Home / AI News / Gemini agent

Google's Gemini Agent: What It Does, Coworker Agents and When You Can Use It

Google's new agent for work runs long tasks, creates AI teammates with their own accounts and connects to your business tools. It is not public yet.

Spark5X · · Google

Google Cloud announced the Gemini agent on October 8, 2026, at its Gemini at Work event. Google calls it a single, universal agent for work: you give it an objective rather than step-by-step instructions, and it plans the work, uses your company's tools and data, and returns finished output in your documents, inbox or code. It is not available to the public yet, so for most businesses the question today is what to prepare, not whether to switch.

1. What is the Gemini agent?

The Gemini agent is Google Cloud's AI agent for businesses. From one prompt box it answers questions, handles knowledge work, creates images and media, and writes and runs code, and it can keep working on a task for hours or days after you close your laptop.

Google describes it running in the cloud with one memory and one set of preferences that follow you across devices. You reach it on the web, on iOS and Android, on Windows and Mac, from the command line, and inside Google Workspace, Microsoft 365 and Slack. You can assign it work, schedule tasks, or have it respond to events as they happen.

Two details stand out. First, it can split a large job into smaller ones and hand each to a temporary sub-agent, then combine the results. Second, Google says it picks between its own Gemini models and Anthropic's Claude models today, with other models to follow, choosing for each task the model that does the job well at the lowest cost.

2. Coworker agents: an AI teammate with its own email address

A coworker agent is a permanent member of your team that Google's agent creates when you describe a role. It gets its own Google Workspace account, with an email address, calendar, Drive and a place in the company directory.

Colleagues can add a coworker agent to a Chat space or mention it by name, as they would a person. It works under its own identity, shows up under its own name in a document's version history, and sees only what has been shared with it. Google's own example is a marketing manager asking a coworker agent for a launch readiness document.

The design choice worth noticing is accountability. Because the agent has its own account, everything it does is recorded against the agent rather than borrowed from an employee's login, which makes it easier to see who did what.

3. Inside Gmail, Docs and Sheets

In Google Workspace, it works inside Gmail, Drive, Docs, Slides, Sheets, Chat and Calendar, and it keeps the same memory and permissions across all of them. Google describes three ways it helps:

  • Personal assistance. It knows your calendar, team and projects, so you can ask it to book a meeting with "the usual team" without listing names, or to research a market, build a model in Sheets and turn it into a slide deck.
  • Proactive delegation. Workspace spots tasks it could take on, such as a request for a slide deck in your inbox, and offers to hand them over in one click. It also points out the messages that matter most and explains why.
  • A member of your team. The coworker agent described above.

4. What it connects to

It connects to the business tools most teams already use, including Salesforce, Slack, Microsoft Office and Teams, Jira, Confluence, ServiceNow, BigQuery, Snowflake and Databricks, plus files on your own computer.

It also works with any Model Context Protocol (MCP) server, the open standard that lets an AI assistant use an outside tool or data source. Companies can publish their own tools and skills, which Google describes as reusable instructions and workflows, to a shared registry so every team's agent can use them. If MCP is new to you, our guide to connecting Google Ads and Keyword Planner to Claude through MCP shows what a connection like this looks like in practice.

5. Is it available yet?

Not publicly. In a follow-up post on October 16, Google Cloud said the new Gemini agent capabilities are not publicly available yet and invited businesses to contact its sales team to discuss how to get ready.

Google's announcement gives no general availability date. The versions built for particular industries are in preview for financial services and legal, with government, healthcare and retail listed as coming soon. If you already use Google Workspace or Google Cloud, your account team or admin console is the place to check when your organisation can turn it on.

6. The controls Google built in

Every agent gets its own verified identity and limited permissions, every action is written to an audit trail, and a spending cap can pause the agent when a project reaches its budget.

Google frames this around four questions: who the agent is, what it may do, what it did, and what it must never touch. Agents run in a sandbox, and all their traffic passes through a gateway where security teams write company-wide rules once, for example that agents may not open documents classified Need to Know. Spending limits are set per project, so a runaway task stops instead of running up a bill, and work resumes with one click.

7. What it means for marketing and growth teams

Agents like this are only as useful as the systems and data they can reach, so the work that pays off now is getting those systems in order.

Three steps make sense before any rollout:

  • Write down the repetitive work. Weekly reports, lead follow-up, campaign briefs and content calendars are the tasks an agent can take on first. A clear, written process is what turns into a reusable skill.
  • Clean up your CRM and shared definitions. An agent that reads your CRM inherits its gaps. Consistent lead stages and agreed definitions, such as what counts as a qualified lead, matter more than which agent you choose. Our CRM automation work starts here.
  • Keep a person approving anything that leaves the company. Let agents draft emails, ads and reports, and have a person approve them before they reach a customer.

None of this ties you to Google: the same preparation works for any AI agent, including the Claude and ChatGPT tools many teams already use. If you want help deciding what to automate first, see our AI automation services.

Frequently asked questions

What is the Gemini agent?

The Gemini agent is Google Cloud's AI agent for businesses, announced on October 8, 2026. From one prompt box it answers questions, handles knowledge work, creates images and media, and writes and runs code, and it can keep working on long tasks in the cloud.

What is a Gemini coworker agent?

A coworker agent is a permanent AI member of a team that Google's agent creates from a role description. It has its own Google Workspace account, including an email address, calendar and Drive, and it sees only what is shared with it.

Does the Gemini agent use Claude models?

Yes. Google says it chooses between its own Gemini models and Anthropic's Claude models today, with other models to follow.

Is the Gemini agent available now?

Not publicly. On October 16, 2026, Google Cloud said the new Gemini agent capabilities are not publicly available yet and asked businesses to contact its sales team to prepare.

More AI news

All AI news →

Get started
Put this to work
Talk to us about using new AI tools in your marketing and growth systems.
    0/1000
    Prefer to pick a time? Book a call →
    🔒 We'll only use your details to reply. See our Privacy Policy.
    Keep reading

    Related AI news

    Request a quoteBook a call