Google is reportedly building what could be its most ambitious AI product to date.
An internal document obtained by Business Insider reveals that employees are testing an AI agent codenamed "Remy" – a tool designed not just to answer questions, but to take actions on your behalf across work, school, and everyday life.
"Remy is your 24/7 personal agent for work, school, and daily life, powered by Gemini," reads the internal description. "It elevates the Gemini app into a true assistant that can take actions on your behalf – not just answer questions or generate content."
Beyond Q&A: What Makes Remy Different
Remy reportedly runs inside a staff-only version of the Gemini app, integrated with a range of Google's own services.
According to two people familiar with the project, it can monitor for things users care about, handle complex tasks proactively, and learn preferences over time – moving well beyond the conversational AI model most users are familiar with today.
That's a meaningful step up from Google's existing "Agent Mode" features, which already allow Gemini to handle multi-step tasks but remain limited in scope and subscription tier. Remy appears designed to be something more foundational – a persistent, always-on layer sitting across Google's ecosystem.
The project is currently at the "dogfooding" stage, the industry term for internal employee testing before a product reaches users.
Whether it makes it to Google I/O later this month remains to be seen, but the annual developer conference is widely expected to put agents front and centre.
The Race Nobody Can Afford to Lose
Remy's emergence is no coincidence. The AI industry is in the middle of a full sprint toward autonomous agents – software that doesn't just respond to prompts, but independently plans and executes tasks on users' behalf.
OpenAI has been especially aggressive here.
Earlier this year, an AI agent widely known in tech circles went viral for its ability to respond to messages and conduct research autonomously – its creator was subsequently hired by Sam Altman's company.
Microsoft, meanwhile, unveiled Copilot Cowork in April, pitching it as a "digital employee" capable of planning tasks and automating workflows. Anthropic has similarly shifted resources toward agentic tooling.
Google is acutely aware of the stakes. CEO Sundar Pichai has positioned agents as the lynchpin of the company's AI monetisation strategy, and Google Cloud CEO Thomas Kurian has argued that Google's enterprise agent infrastructure offers capabilities "that nobody else offers."
