Google unveils the Gemini agent for work

What Google announced

On 8 October 2026, at its Gemini at Work event, Google Cloud unveiled the Gemini agent, which it describes as ‘a universal agent for work’. According to Google’s official blog (opens in a new tab) and the launch post on the Google Cloud blog (opens in a new tab), the agent ‘plans the work, uses skills and tools, connects to your systems, and brings back something finished’, inside the documents and inbox people already use.

The pieces Google describes include:

• Long-running work: the agent runs in the cloud and can carry on with tasks that take hours or days, with a single memory and context, from any device or channel.
• Workspace integration: it works inside Gmail, Drive, Docs, Slides, Sheets, Chat and Calendar. Google also describes ‘coworker agents’ that act under their own identity and see only what the team shares with them.
• Models used: according to Google, the agent orchestrates across the Gemini family and Anthropic’s Claude models today, with other models to follow.
• Governance and cost: an identity for each agent, an audit log, an isolated sandbox and per-project spending caps that pause the agent when they are hit.
• Industry versions: according to Google, the financial services and legal versions are in preview, with government, healthcare and retail ‘coming soon’.

The announcement gives no availability date for the agent, and no plans, prices or regions. Nor does it say which features will reach each type of Workspace account. Wait for that information before planning anything.

What it means for your business

For a small business or a local council, there is nothing to install yet. The announcement shows where the product is heading: assistants that carry out tasks on email, calendars and documents. So, as well as checking answers, you have to decide what the agent is allowed to do.

Three points to watch. First, access: an agent should only see what it needs, and that depends on how tidy your Drive and mailbox permissions are. Second, data protection: if the agent handles personal data, the GDPR still applies, so find out who the data is shared with and where it is processed. Third, human oversight: depending on the use, the EU AI Act calls for transparency and human control, and a delegated task still needs someone accountable for it.

How to try it this week

• List three repetitive team tasks (summarising email threads, drafting meeting minutes, tidying a spreadsheet) and mark the ones with no personal or confidential data.
• Check who can open which shared folders. If access is too wide, fix it before you connect any assistant.
• Try the assistant you already pay for on a made-up or public document, and note what it does well, what it makes up and how much time it saves.
• Write down who checks the result before it leaves the organisation.
• Ask your Workspace admin or supplier what will be available on your plan, since the announcement does not say.

IAescola designs tailored training for teams that want to bring in assistants and agents with clear permissions and responsible use of data.