Google is building a new AI personal agent for Gemini, internally codenamed 'Remy' or 'OpenClaw,' that can take actions on the user's behalf across apps, services, and devices, according to reporting by Business Insider. The development represents Google's most ambitious attempt yet to move Gemini beyond a conversational assistant into a true agentic system capable of executing multi-step tasks in the real world — booking appointments, managing files, sending emails, and interacting with third-party services without requiring the user to be present at every step.

"The agent is designed to act on users' behalf — not just answer questions, but complete tasks across the digital environment."

— Business Insider

Gemini 3.1 Arrives in Smart Homes and Cars

The Remy announcement comes alongside a series of concrete Gemini upgrades announced today. Google has rolled out Gemini 3.1 for its Home voice assistant, bringing multi-step task handling to smart home devices. The update allows users to issue complex, multi-part commands and have them executed as a single coherent workflow rather than a sequence of separate commands. Google has also announced that Gemini is coming to cars with Google built-in, replacing Google Assistant in eligible vehicles.

On the developer side, Google has expanded the Gemini API File Search tool with three major updates: multimodal support, custom metadata filtering, and page-level citations. Multimodal support means that developers can now search across documents, images, and other media types in a single query, rather than being limited to text-based retrieval.

UK Workers Vote to Unionize

Not all of today's Google AI news was about product launches. In a significant development reported exclusively by The Guardian, workers at Google DeepMind's UK offices have voted to unionize, citing concerns about the company's military AI contracts and the broader direction of AI development. One worker pointed specifically to the Iran war and the Pentagon's feud with Anthropic as indications that the AI industry is moving in a direction that conflicts with the values many researchers joined the field to advance.

The unionization vote is the latest sign of growing labor tension within the AI industry, where researchers and engineers who built their careers on the promise of beneficial AI are increasingly confronting the reality that their work is being deployed in military and surveillance contexts. For Google, which has faced employee protests over military contracts before — most notably the 2018 Project Maven controversy — the DeepMind vote represents a fresh challenge to its ability to pursue government contracts without internal dissent.