At a Google Cloud event on Thursday, Google announced a new unified AI agent powered by Gemini that can go beyond answering questions to plan and execute tasks on behalf of users through a single interface.

 

The launch reflects a broader shift in AI, from conversational chatbots to autonomous agents capable of completing tasks such as writing code, scheduling meetings, booking appointments, and organizing travel. Google’s move follows the emergence of other AI agents, including Meta’s Muse, Instinct, and OpenAI’s recently introduced Dots.

 

Google has a significant advantage in expanding its agentic AI offering. CEO Sundar Pichai revealed that Gemini has more than 1 billion monthly active users, while nearly 90% of Fortune 100 companies use Gemini Enterprise at work.

 

Given its strong enterprise presence, Google will initially make the new agent available to businesses before expanding to consumers. Pichai said this approach would help the company address key challenges involving security, scalability, and performance.

 

According to Google Cloud CEO Thomas Kurian, the agent can be assigned “objectives, not just instructions.” Instead of simply responding to prompts, it can develop plans, use specialized skills and tools, and interact with internal business systems to achieve specific goals.

 

The agent automatically selects the most suitable AI model for each task. However, users can also choose a model manually, including third-party options such as Anthropic’s Claude. Google plans to expand these choices to include open-source and additional private models.

 

Businesses can provide files, folders, and project-specific resources to give the agent the context it needs. It can connect with a wide range of enterprise platforms, including Google Workspace, Microsoft 365, Slack, Jira, Confluence, Git, BigQuery, Databricks, Postgres, and Snowflake. It also supports secure connections to Model Context Protocol (MCP) servers inside or outside an organization's network.

 

To help users monitor its work, Google is introducing a Tasks Inbox interface. It allows employees to track progress, see how work is delegated to subagents, and monitor the use of specialized skills and code.

 

Another notable feature is that the agent can have its own Google Workspace account, email address, and working context. It can understand team structures, time zones, calendars, and approval requirements, making it easier to collaborate within an organization. Employees can interact with it by tagging it, emailing it, sharing resources, or adding it to group chats.

 

The agent also maintains an audit trail that attributes its actions to the AI agent itself rather than to an individual employee, helping businesses improve accountability and oversight.

 

Users will be able to access the agent across iOS and Android devices, Windows and Mac computers, command-line interfaces, Google Workspace, Microsoft 365, ServiceNow, and Slack.

 

Early testers include sportswear company On, Shopify, and PayPal. Google also highlighted enterprise customers such as BNP Paribas, Merck, Ulta Beauty, Orange Spain, and Wesfarmers.

 

To help organizations manage costs, Google is introducing flexible spending controls, including multi-model orchestration, intelligent model routing, and real-time spending limits.

 

The bigger picture: Google’s latest move signals a transition from AI assistants that provide information to AI agents that can execute complex workflows across business applications. If Google can deliver the right balance of autonomy, security, accountability, and cost control, agentic AI could become a more integrated part of everyday enterprise operations.