Google Launches the Gemini Agent, Its New Universal Agent for Work

Google launches the Gemini agent: a friendly robot at a laptop with email, calendar, document, checklist, and chart panels floating around it

Short Answer: What did Google announce with the Gemini agent?

On October 8, 2026, Google Cloud announced the Gemini agent, a single agent for work that answers questions, handles knowledge work, creates images and media, and writes and runs code from one prompt box. It works across Google Workspace, Microsoft 365, Slack, and the command line, and can run jobs on Gemini or Anthropic’s Claude models. Google is rolling it out to businesses first, with consumers to follow and no date given.

Quick Summary

  • Thomas Kurian, CEO of Google Cloud, announced the agent at Gemini at Work 2026. It plans the work, uses skills and tools, connects to company systems, and returns finished output inside the documents, inbox, and developer environments people already use.
  • It runs in the cloud with one set of memories across devices, and long jobs keep running after you close your laptop. It can create temporary sub-agents and persistent “coworker agents” with their own email addresses.
  • Each job runs on the model that fits best. Google says that means Gemini models and Anthropic’s Claude models today, with other models planned.
  • Google says every agent gets its own identity and audit trail, runs in a sandbox behind an Agent Gateway firewall, and can be paused by a per-project spend cap.
  • Google’s announcement does not give pricing or a general availability date for the agent. Gemini for Financial Services and Legal are in preview; Government, Healthcare, and Retail are coming soon.

Google announced the Gemini agent on October 8, 2026, in a keynote from Google Cloud CEO Thomas Kurian that Google published as a blog post titled “Welcome to Gemini at Work 2026”. TechCrunch covered it the same day. According to TechCrunch, Google CEO Sundar Pichai said Gemini has more than 1 billion monthly active users and that nearly 90% of the Fortune 100 use Gemini Enterprise. Those are Google’s figures.

The launch comes as agents become a bigger part of the AI landscape. We have covered the consumer side with Meta Muse; this is Google’s agent for the workplace. If the term is new, start with our explainers on what an AI agent is and multi-agent systems.

What Is the Gemini Agent?

The Gemini agent is Google’s name for one agent that does the jobs that used to be spread across a chatbot, an assistant, an image tool, and a coding tool. Google says it answers questions, handles knowledge work, creates images and media, and writes and runs code, all in one agent and one API.

The pitch is delegation. Google’s post says users give the agent “objectives, not instructions,” and come back to finished work. Google says it can act as a personal assistant, or as a team member working on behalf of a group, such as a project manager on a team or an analyst in a finance department.

How Does the Gemini Agent Work?

Google describes the agent through six design principles. This table summarizes them in our words.

Capability What Google says
Unified agent Chats, works on objectives you assign, runs scheduled tasks, responds to events, and generates code from one interface.
Access everywhere Available on web, iOS, Android, Windows, Mac, the command line, Google Workspace, Microsoft 365, and Slack, and can run as a headless agent inside third-party apps. TechCrunch adds ServiceNow.
Persistent execution Runs in the cloud with one set of memories and one personalization graph across devices. Work that takes hours or days continues after you close your laptop.
Multi-agent orchestration Creates temporary sub-agents, each with its own identity, for parallel and sequential steps. Coworker agents have a persistent role, their own @agents.company.com email address, and their own storage.
Deep context Learns from each interaction. Teams can create projects with dedicated context, skills, and tools.
Model choice Runs each job on the model that fits best, across Gemini and Anthropic’s Claude models today, with other private and open models planned. TechCrunch reports users can also pick the model themselves.

TechCrunch adds that users can follow along from a tasks inbox, where they can see the agent’s thinking process, delegation to sub-agents, the skills it loads, code, and progress.

What Can the Gemini Agent Connect To?

Google groups the agent’s abilities into tools, skills, and memory.

  • Tools: Google lists collaboration tools (Confluence, Microsoft Office, Teams, Slack, Workspace), development tools (Git, Jira), enterprise platforms (Salesforce, ServiceNow), databases (BigQuery, Databricks, Postgres, Snowflake), and files on a user’s own desktop. It can also work with any Model Context Protocol (MCP) server, inside or outside a company’s network, and a tools registry lets teams publish tools for the rest of the company.
  • Skills: Reusable instruction sets that teach the agent how to do multi-step tasks. Gemini ships with a global skills library, and teams can publish custom skills to a shared registry. Google says skills improve consistency and save token costs.
  • Memory: Google describes four kinds: session memory for the task in front of it, semantic memory (a knowledge base it builds), procedural memory for how a job gets done (including skills it writes for itself), and episodic memory of what it has done before.

How Does the Gemini Agent Work Inside Google Workspace?

Google says the agent works inline in Gmail, Drive, Docs, Slides, Sheets, Chat, and Calendar, carrying the same memory, skills, and controls it has elsewhere. It works in three ways:

  • Personal assistance: It arrives knowing your calendar, team, and projects. Google’s example is asking it to set up a meeting with the usual regional event leads without supplying names or email addresses; it works out who they are, checks calendars, and starts an email thread, even with external participants.
  • Proactive delegation: Workspace can recognize a delegatable task, such as a manager asking for a project update as a slide deck, and offer a single click to hand it to Gemini.
  • A member of your team: You describe a role and Gemini creates a coworker agent with its own Workspace account, including an email address, calendar, Drive, and a place in the company directory. Colleagues add it to a Chat space or @mention it.

Google’s coworker example is a marketing manager asking an events coordinator agent to draft a launch readiness document and post it to the group, or tagging the agent in a document comment, where it can suggest an edit and reply under its own name in version history. Google says a coworker agent acts under its own identity rather than yours and sees only what you share with it.

