For the past three years, AI has been something you talk to. Google just made it something you delegate to. At its Gemini at Work 2026 event on October 8, Google Cloud CEO Thomas Kurian unveiled the Gemini agent — a single, universal AI system designed to take a goal, go off and do the work, and return with the finished result. It has its own email address. Its own calendar. And it can work for hours or days without you ever opening a laptop.
This isn’t a chatbot upgrade. It’s Google’s most direct challenge yet to Microsoft, OpenAI, and Anthropic in the race to build the AI coworker that actually does the job.
What Exactly Is the Gemini Agent?
The Gemini agent is not a new app. It’s a new layer that runs across the apps you already use.
According to Google’s announcement, it is “a single, universal agent for work that answers your questions, handles your knowledge work, creates your images and media, and writes and runs code — all in a single agent and a single API.”
The core idea is delegation. Instead of giving an AI step-by-step instructions, you give it an objective. The agent figures out how to achieve it. As Google puts it: “give it objectives, not instructions.”
For example, if your manager emails you asking for the latest project update as a slide deck, Workspace Intelligence recognises that as a delegatable task and gives you a single-click option to pass it to Gemini. Gemini then builds the deck for you.
What Can It Actually Do?
The Gemini agent handles four broad categories of work:
Knowledge work. It can research information, synthesize documents, and pull together reports. It connects to your systems and brings back something finished — inside the documents, inbox, and developer environments you already work in.
Content creation. It generates images and media, drafts documents, and builds presentations. It can work across Gmail, Google Docs, Sheets, Slides, Drive, Chat, and Calendar — carrying the same memory and context wherever it operates.
Coding. It writes and runs code. For developers, it can research issues, scope fixes, build and test them, and return completed pull requests with video demonstrations of the changes.
Coordination. It can set up meetings by figuring out who should attend based on your chat spaces and past events, checking their calendars, and coordinating a time that works — even with external participants. You don’t need to supply a single name or email address.
The Coworker Agent: An AI With Its Own Email Address
The most striking feature is what Google calls coworker agents. A manager can describe a role — say, an events coordinator or a financial analyst — and Gemini creates an agent with its own Workspace account: an email address, a calendar, Drive storage, and a listing in the company directory.
Employees can then assign work to the agent much as they would to a colleague, including by mentioning it in Google Chat or requesting document edits. Because execution takes place in the cloud, you can start an assignment on one device, leave it running, and resume from another without reconstructing the context. The agent can manage workflows lasting multiple days, coordinate parallel assignments, and respond to scheduled tasks.
Google says the agent operates 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.”
How It Works Under the Hood
The Gemini agent is not just Gemini. It’s a multi-model orchestration layer.
Google says the agent “runs each job on the model that fits best.” That includes both the Gemini and Claude families of models today, with support for other private and open models in the future. The goal is to deliver optimal quality while lowering costs.
The agent uses sub-agents to tackle multi-step tasks, coordinating workflows that can run in parallel or sequentially. It runs in the cloud with a single set of memories, context, and one personalization graph. Google describes four kinds of memory:
Session memory: for the task in front of it, even when it runs for days
Semantic memory: a structured knowledge base it builds as it reads documents, talks to people, and works with other agents
Procedural memory: for how a job gets done, including skills it writes for itself
Episodic memory: everything it has done before
Safety and Cost: Treating Agents Like Employees
Giving software the ability to act on its own raises obvious safety questions. Google’s answer is to treat each agent like an employee.
Every agent gets a cryptographically attested identity — a digital ID verified through cryptography — and every action lands in an audit trail attributed to the agent rather than a person. All agents run inside an Agent Sandbox, a walled-off environment meant to keep software away from the rest of a system, and all their traffic passes through Agent Gateway, which Google describes as an “AI network firewall.”
On cost, Google says nearly 500 of its customers each processed more than one trillion tokens over the past year. The agent includes real-time spend caps: when a project hits its limit, the agent pauses until someone clicks to resume.
What It Means for Work
The Gemini agent represents a fundamental shift in how Google thinks about AI in the workplace. Work, according to Kurian, “now starts in the prompt window.”
But the agent isn’t just for individuals. It’s designed to become part of the organisational fabric — a persistent digital coworker with its own identity, permissions, and memory, operating across Google Workspace, Microsoft 365, Slack, and any system that speaks the Model Context Protocol.
The competitive context is intense. Microsoft introduced an updated Copilot on September 25. OpenAI introduced its always-on “dots” agents four days later. Anthropic announced native Claude integration with Google Docs, Sheets, and Slides on October 6. And SpaceXAI expanded its persistent Grok Bot agents in September.
Google’s answer is not to build a better chatbot. It’s to build the agent that runs the work — and to give it an email address so it can be treated like anyone else on the team.
The Gemini agent
is currently in private preview. Wide availability is coming soon for Workspace customers on select Business and Enterprise plans. Google has not published pricing or a general availability date. Financial services and legal versions are in preview, with government, healthcare, and retail listed as coming soon.
For now, the promise is clear: an AI that doesn’t just answer your questions, but takes your objectives, plans the work, uses the tools, connects to the systems, and brings back something finished.
The chatbot era is over. The coworker era has begun. And this one has its own inbox.
