TechForge

October 9, 2026

  • Google Cloud’s Gemini AI agent automates tasks across enterprise applications.
  • Google adds security and spending controls for AI agents.

 

Google Cloud has introduced a new Gemini AI agent that can carry out multi-step workplace tasks across enterprise applications. Announced at Gemini at Work 2026, the agent allows employees to assign work, generate content, analyse data, and execute code through a single interface.

According to Google, the agent can plan and execute tasks using connected applications and company data, with work continuing in the cloud after an employee closes their device. It can also coordinate specialised agents to complete different stages of an assignment.

The launch builds on Google’s April 2026 expansion of Gemini Enterprise, which introduced tools for agent development, orchestration, and governance. The new agent combines these capabilities in a single interface for assigning and managing workplace tasks.

Gemini takes on more workplace tasks

The agent works within Google Workspace applications, including Gmail, Drive, Docs, Sheets, Slides, Chat, and Calendar. It also connects to external platforms such as Microsoft 365, Slack, Salesforce, ServiceNow, and Jira.

Google said the agent can operate through third-party applications without requiring a dedicated interface. It supports connections to enterprise databases, including BigQuery, Databricks, PostgreSQL, and Snowflake, as well as tools connected through Model Context Protocol (MCP) servers.

The agent maintains a shared history and context across devices and applications through its cloud-based architecture. Employees can therefore access assigned tasks from desktop or mobile devices without having to provide the same information again.

Google demonstrated this capability using an email requesting a project update in presentation format. Workspace Intelligence identifies the request and suggests delegating it to Gemini, which can then gather information from connected applications to prepare the presentation.

The company is also introducing what it calls coworker agents, which can carry out defined responsibilities for a team. These agents receive their own email addresses, calendars, storage, and identities within an organisation.

Unlike temporary agents created for individual assignments, coworker agents maintain defined roles across multiple work sessions. Employees can interact with them through Google Chat or assign work by mentioning them in documents, with access limited by existing sharing permissions.

Google also provides a tasks inbox where employees can monitor assigned work. The interface displays task progress, delegated assignments, selected skills, and code execution.

Gemini uses reusable instructions, known as skills, to carry out specific workflows. It can also create temporary agents to complete different parts of an assignment simultaneously or in sequence.

The agent can also select between Google’s Gemini models and Anthropic’s Claude models, with support for additional private and open models planned. Google’s Smart Routing system chooses models based on task requirements, performance, and cost.

Beyond general workplace tasks, Google is extending Gemini’s capabilities to data analysis and engineering. The agent can generate SQL and PySpark code, prepare reports using BigQuery, and assist developers with data pipelines and machine learning workflows.

Gemini also integrates with Google’s Knowledge Catalog to access information about datasets, business definitions, and data relationships. According to Google, the agent uses this information to identify relevant datasets and generate queries, while data engineers can use it to write and test code.

Security and cost controls for AI agents

Google has outlined security and governance controls for agents that access enterprise applications and internal company data. These include identity management, access permissions, audit logs, and restrictions on how agents interact with connected systems.

Google said each agent receives a cryptographically verified identity with permissions approved by security administrators. Its activities are recorded in audit logs, allowing organisations to track actions performed across connected systems.

When an agent accesses external applications, identity and authorisation information can be exchanged using standards such as OAuth. Administrators can control which resources an agent is permitted to access and what actions it can perform.

Google said its Agent Sandbox isolates agent execution, while Agent Gateway enforces network and access policies. Administrators can apply restrictions across an organisation’s agents, including limits on access to documents with specific security classifications.

Google has also detailed spending controls for agents that perform longer-running tasks and use multiple AI models. The controls cover token consumption and computing resources used during code execution.

Through Google Cloud’s billing console, administrators can establish project-level spending limits. According to the company, Gemini monitors usage and sandbox execution costs, pausing agent activity when the configured limit is reached.

Administrators can resume paused tasks and monitor AI spending by project, allowing costs to be allocated to individual departments. The spending controls also work alongside Smart Routing, with organisations able to select models manually when needed.

How businesses are using Gemini Enterprise

Google also provided examples of organisations deploying its existing Gemini Enterprise platform, including companies in financial services. These deployments involve existing products rather than confirmed production use of the newly announced universal Gemini agent.

According to Google, Singapore’s DBS Bank is deploying structured chains of 70 to 80 specialised agents to prepare corporate credit memorandums. The agents combine financial, regulatory, and market information, with employees remaining involved in reviewing the results.

Australian financial technology company Airwallex has deployed Gemini Enterprise to support customer service and dispute processing. According to figures published by Google, the company reduced dispute-handling time by 40% and resolved customer issues 10 times faster.

Airwallex is also developing AI workflows for treasury management, payables, and payments. The reported performance figures come from Google and have not been independently verified.

Google has not provided comparable production performance figures for the newly announced universal Gemini agent.

Google is also expanding its industry-specific Gemini offerings, starting with financial services and legal applications. The financial services offering includes tools for investment research, credit analysis, and risk modelling, with access to data from providers such as LSEG, S&P Global, and FactSet.

Additional offerings are planned for government, healthcare, and retail. The new Gemini agent is initially available to enterprise customers through a private preview, with no general availability date announced.

 

 

 

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events, click here for more information.

Tech Wire Asia is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.

About the Author

Muhammad Zulhusni

As a tech journalist, Zul focuses on topics including cloud computing, cybersecurity, and disruptive technology in the enterprise industry. He has expertise in moderating webinars and presenting content on video, in addition to having a background in networking technology.

Related

November 13, 2026

November 13, 2026

November 11, 2026

November 11, 2026

Join our Community

Subscribe now to get all our premium content and latest tech news delivered straight to your inbox

Popular

35678 view(s)
28211 view(s)
25631 view(s)
20740 view(s)

Subscribe

All our premium content and latest tech news delivered straight to your inbox

This field is for validation purposes and should be left unchanged.