Google Launches "Gemini agent" for Business: Its Own Work Email and Calendar, Optional Claude Models, and Hard AI Spend Caps per Project

At Gemini at Work 2026 on October 8, Google Cloud launched Gemini agent, pitched as a single universal agent for work: it takes objectives rather than just instructions, plans tasks, uses skills and tools, and keeps running in the cloud for hours or days. It is launching for businesses first, with consumers to follow, and no dates or prices were given. Gemini agent can have its own Workspace account with email, calendar, Drive and a directory identity, and every action goes into an audit trail attributed to the agent; it picks a model by default but can use Anthropic's Claude. It connects to Workspace, Microsoft 365, Slack, Jira, BigQuery, Snowflake and other systems, plus any MCP server. Companies can set a hard AI spend cap per project in the Cloud Billing console, and the agent pauses when it is reached. Versions for financial services and legal are in preview.

More than a chat box

On October 8, at Gemini at Work 2026, Google Cloud launched Gemini agent. Google Cloud CEO Thomas Kurian says you can give it "objectives, not just instructions": it plans tasks itself, uses skills and tools, connects to internal company systems, and keeps running in the cloud for hours or even days, whether or not your laptop is open.

It is launching for businesses first, with a consumer version to follow, and Google gave no dates or prices. Versions for financial services and legal are in preview, with government, healthcare and retail "coming soon."

An identity like a new colleague

The most distinctive design choice is identity. Gemini agent can have its own Workspace account, with email, calendar, Drive and an entry in the company directory; coworkers can email it, tag it in group chats or share files with it, and it sees only what people share. Every action is written to an audit trail attributed to the agent, not to an employee. All tasks run in a sandbox with its own network boundary, and traffic passes through an Agent Gateway that enforces company policy.

On models, it chooses automatically by default or can be set manually; options currently include the Gemini family and Anthropic's Claude, with open and private models to come. It connects to Workspace, Microsoft 365, Slack, Jira, Confluence, Git, BigQuery, Databricks, Postgres, Snowflake and more, and to any MCP server.

What it means

For enterprise IT, the most practical piece may be cost control: you can set a hard AI spend limit for a project in the Cloud Billing console, and when it is reached the agent pauses until you resume it with one click; spending is tracked per project, so it can be charged back to departments. Agents with their own accounts and auditable actions are becoming standard for big-vendor enterprise agents, and that is easier to govern than handing an employee's own permissions to an agent. But with no pricing and no consumer timeline, for now this is mainly Google's move in the race to be the enterprise agent platform. Google says Gemini has more than 1 billion monthly users and nearly 90% of the Fortune 100 use Gemini Enterprise, both company figures.

via: Google Cloud blog, TechCrunch report