Google Cloud has announced the Gemini agent, which it describes as a single, universal agent for work. It was unveiled at the Gemini at Work 2026 event on 8 October 2026, in a keynote by Google Cloud chief executive Thomas Kurian. Google says it plans tasks, uses tools, and returns finished work inside the documents, inbox and developer tools that people already use.
This article sets out what Google says the agent does, which parts are in preview, and which questions the announcement leaves open. Every claim here is Google's own, taken from its announcement posts, and has not been tested independently.
What the Gemini agent is
Google says the Gemini agent combines answering questions, knowledge work, image and media creation, and writing and running code in a single agent and a single API. The idea is that you give it an objective and come back to finished work, rather than giving it step by step instructions.
It runs in the cloud, so Google says it keeps one set of memories and context across web, phone and desktop apps, the command line, Google Workspace, Microsoft 365 and Slack. Long jobs keep running after you close your laptop. It can also create temporary sub-agents for multi-step tasks, and act as a coworker agent with its own identity, email address and storage.
On models, Google says the agent chooses the best one for each job, drawing on its own Gemini family and on Claude models from Anthropic today, with other models promised later. The aim is better accuracy on hard tasks and lower cost on simple ones.
Inside Workspace and for data teams
Inside Google Workspace, the agent works in Gmail, Drive, Docs, Slides, Sheets, Chat and Calendar. Google gives examples such as setting up a meeting with a usual team without being told names, or suggesting that an emailed request for a slide deck be handed to the agent with one click.
For data teams, Google announced skills that let people ask plain language questions and get operational reports. Engineers can describe an outcome and the agent generates PySpark code, trains models and fixes pipeline issues. Google also describes tools called Knowledge Catalog, Smart Storage and a Borderless Lakehouse that sit underneath.
Preview, security and cost
Industry versions for financial services and legal are in preview, and Google says versions for government, healthcare and retail are coming soon. The financial services version draws on data providers and SEC filings and shows confidence scores and source citations. None of this is advice for investors, and Google makes no claim that it replaces professional judgement.
On security, Google says each agent gets its own identity, role based permissions and an audit trail, runs inside a sandbox, and has its traffic checked by a component called Agent Gateway that enforces company policy. On cost, it lists smart routing between models and real time spend caps that pause a project's agent when a limit is hit.
Customers named by Google
Google's post lists many customers using Gemini Enterprise, the product the agent builds on. They include BNP Paribas, Ryanair, Lloyds Banking Group, Verizon, The Home Depot and PayPal. In India and Asia, it names Tata Steel, which it says deployed more than 300 specialised agents in nine months and cut customer complaint turnaround times by 50%, and lists TCS among its Asia Pacific leaders.
These are results as reported by Google and its customers. They are not a measure of what a different company would see.
What Google has not said
The announcement does not give prices for the Gemini agent, so none are quoted here. It does not state which regions or languages will get it first, or give a general availability date for the agent itself, and it does not describe India specific pricing or data residency. Businesses weighing it up should ask Google Cloud directly and read its product pages.
If you follow Google's AI tools, our reports on the SynthID Detector opening to everyone, Gemini Live guided vision tested in India and the latest ChatGPT rollout give wider context. This is information, not advice.
