Google is pushing artificial intelligence deeper into everyday business operations with a new version of Gemini that can take responsibility for tasks, coordinate work across applications, and deliver completed results with less human input.
Announced at the company’s Gemini at Work 2026 event on Thursday, October 8, the Gemini agent represents a shift from AI tools that primarily answer questions to systems capable of planning and executing assignments on behalf of individuals and entire teams.
Google is initially targeting businesses, where the technology could help employees manage complex workflows, analyse information, develop software and coordinate projects through a single interface.
The announcement comes as technology companies increasingly compete to develop AI agents that can perform practical tasks rather than simply generate responses to prompts.
From Giving Instructions to Assigning Goals
Google Cloud CEO Thomas Kurian said the new system allows users to define the outcome they want instead of providing detailed instructions for every step.
“You give it objectives, not instructions. It plans the work, uses custom skills and tools, connects to your systems, and brings back something finished,” Kurian explained during the event.
The agent can break complicated assignments into smaller tasks, deploy specialised sub-agents and work across connected business applications to complete a project.
For example, an employee could ask Gemini to research an industry, organise the findings into a financial model and prepare a presentation without manually transferring information between applications.
Google says the agent can continue working in the cloud even after users close their computers, allowing certain assignments to run for extended periods.
Businesses Already Have a Large Gemini User Base
Google is introducing the technology to enterprise customers first, drawing on its existing presence in the corporate market.
Chief executive Sundar Pichai said Gemini has more than 1 billion monthly active users, while nearly 90 per cent of Fortune 100 companies use Gemini Enterprise for work, according to figures presented at the event.
The company believes business customers offer a practical environment for developing and deploying more advanced AI agents, although security, reliability and performance remain important considerations.
Early testers include sportswear brand On, e-commerce platform Shopify and payments company PayPal.
Other organisations using Google’s enterprise AI products include BNP Paribas, Bradesco, Merck, Orange Spain, Santee Cooper, SOMPO, Ulta Beauty and Wesfarmers.
Google said the new agent was available in preview to selected customers, with broader access planned.
One Agent Across Multiple Workplace Platforms
Rather than requiring employees to switch between separate AI tools, Google is positioning Gemini as a central system that can operate across existing workplace software.
The agent can connect to services including:
-
Google Workspace and Microsoft 365
-
Slack, Jira and Confluence
-
Git and ServiceNow
-
BigQuery, Databricks, PostgreSQL and Snowflake
It also supports connections to systems using the Model Context Protocol (MCP), enabling it to interact securely with compatible tools inside or outside an organisation’s network.
Users can access Gemini through mobile devices, Windows and Mac computers, command-line interfaces and supported workplace applications.
The agent can also work with files, folders and project-specific resources supplied by employees, allowing it to draw on relevant information when completing assignments.
AI Coworkers Get Their Own Digital Identity
One of the more significant changes is Google's plan to give certain agents their own workplace identities.
A team-based agent can have its own email address, calendar, and account within Google Workspace, allowing employees to interact with it much as they would with a colleague.
Workers can assign tasks by mentioning the agent in a group chat, sharing documents or sending it an email. The agent can then carry out its responsibilities using the information and permissions made available to it.
Google says actions taken by the system are recorded in an audit trail under the agent's identity rather than being attributed to an individual employee.
This approach is intended to help organisations track automated work and establish clearer accountability as AI systems become more involved in business operations.
However, the technology also makes access controls and oversight essential, particularly when agents can interact with confidential documents, corporate databases and internal communication channels.
Companies Can Choose Different AI Models
Google is also allowing businesses to use models beyond its own Gemini family.
The agent can select a suitable model for a particular assignment automatically, while users can make their own selections when needed. Support for Anthropic's Claude models is included at launch, with additional open-source and proprietary models planned for the future.
This separates the agent's ability to manage a task from the underlying model used to perform it.
The approach could help companies balance performance, accuracy and cost instead of relying on one model for every type of assignment.
Google is also introducing flexible spending controls, including smart routing, multi-model coordination and real-time expenditure limits. These features are designed to help businesses monitor AI usage and prevent automated workloads from exceeding their budgets.
Data Analysis and Industry-Specific Tools
Beyond general office work, Google is developing specialised capabilities for technical teams and particular industries.
Data engineers and machine-learning specialists will be able to use Gemini to generate code, work with notebooks, train models and troubleshoot data pipelines. Other employees will be able to request operational reports using everyday language.
Google is also introducing industry-focused capabilities for financial services and legal work, with further specialisations planned for sectors including healthcare, retail and government.
Examples shared by the company include Bradesco's use of enterprise AI to accelerate financial document reviews and Orange Spain's deployment of custom agents across departments such as human resources, IT, sales and customer service.
These examples illustrate potential productivity benefits, although results will vary according to the tasks, systems and safeguards involved.
Security and Governance Remain Key Challenges
Giving AI systems the ability to act independently introduces risks that go beyond those associated with conventional chatbots.
An agent that can access business systems, execute code or communicate with colleagues needs clear boundaries governing what it can see and do.
Google says its enterprise offering includes individual agent identities, role-based permissions, activity monitoring and sandboxing. Its Agent Gateway is designed to enforce organisational policies on traffic moving between agents and other systems.
These safeguards are intended to help companies monitor automated activity and limit access to sensitive information.
Even with these controls, businesses will need to determine which tasks can be delegated safely and where human approval remains necessary.
Industry analysts have also questioned whether Google's expanding collection of Gemini products will make it easier for businesses to manage AI or introduce further confusion about which tools they need.
What Google's Move Means for the AI Market
The launch signals a broader change in how AI companies are positioning their products: the focus is moving from generating information towards completing work.
For businesses, successful AI agents could reduce repetitive administrative tasks, accelerate research and help employees coordinate complicated projects. They could also introduce new costs, security concerns and questions about accountability.
Google's existing user base gives it a substantial platform from which to expand, but adoption will depend on whether organisations can demonstrate practical benefits while maintaining control over their data and automated systems.
For now, the company is concentrating on enterprise customers as it develops the technology. Its longer-term challenge will be turning the promise of autonomous digital coworkers into dependable tools that businesses can use at scale.
Source: Google Cloud — Gemini at Work 2026