Google launches Gemini agent to automate tasks for businesses
Google has announced a universal Gemini AI agent for enterprise users that can complete tasks across apps and act as a virtual coworker.
Google has announced a new unified Gemini AI agent designed to help businesses automate complex workflows. Revealed at a company event on Thursday, the tool marks Google's transition into the "agentic" era of artificial intelligence. Instead of just answering questions, this new agent can take ownership of assigned tasks and complete them across different applications from a single interface.
What is the new Gemini agent?
The new tool is a universal agent that runs in the cloud, allowing it to maintain the same context across mobile devices, desktops, and web browsers. Unlike traditional chatbots that require step-by-step instructions, this agent can be given broad objectives. It then plans the work, uses specialized skills, and delegates tasks to smaller sub-agents to achieve the goal.
To make collaboration easier, the AI agent can be treated like an actual employee. It receives its own corporate email address and Workspace account. Users can interact with it by sending an email, tagging it in a message, or adding it to a group chat. When the agent takes action, it leaves an audit trail attributed to itself rather than a human coworker.
How does it connect to business systems?
The agent is built to integrate deeply with a company's existing data and software. It can connect to various platforms, including:
Users can monitor the agent's progress through a "tasks inbox" interface. This dashboard displays the agent's thinking process, shows which sub-agents have been assigned to specific tasks, and tracks overall progress.
- Google Workspace apps like Gmail, Drive, Docs, Sheets, and Calendar
- Third-party productivity tools like Microsoft 365, Slack, and ServiceNow
- Developer and project management platforms such as Jira, Confluence, and Git
- Data warehouses and databases like BigQuery, Databricks, Snowflake, and Postgres
- Any Model Context Protocol (MCP) server inside or outside the company network
Which AI models does it use?
By default, the agent automatically selects the best AI model to complete a given task. However, businesses can also choose their preferred models. This includes third-party options, starting with Anthropic's Claude models, with plans to add open-source and other private models in the future.
To help companies manage the costs of running these models, Google is introducing flexible spending options. These include smart routing, multi-model orchestration, and real-time spending limits.
Why it matters for businesses
Google is focusing its initial rollout on corporate clients to address complex challenges around security, scale, and performance before releasing the tool to consumers. The company already has a massive footprint in the enterprise market, with Google CEO Sundar Pichai noting that nearly 90% of Fortune 100 businesses use Gemini Enterprise.
Early testers of the agent include brands like Shopify, PayPal, and On. The tool is currently available in a private preview for enterprise customers, offering a look at how businesses might soon use autonomous digital assistants to handle daily administrative workloads.
Comments 0
No comments yet. Start the conversation.