, the company is internally testing a Google Remy AI agent, a Gemini-powered tool designed to actually take actions on your behalf, not just generate responses.
Two people familiar with the project told Business Insider that Google employees are currently testing Remy in a staff-only version of the Gemini app. An internal document describes it as a “24/7 personal agent for work, school, and daily life.” It can monitor things you care about, handle complex tasks in the background, and learn your preferences over time.
The data sources it reportedly draws from include your chats, connected apps, personal context, location, and something described as “Agent files.” This means that Remy could, in theory, send messages on your behalf, share documents, or make purchases, all without you having to initiate each step manually.
This Is a Different Kind of Gemini Feature
What separates the Google Remy AI agent from what Gemini already does is the proactive part. Google has been steadily expanding what Gemini can do, from controlling your computer to running agentic multi-step workflows. However, those features still require you to prompt them.
Remy sounds more like Google Labs CC, another experimental agent that proactively organizes your day, but with a much wider scope and deeper integration across Google’s services. The reported list of connected apps includes Gmail, Calendar, Docs, Drive, Keep, Tasks, GitHub, WhatsApp, Spotify, Google Photos, and more.
There’s no confirmed public release date, and Google hasn’t officially acknowledged the project exists. That said, Google I/O kicks off May 19, and AI agents are widely expected to be a central theme. Whether Remy makes an appearance there or stays behind closed doors a while longer is an open question.