Google announced a unified Gemini agent for businesses on Oct. 8 at its Gemini at Work event, TechCrunch’s Sarah Perez reported. The agent is meant to carry out work for users from a single interface, and Google will offer it to businesses before consumers.
Google Cloud CEO Thomas Kurian said the agent can be given “objectives, not just instructions,” according to TechCrunch. In Google’s own write-up of his keynote, Kurian described it as one agent that answers questions, handles knowledge work, creates images and media, and writes and runs code, “all in a single agent and a single API.”
Google CEO Sundar Pichai said Gemini has more than 1 billion monthly active users and that nearly 90% of Fortune 100 companies use Gemini Enterprise, TechCrunch reported. He said starting with businesses lets Google work through the “harder problems around security, scale, and performance” of running agents like this one.
How it works
The agent plans the work, uses custom skills and tools, and connects to a company’s internal systems, Kurian said. TechCrunch listed systems including Google Workspace, Microsoft 365, Slack, Jira, BigQuery and Snowflake. It can also work with any server that uses the Model Context Protocol (MCP), an open standard for connecting AI tools to data.
The agent gets its own Workspace account and email address, like a co-worker, Google said. It knows who’s on which team and who needs to approve what, along with colleagues’ time zones and calendars. Users can tag it, email it, share files with it or add it to a group chat, and its actions leave an audit trail attributed to the agent rather than a person, TechCrunch reported.
Google’s blog says the agent runs in the cloud, so work that takes hours or days keeps going after a user closes a laptop. It can create temporary subagents, each with its own identity, to split up a multi-step job. Longer-lived “coworker agents” get their own @agents email addresses and storage and only see the context people give them, according to the post.
Users can follow progress in a “tasks inbox,” which shows the agent’s reasoning, subagent handoffs, loaded skills, code and progress, according to TechCrunch.
Which models it uses
By default, the agent picks a model for each job. Users can choose one themselves, including Anthropic’s Claude models, and Google says open and other private models will be added later. “The best model for the task is not always the largest one,” Kurian wrote.
Google also announced cost controls, including routing jobs across models and real-time spending caps, TechCrunch reported.
Early testers included the sportswear brand On, Shopify and PayPal. Google’s blog says PayPal routes 10 million multi-model requests every week. It also says nearly 500 Google Cloud customers each processed more than one trillion tokens in the past year.
TechCrunch’s report didn’t give a date for the consumer version, and Google’s post didn’t either.
Source: TechCrunch, Sarah Perez, “Google brings agentic AI to Gemini, starting with businesses”
Additional sources: Google Cloud Blog, Thomas Kurian, “Welcome to Gemini at Work 2026: Introducing the Gemini agent”
