Google Launches Agentic AI for Gemini, Starting With Businesses
▼ Summary
– Google announced the launch of a unified Gemini agent capable of performing tasks autonomously rather than just answering questions.
– The company plans to initially roll out this agentic AI to businesses to address security, scale, and performance challenges before targeting consumers.
– The new agent can plan work based on objectives, connect to various internal systems like Google Workspace and Microsoft 365, and use third-party models such as Anthropic’s Claude.
– Users can monitor the agent’s progress through a tasks inbox interface that displays its thinking process, delegation, and audit trails.
– Gemini operates with its own Workspace account, allowing it to interact with colleagues via email or chat while maintaining context about teams and schedules.
Google has officially entered the era of agentic AI, announcing a new unified agent for its Gemini platform that goes beyond simple conversation to execute complex tasks on behalf of users. Unveiled at a Google Cloud event, this tool represents a significant shift in how artificial intelligence integrates into professional workflows, allowing the system to take ownership of assigned duties rather than merely providing information.
The launch aligns with a broader industry trend where AI tools are evolving from passive conversational interfaces into active agents capable of generating code, scheduling meetings, booking travel, and managing other operational responsibilities. This development follows similar moves by competitors, including Meta’s Muse, messaging-based agents like Instinct, and OpenAI’s recent introduction of Dots within ChatGPT.
Enterprise-First Strategy and Scale
Despite the growing consumer interest in AI agents, Google is prioritizing enterprise adoption for its initial rollout. The company aims to address the complex challenges associated with deploying powerful autonomous systems, particularly regarding security, scalability, and performance.
Sundar Pichai, CEO of Google, highlighted the massive existing footprint of Gemini as a foundation for this expansion. He noted that the platform currently boasts over 1 billion monthly active users. Furthermore, he emphasized that nearly 90% of Fortune 100 companies already utilize Gemini Enterprise in their daily operations. By targeting businesses first, Google intends to refine the technology in high-stakes environments before expanding it to the general public.
Thomas Kurian, CEO of Google Cloud, described the fundamental difference between the new agent and previous iterations. He explained that the system can be given “objectives, not just instructions.” This capability allows the AI to independently plan workflows, utilize custom skills and tools, and interface directly with internal business systems to achieve specific goals.
Integration and Model Flexibility
The new agent is designed to operate seamlessly within existing corporate ecosystems. It connects to a wide array of data sources and productivity platforms, including Google Workspace, Microsoft 365, Slack, Jira, Confluence, Git, BigQuery, Databricks, Postgres, and Snowflake. Additionally, it supports secure connections to any Model Context Protocol (MCP) server, whether hosted inside or outside the company’s network.
Users retain control over which AI models power these tasks. While the system defaults to selecting the most appropriate model for a given job, operators can manually override this choice. Initial third-party support includes Anthropic’s Claude models, with plans to expand the selection to include open-source and private models in the future.
The interface for interacting with the agent is equally robust. Users can submit requests containing attachments such as files, folders, or specialized project workstreams. A dedicated “tasks inbox” provides transparency into the agent’s operations, displaying its reasoning process, task delegation to subagents, loading of specialized skills, generated code, and real-time progress.
Operational Identity and Accessibility
A distinctive feature of the Gemini agent is its creation of a dedicated Workspace account, effectively treating it as a colleague within the organization. This identity grants the AI its own email address and contextual awareness of the company structure. It understands team compositions, employee time zones, approval hierarchies, and calendar availability. Interaction is flexible, allowing users to tag the agent in chats, send emails to it, share documents, or add it to group conversations. As it executes actions, the agent maintains an audit trail attributed to itself rather than the human user, ensuring clear accountability.
The agent is accessible across multiple platforms, including iOS and Android mobile devices, Windows and Mac desktops, command-line interfaces, and integrated environments within ServiceNow and Slack.
Early testing involved major brands such as sportswear manufacturer On, e-commerce platform Shopify, and financial services giant PayPal. Broader Gemini Enterprise customers cited include BNP Paribas, Bradesco, Merck, Orange Spain, Santee Cooper, SOMPO, Ulta Beauty, and Wesfarmers. To assist enterprises in managing expenses, Google introduced flexible spending controls, featuring multi-model orchestration, smart routing, and real-time spend caps to help organizations maintain better oversight of their AI costs.
(Source: TechCrunch)
