Google Cloud's new Gemini agent wants to be the only AI tool your company needs

Most companies now juggle several AI tools: one chatbot for questions, another for writing, a separate assistant for code. Google Cloud's pitch at its Gemini at Work 2026 event was that all of that should collapse into one box.

CEO Thomas Kurian introduced the Gemini agent, which Google calls "a single, universal agent for work." Its slogan is simple: "You give it objectives, not instructions."

What the Gemini agent does

Instead of answering one prompt at a time, the agent is meant to take a goal, plan the work, use tools and come back with something finished. Google lists four jobs it handles from the same prompt box:

• answering questions,
• knowledge work such as research and reports,
• creating images and other media,
• writing and running code.

It works inline in Gmail, Drive, Docs, Sheets, Slides, Chat and Calendar, and it can also be reached from the web, iOS and Android, Windows and Mac desktops, the command line, Microsoft 365 or Slack. Because it runs in the cloud, it keeps one memory across all of them, and long jobs keep running after you close your laptop.

Sub-agents and "coworker agents"

For bigger tasks, the agent can create a temporary roster of sub-agents, each with its own identity, and coordinate them over hours or days.

Google goes a step further with coworker agents: AI team members with a fixed role, such as an analyst in the finance department. They get their own @agents.company.com email address, their own storage, and only the context people choose to share with them.

It doesn't only use Gemini models

The detail that stands out is model choice. Google separates the agent from the model underneath and says it runs each job on whichever model fits best, "orchestrating across our Gemini family of models and Claude models from Anthropic today," with other private and open models planned.

The argument is about cost as much as quality. Simple tasks don't need the biggest model, and Google is adding smart routing and real-time spend caps so companies can keep bills under control.

Connected to the software companies already use

The agent can plug into tools such as Confluence, Microsoft Office, Teams, Slack, Git, Jira, Salesforce, ServiceNow, BigQuery, Databricks, Postgres and Snowflake, plus any MCP server. It also uses reusable skills, written instructions that teach it how a specific task gets done. Both ideas are explained in plain English in our guide to MCP and Agent Skills.

How big is Google's enterprise AI business?

Kurian shared a few numbers to show the scale:

These are Google's own figures and haven't been independently checked.

What it means for you

If your company runs on Google Workspace, expect this agent to show up where you already work. The practical questions for IT teams will be permissions and cost: the agent acts with each user's access, so what it can reach depends on what people can already open. For a wider look at the idea, read What is an AI agent?

Sources

• Gemini at Work 2026: Introducing Gemini agent (Google Cloud)
• Google Cloud launches Gemini agent (Google)