At a Google Cloud event, Google announced an agentic version of its Gemini AI that can perform tasks on users’ behalf from a single interface, not just answer questions. The company positioned the move as part of a broader shift from conversational AIs toward systems that can take ownership of assigned work, such as generating code, scheduling meetings, booking appointments and travel.
The launch follows growing interest in consumer-facing and messaging‑based agents — for example Meta’s Muse, Instinct and ChatGPT’s Dots — but Google emphasized Gemini’s existing enterprise footprint as a competitive advantage.
Enterprise-first rollout and scale
Sundar Pichai, CEO of Google, said Gemini currently has over 1 billion monthly active users. He also noted that nearly 90% of Fortune 100 companies now use Gemini Enterprise at work. Given that adoption, Google will initially make the agent available to businesses so it can address "harder problems around security, scale, and performance" before a broader consumer release.
Thomas Kurian, CEO of Google Cloud, explained that the new agent accepts "objectives, not just instructions": it can plan tasks, leverage custom skills and tools, and connect to internal business systems to accomplish goals.
Models, integrations and extensibility
By default the agent selects the most appropriate model for a task, but users can override that choice and pick other models — including third‑party models, starting with Anthropic’s Claude models. Google said it will expand the model picker to include open‑source models and other private models over time.
Requests can include attachments such as files, folders or project bundles designed for specific workstreams. The agent can connect to a company’s data and systems, including Google Workspace, Microsoft 365, Slack, Jira, Confluence, Git, BigQuery, Databricks, Postgres, Snowflake and others. It can also work securely with any Model Context Protocol (MCP) server inside or outside a company’s network.
Visibility, an in‑product account and audit trails
Users monitor the agent from a "tasks inbox" interface that displays the agent’s reasoning, delegation to subagents, special skills and code it loads, and progress on assignments. Google said the agent will have its own Workspace account — effectively a co‑worker identity with its own email address and context. The agent understands team membership, time zones, approval requirements and calendar items within the company.
People can invoke the agent by tagging it, emailing it, sharing with it, or adding it to a group chat. When the agent takes actions, it writes an audit trail attributed to the agent rather than to an individual user.
Platforms, testers and customers
The agent will be accessible from iOS and Android mobile devices, Windows and Mac desktops, the command line interface, Google Workspace, Microsoft 365, ServiceNow and Slack.
Early testers included On, Shopify and PayPal. Google listed Gemini Enterprise customers such as BNP Paribas, Bradesco, Merck, Orange Spain, Santee Cooper, SOMPO, Ulta Beauty and Wesfarmers.
Cost controls
To help businesses manage AI spending, Google announced new flexible spending options including multi‑model orchestration, smart routing and real‑time spend caps.
