Google employees are testing an internal AI agent codenamed “Remy” that would turn the Gemini app into a fully autonomous personal assistant, according to Business Insider, which reviewed an internal document and spoke with two people familiar with the project.
“Remy is your 24/7 personal agent for work, school, and daily life, powered by Gemini,” reads the internal description obtained by Business Insider. “It elevates the Gemini app into a true assistant that can take actions on your behalf, not just answer questions or generate content.”
What Remy Does
The agent is designed to operate across Google’s existing service layer: Gmail, Calendar, Drive, Android, and Search. Rather than answering prompts, Remy breaks objectives into steps and executes them autonomously. An internal description states that Remy can “monitor for things that matter to you, handle complex tasks proactively, and learn your preferences over time,” according to Business Insider.
Google currently offers “Agent Mode” features in Gemini that can perform multi-step tasks, but access varies by subscription tier and region. Remy represents a more advanced iteration: a persistent agent embedded across Google’s entire consumer stack rather than a feature layered onto search.
A Google spokesperson declined to comment. Two people familiar with the project confirmed that Remy is in an active “dogfooding” phase, meaning employees are testing it internally before any public release, per Business Insider.
The Distribution Asymmetry
The competitive dynamics here are straightforward. OpenClaw proved consumer demand for autonomous personal agents through messaging interfaces. Google is now building the same category of product with one structural advantage that open-source projects cannot replicate: native integration with services that billions of people already use daily.
Gmail, Calendar, and Drive are not third-party connections Remy needs to negotiate. They are first-party services on the same infrastructure. That eliminates the authentication friction, API rate limits, and permission complexity that agents like OpenClaw must navigate when connecting to external tools.
Android Central reported that internal testing suggests Remy could manage work, school, and personal tasks directly inside the Gemini app, with new code found in the latest Google app beta hinting at upgraded planning and autonomous agent features.
Google I/O Timing
Google’s annual I/O developer conference is scheduled later this month. DeepMind CEO Demis Hassabis has repeatedly discussed his vision for building a digital assistant, and agents are expected to be a major focus of the event.
The timing also follows Meta’s development of “Hatch,” its own agentic assistant, and OpenAI’s hiring of OpenClaw creator Peter Steinberger in February. The personal agent category that OpenClaw validated is now a three-way race among the largest technology companies, each bringing different integration advantages: Google has the productivity suite, Meta has the social graph, and OpenAI has the model infrastructure.
No public release timeline for Remy has been disclosed. But the shift from conversational AI to autonomous agent execution inside the apps where users already spend their time represents Google’s clearest bet yet that agents, not chatbots, are the interface that wins consumer AI.