Book a call

News · October 8, 2026

Google Cloud bets on one work agent that runs Gemini and Claude

// ai.news :: 2026

At Gemini at Work 2026, Google Cloud introduced a single Gemini agent for chat, long-running tasks and code, with coworker agents that get their own identities and a choice of Google or Anthropic models per job. For IT leaders, identity, spend caps and audit trails are the parts to study first.

What happened

  • Google Cloud announced the Gemini agent on October 8, 2026, describing it as one agent that answers questions, handles knowledge work, creates media and writes and runs code from a single prompt box and API, and keeps working in the cloud on tasks that last hours or days.
  • According to Google Cloud, teams can create persistent coworker agents with their own identities, including company email addresses and storage, which can be added to chats or tagged in document comments, and the agent can spin up temporary sub-agents for parallel steps.
  • Google said each job can run on Gemini models or on Anthropic's Claude models, with Smart Routing choosing a model by performance and cost, and hard spending caps per project that pause the agent when reached.
  • For governance, Google Cloud listed an attested identity for every agent, administrator-approved permissions, action logs attributed to the agent, an isolated Agent Sandbox for code and an Agent Gateway that enforces organization-wide policies.
  • Unite.AI reported that Financial Services and Legal versions are in preview, with Government, Healthcare and Retail listed as coming soon. Google's post gives no pricing or general-availability date.

Why it matters

Agents with their own email addresses and accounts behave more like new employees than like software features. Treat them that way: a named owner, a defined role, least-privilege access, and offboarding when the job is done. Identity and audit design should come before the first pilot.

Multi-model routing inside one platform is useful, but it moves model choice into the vendor's hands. Keep your own evaluation tasks and check which model actually handled important jobs, especially where quality or data location matters.

Hard spend caps per project are a good default and worth copying in any agent setup, including ones you build yourself. Without pricing or a GA date, this is a time to run a limited pilot on a non-critical workflow, not to plan a company-wide rollout.

Sources

  • Google Cloud Blog: Gemini at Work 2026: Introducing the Gemini agent
  • Unite.AI: Google Cloud Unveils Gemini, Its Universal Agent for Work

This post is CMM’s own summary and commentary on publicly reported news. Facts are attributed to the sources linked above; please read them for full details. CMM is not affiliated with these publishers or with the companies mentioned, and trademarks belong to their owners. Prepared with AI assistance. Corrections or requests: [email protected]. See our editorial policy.

Want help with this?

Tell us what you need, and we will suggest the right team model on a short call.

Talk to us