Google Cloud on Thursday introduced the Gemini agent, a single universal AI agent built to do your work, not just talk about it.
Unveiled at the company's Gemini at Work 2026 keynote, the agent can answer questions, handle tasks, create content and write code — planning the work itself and returning finished results instead of waiting for the next prompt.
The launch drops Google straight into the middle of an accelerating AI agent arms race.
OpenAI released always-on business agents called dots in September, and Meta rolled out Muse, a personal agent that can shop, book travel, send emails and make payments.
Google's entry is aimed squarely at the enterprise, where nearly 80 percent of Google Cloud customers already use the company's AI products, according to the company.
A coworker, not a chatbot
The pitch is that you hand Gemini an objective instead of a long list of instructions.
It plans the work, picks the right skills and tools, connects a company's business systems and returns a finished result in documents, email or developer environments.
The agent works across the web, Android, iOS, Windows and Mac — and keeps working in the cloud after you close your laptop, with longer tasks able to run for hours or days without another briefing.
Notably, the agent is separate from the model underneath it.
Gemini orchestrates across Google's own Gemini model family and Anthropic's Claude models today, matching the model to each task for quality and cost, with other private and open models planned.
A tools layer connects it to Confluence, Microsoft Office and Teams, Slack, Workspace, Git, Jira, Salesforce, ServiceNow, BigQuery, Databricks, Postgres, Snowflake and any Model Context Protocol server inside or outside the company network.
Google also ships a global library of reusable "skills" — modular prompts, knowledge and workflows — plus an enterprise registry where teams can publish their own.
Your agent gets an email address
The most striking feature is the coworker mode.
Users can create persistent coworker agents with their own identities: an actual Workspace account with an email address, calendar, Drive storage and a presence in the company directory.
The agent acts under its own identity rather than the user's, sees only what is shared with it and follows the team's existing sharing and membership settings.
It can even spin up temporary sub-agents with their own identities to split up complicated jobs.
Google says the agent keeps four kinds of memory — session, semantic, procedural and episodic — and onboards itself the way a new hire would, learning the user, the tools and the team before starting real work. Governance tooling covers identity, authorization, permissions, auditing and policy management, with real-time spending caps to control costs.
The agent wars are officially on
"Gemini can function as your personal assistant or as a team member, where it works on behalf of a group of people, like a project manager within a team, or on behalf of a specific role in an organization, like an analyst in your finance department," Google Cloud CEO Thomas Kurian wrote in a blog post announcing the launch.
Industry-specific versions for financial services and legal work are already in preview, with government, healthcare and retail editions on the way.
Wall Street barely blinked at the news — Alphabet shares were up only marginally in early trading, suggesting investors saw Google's entry into the agent race as inevitable. But for the office workers drowning in busywork, an AI that holds an email address and keeps working after hours may be exactly the point.
By Hamza Faisal
