Google Announces a Single Gemini Agent for Work, With Claude in the Model Picker, in Private Preview

Google's new work agent can run on a rival's model and work inside a rival's office suite, but only selected customers can use it so far. At Gemini at Work 2026 on October 8, Google Cloud announced the Gemini agent, which it calls "a universal agent for work." In Google's description, it plans the work, uses skills and tools, connects to customers' business systems, and "brings back something finished" inside the documents, inboxes and developer environments people already use.

Who can use it, and when

The launch is narrower than the keynote framing:

• Private preview only. A Google Cloud spokesperson told VentureBeat the agent is available "in private preview to a select group of customers," with wider availability promised soon but no date.
• Businesses first. TechCrunch reports Google will bring the agent to businesses before consumers; CEO Sundar Pichai said that lets it solve the "harder problems around security, scale, and performance."
• Price. Google told VentureBeat the agent will be included at no additional charge where Gemini Enterprise is available, including through existing Workspace business and enterprise subscriptions, and that it will share more pricing detail closer to the wider release. VentureBeat notes Google has not said whether persistent agents with their own Workspace accounts need extra licences.

What Google says it does

• Objectives, not prompts. Google Cloud CEO Thomas Kurian said the agent can be given "objectives, not just instructions," per TechCrunch.
• Model choice. By default it picks a model; users can choose third-party models, starting with Anthropic's Claude, with open-source and other private models planned. VentureBeat says administrators can restrict which models agents may use.
• Connections. TechCrunch lists Google Workspace, Microsoft 365, Slack, Jira, Confluence, Git, BigQuery, Databricks, Postgres and Snowflake, plus any Model Context Protocol (MCP) server.
• Its own identity. The agent gets its own Workspace account and email address, works from a "tasks inbox" that shows its steps and subagents, and writes an audit trail attributed to the agent rather than a person, according to TechCrunch.
• Where it runs. iOS and Android, Windows and Mac, the command line, Workspace, Microsoft 365, ServiceNow and Slack.
• Cost controls. Google cites multi-model orchestration, smart routing and real-time spend caps. VentureBeat notes project-level spending limits were first announced in August.

TechCrunch reports the agent's early testers included On, Shopify and PayPal. Pichai said Gemini has more than 1 billion monthly active users and that nearly 90% of Fortune 100 companies use Gemini Enterprise; VentureBeat notes that customer examples such as BNP Paribas and SOMPO describe existing Gemini Enterprise deployments and do not necessarily represent adoption of the new agent.

What it signals

The notable choice is putting Claude in the picker. Two days earlier, Anthropic launched Claude for Google Workspace in beta; Google is now offering Claude inside its own agent. That suggests Google wants enterprises to standardise on its agent layer while models can be swapped underneath, a reading VentureBeat also draws, in the same contest as OpenAI's Dots agents.

Open questions remain, per VentureBeat: Google said prompts, documents and chat histories stay inside Gemini Enterprise when customers use third-party models such as Claude, but did not say whether data is sent to outside model providers for processing, or what retention and residency terms apply. Google also published no independent benchmarks for how reliably the agent completes long, cross-application work.

Limits

• All capability descriptions are Google's, via its blog post, TechCrunch's event coverage and VentureBeat's briefing; none has been independently tested.
• The availability, pricing and data-handling details come from a Google Cloud spokesperson quoted by VentureBeat; Google's own blog post does not state them.
• Google's blog post points to a longer Google Cloud blog post, which we did not read.

Sources

• Google (blog.google): Google Cloud introduces the Gemini agent (October 8, 2026)
• TechCrunch: Google brings agentic AI to Gemini, starting with businesses (October 8, 2026)
• VentureBeat: Google Cloud unveils persistent Gemini Agents for long-running tasks, and they get their own Gmail, Calendar, and Drive storage (October 8, 2026)
• Times of AI: Anthropic Puts Claude in Google Docs, Sheets and Slides — Marketplace Beta, Not a Partnership
• Times of AI: Dots Runs Always-On Agents on the Astra OpenAI Just Refused to Upgrade