How Does Google Say It Governs Agents and Controls Costs?

Google frames governance as four questions: who the agent is, what it may do, what it did, and what it should never touch. Its answers, according to the announcement:

  • Identity: every agent gets a cryptographically attested identity with least-privilege permissions, stamped into the logs of its work.
  • Permissions: fine-grained, role-based access approved by security administrators, with identity passed to external systems through standards such as OAuth.
  • Auditing: every action is written to an audit trail attributed to the agent, not to a person.
  • Policy: agents run in an Agent Sandbox with its own network boundary, and all traffic passes through Agent Gateway, which Google calls an AI network firewall. An organization writes a policy once, such as agents may not open documents classified Need to Know, and it applies to every agent.

On cost, Google points to multi-model orchestration, Smart Routing that sends each workload to the model with the best performance for the price, and real-time spend caps set per project in the Cloud Billing Console. If a cap triggers, that project’s agent pauses and can be resumed with one click. Because tracking is per project, companies can charge AI costs back to specific departments.

What Is Available Now, and What Has Google Not Said?

Item What the announcement says
Who gets it first Businesses. TechCrunch reports consumers will get the agent later, with no date given.
Industry versions Gemini for Financial Services and Legal are in preview. Government, Healthcare, and Retail are coming soon.
Gemini in Workspace Described as working inline across Gmail, Drive, Docs, Slides, Sheets, Chat, and Calendar. The announcement does not give a rollout date.
Model picker Gemini and Claude models today. Open-source and other private models are planned.
Pricing and general availability Not stated. Google says new flexible spending options help control costs but gives no prices.

That gap matters for planning. An announcement of this size can mean different things for different customers, so check with your Google Cloud or Workspace contact before assuming your organization has access.

What Results Is Google Citing?

Google lists many customer stories. It places them under Gemini Enterprise, so they are not necessarily results from the new agent itself, and they are Google-reported rather than independently verified. Examples include:

  • Ulta Beauty says its AI shopping assistant, built with Gemini Enterprise, boosted digital sales conversion rates three-fold.
  • Wesfarmers says shopping agents that help customers discover products and build carts raised conversion rates up to three times at Kmart and Officeworks, and an internal Bunnings agent saved half a million hours of administrative work.
  • Bradesco says document review dropped from one hour to five minutes. Snap says diagnostic troubleshooting dropped from 30 minutes to 30 seconds. Commerzbank says document review dropped from 20 hours to one.

Google also says early testers of the agent’s dynamic model selection included the sportswear brand On, and notes that Shopify and PayPal already use multi-model approaches at scale.

What Does This Mean for Marketers and AEO?

Google’s announcement is about enterprise work, not public search, so we want to be careful about how far to take it. These are our observations, not Google’s claims.

Research may start inside work agents. Google’s own Workspace example has an agent researching market trends, building a financial model in Sheets, and producing a deck. The announcement does not say which sources an agent uses for that kind of research. But if more buyer research starts in a work agent rather than a search box, the public information about your brand that an agent can retrieve and trust matters more. Our guide to agentic search covers how that behavior differs from a standard query.

Retailers are building shopping agents on Gemini Enterprise. Ulta and Wesfarmers are examples Google cites. Our coverage of the NielsenIQ finding that 51% of U.S. consumers used an AI tool to shop in the past month shows why that is relevant. Clear, consistent product facts help any agent describe your products correctly, and our guide to writing product descriptions for AI search is a practical starting point.

Agents will show up as participants. A coworker agent has its own email address, appears in a company directory, and posts in chat spaces. For marketing teams, that is a working model for delegating drafts, reports, and meeting coordination, with the audit trail Google describes. Whether it fits your workflow depends on pricing and availability that Google has not yet given.

The practical step is to watch for access in your own tools and to test with a bounded task first, such as a launch checklist or a weekly report, and to check the output as you would a new colleague’s.

Frequently Asked Questions

What is the Gemini agent?

It is a single agent from Google Cloud that answers questions, handles knowledge work, creates images and media, and writes and runs code from one prompt box. It plans work, uses skills and tools, connects to company systems, and returns finished output inside the apps people already use.

When did Google announce the Gemini agent?

Google Cloud announced it on October 8, 2026, at its Gemini at Work 2026 event.

Who can use the Gemini agent?

Google is starting with businesses. TechCrunch reports consumers will get it later, with no date given. Google’s announcement does not state pricing or a general availability date.

Does the Gemini agent use Claude?

Yes, in part. Google says the agent runs each job on the model that fits best, across Gemini models and Anthropic’s Claude models today, with other private and open models planned.

Can the Gemini agent have its own email address?

A coworker agent can. Google says coworker agents have their own @agents.company.com email addresses and storage, and in Workspace they receive their own account with a calendar, Drive, and a place in the company directory.

How does Google keep the agent under control?

Google says each agent has its own attested identity, role-based permissions, and an audit trail, runs in a sandbox behind an Agent Gateway firewall, and can be paused by a per-project spend cap.

Sources: Google Cloud: Welcome to Gemini at Work 2026, Introducing the Gemini agent, TechCrunch: Google brings agentic AI to Gemini, starting with businesses.

About the author

Kai Williams

Kai Williams has been in marketing for years, with a long background in SEO before AEO had a name. He stepped into Answer Engine Optimization the moment AI started reshaping how people search, and has been tracking the shift ever since. At Prompt Insider, he covers AEO, AI marketing, and the future of search, breaking down what is changing and what brands need to do about it.

Get the insider edge

AI news, AEO tactics, and tool reviews — straight to your inbox.