The company also moved Claude Docs, Slides and Design out of beta, making them generally available on all Claude plans.
Google introduces Gemini task agent with businesses first in line
The company plans a later consumer rollout and says the agent will support third-party models, starting with Anthropic’s Claude.

Google unveiled a Gemini agent designed to carry out work as well as respond to questions, with businesses getting access before consumers. Announced at a Google Cloud event, the system brings those capabilities together in one interface, the company said.
Google Cloud CEO Thomas Kurian said the agent can work toward assigned goals by planning tasks, drawing on specialized tools and skills, and accessing internal corporate systems. Google CEO Sundar Pichai said starting with businesses would allow Google to tackle challenges involving security, scale and performance.
Google said automatic model selection will be the default, with users able to override it. Anthropic’s Claude models will be among the initial third-party choices; support for open source models and additional private models is planned.
The company said employees can supply attachments and connect the agent to services such as Slack, Snowflake, Microsoft 365 and Google Workspace. A task inbox will show progress, code and work delegated to subagents.
Google said the agent will have a dedicated Workspace account and email address. Employees can bring it into conversations or share materials with it, while an audit trail will identify its actions separately from those of people.
Early testing involved On, PayPal and Shopify. Pichai put Gemini’s monthly active user count at over 1 billion and said nearly 90% of Fortune 100 businesses use Gemini Enterprise. Google also announced spending options that include real-time caps to help businesses control AI costs.
