Google has launched a new Gemini agent for work, aiming to turn Gemini from a workplace chatbot into a universal AI agent that can plan, execute and complete tasks across an organisation’s apps and data.
Announced at Gemini at Work 2026, the new agent can work across Gmail, Drive, Docs, Slides, Sheets, Chat and Calendar, while also connecting to wider enterprise systems. Google says users can delegate an outcome rather than giving step-by-step instructions and allow Gemini to plan the work, choose tools and return a finished result.
What Google announced
Google Cloud CEO Thomas Kurian introduced the Gemini agent on 8 October as a single, universal agent for work.
Google says the agent can:
- answer questions using business context;
- handle knowledge work;
- create images and other media;
- write and run code;
- use skills and tools;
- connect to enterprise systems;
- schedule tasks; and
- respond to events.
The central idea is that workers should be able to delegate a result rather than manually move between separate AI tools and applications.
Official source: Google Cloud — Gemini at Work 2026.
Gemini now works inline across Google Workspace
The most significant practical change is how deeply Google is embedding the agent into Workspace.
Gemini can operate directly inside:
- Gmail;
- Google Drive;
- Docs;
- Slides;
- Sheets;
- Chat; and
- Calendar.
Google says the same memory, skills and enterprise controls can follow the user across those applications.
That means a worker could ask Gemini to research a topic, build a financial model in Sheets, create a presentation in Slides and coordinate a meeting without repeatedly explaining the project in each application.
Google wants users to assign outcomes, not prompts
The new language Google is using is important.
Instead of telling Gemini exactly how to perform every step, users are expected to give it an objective.
For example, a user could ask the agent to prepare for a regional planning meeting. Gemini could identify relevant people from prior Chat threads, check calendars, review documents, create meeting material and coordinate the event.
Google says Gemini can act as:
- a personal assistant for one employee;
- a team member working on behalf of a group; or
- a role-based agent, such as a finance analyst or project manager.
This is different from Google’s Gemini model announcements
The Gemini agent should not be confused with a new underlying model release.
Google continues to develop different Gemini models, but this announcement is primarily about orchestration: one agent that can choose models, use tools, access authorised business context and complete work across applications.
That makes this a product-and-platform story rather than simply another benchmark race against OpenAI or Anthropic.
Related NextNews coverage: Google Gemini 4 Argon and the AI model race.
Gemini can choose the model for the job
Google says the agent can select the best model for a particular task and use multi-model orchestration rather than relying on one model for everything.
That is partly a performance decision and partly a cost-control strategy.
Google is also introducing Smart Routing, real-time spend caps and other controls intended to help businesses manage AI expenditure as agent usage scales.
Security and governance are central to the pitch
Enterprise AI agents are only useful if organisations can control what they are allowed to do.
Google says Gemini’s workplace agent is governed through:
- identity and policy management;
- authorisation and permission controls;
- secure sandboxing;
- network gateways; and
- enterprise administration.
These controls matter because an agent that can act across email, files, calendars and enterprise systems has broader potential access than a conventional chatbot.
The security question is therefore not just whether Gemini generates an accurate answer. It is whether it performs the right action with the right permissions.
Google is also adding specialist business skills
Gemini at Work includes new data and analytics capabilities designed to let both technical and non-technical employees ask operational questions in natural language and receive actionable results.
Google is also adding more specialised tools and knowledge for industries including financial services and legal work.
That suggests Google sees the universal Gemini agent as the front door, with industry-specific skills underneath it.
Google says enterprise AI adoption is already widespread
Google says nearly 80% of Google Cloud customers now use its AI products and nearly 90% of the Fortune 100 use Gemini Enterprise.
Those figures are Google’s own customer metrics and should not be interpreted as independent measures of market share.
Google also says nearly 500 Cloud customers each processed more than one trillion tokens over the past year.
The point Google is making is that large organisations are moving beyond experiments and are now trying to integrate AI directly into operating workflows.
Why this matters for Microsoft and OpenAI
The new Gemini agent intensifies the competition to become the main AI layer inside everyday work.
Microsoft is pushing Copilot and agentic workflows deeply into Microsoft 365 and Windows. OpenAI is expanding agents that can research, code and take action across tools. Anthropic is positioning Claude as an enterprise assistant with increasingly capable tool use.
Google’s advantage is its installed Workspace base and direct control over Gmail, Drive, Docs, Sheets, Slides and Calendar.
If an AI agent can operate naturally inside those applications, Google does not need users to leave their existing workflow to access AI.
The biggest challenge is trust
The more capable an agent becomes, the more serious its mistakes can become.
A chatbot that produces a poor paragraph creates inconvenience. An agent that sends the wrong email, updates the wrong spreadsheet, changes a calendar or acts on inaccurate business data can create operational problems.
That is why permission controls, approval points, audit logs and administration are likely to matter just as much as model intelligence.
Organisations will also need clear internal policies about what Gemini can access, which tasks it can complete automatically and which actions still require human review.
What this could mean for smaller businesses
Google is also pitching the new agent to small and midsize businesses.
The company argues that lean teams can use Gemini as a personal assistant or team member to research, draft, coordinate work and analyse data without hiring specialist staff for every task.
That may be attractive to smaller businesses, but the value will depend on pricing, integration complexity, data governance and whether the agent reliably saves more time than it creates in review work.
Google says small and midsize business use of its Cloud AI tools has already increased more than fivefold year-on-year.
Source: Google Cloud — Gemini for small and midsize businesses.
The bottom line
Google’s new Gemini agent for work is a clear attempt to move enterprise AI beyond the chat window.
The promise is a single agent that understands business context, operates across Workspace, connects to enterprise systems, chooses the right model and delivers completed work instead of isolated answers.
If it works reliably, the biggest change may not be that employees use AI more often. It may be that AI begins carrying out entire chains of work that previously required people to move manually between email, documents, spreadsheets, calendars and internal systems.
The competitive question now is which company can make that experience useful enough — and trustworthy enough — for businesses to delegate real work.
Important disclaimer
This article is provided for general news, technology and informational purposes only. It does not constitute cybersecurity, privacy, procurement, legal, operational, financial or other professional advice and should not be relied upon as advice tailored to your organisation.
AI features, availability, pricing, integrations, security controls and product capabilities can change. Organisations should verify current information with Google and obtain independent advice from appropriately qualified technology, cybersecurity, privacy, legal, procurement and governance professionals before deploying AI agents in business-critical workflows.
Product capabilities and adoption figures attributed to Google are based on Google’s published information and have not been independently verified by NextNews. See the NextNews disclaimer.
Disclaimer
NextNews strives for accurate news, but use it with caution—content changes often, external links may be iffy, and technical glitches happen. See the full disclaimer for details.