Google is turning Gemini from a chatbot into a coworker. At a Google Cloud event on Thursday, the company launched a unified Gemini agent that can answer questions and also complete work on a user’s behalf, all from one interface, according to TechCrunch AI. Businesses get it first. Consumers will get it later.
The timing makes sense. AI tools are moving past conversation and into doing real work: writing code, scheduling meetings, booking travel. Google is entering a busy field that already includes Meta’s Muse, messaging agents like Instinct, and ChatGPT’s recently launched Dots. Google does have reach on its side, though. CEO Sundar Pichai said Gemini has more than 1 billion monthly active users, and that nearly 90% of Fortune 100 companies use Gemini Enterprise.
7 Things to Know About Google’s New Agent
- It takes objectives, not just instructions. Google Cloud CEO Thomas Kurian said the agent can be given “objectives, not just instructions.” You hand it a goal, and it plans the work, picks custom skills and tools, and connects to internal systems to finish the job. That’s the real difference between an agent and a smart autocomplete.
- It’s a real member of the team, with its own account. This is the most interesting design choice. The agent gets its own Google Workspace account and email address. It knows who sits on which team, their time zones, who signs off on what, and what’s on people’s calendars. You can tag it, email it, share files with it, or add it to a group chat, the same way you’d loop in a colleague.
- Claude is in the model picker. By default, the agent chooses the best model for each task. Users can override that and pick a model themselves, and the third-party options start with Anthropic’s Claude models. Google says open source and other private models will come later. A Google product that openly offers a competitor’s model tells you enterprise buyers care more about flexibility than brand loyalty.
- It plugs into the tools companies already use. The agent connects to Google Workspace, Microsoft 365, Slack, Jira, Confluence, Git, BigQuery, Databricks, Postgres, Snowflake, and more. It also works with any Model Context Protocol (MCP) server, the open standard for linking AI agents to outside tools and data, whether that server sits inside or outside the company network.
- A “tasks inbox” shows its work. Users can follow along in a tasks inbox that shows Gemini’s reasoning, the jobs it hands off to subagents (smaller helper agents that each take a piece of the job), the skills it loads, the code it writes, and its progress. Requests can include files, folders, or project bundles that pair files with specific skills.
- Every action leaves an audit trail. When the agent does something, the record is attributed to the agent, not a person. For compliance teams, that’s a big deal. Tracing who did what is one of the biggest unsolved problems with autonomous AI at work.
- It runs almost everywhere. The agent works on iOS and Android, Windows and Mac desktops, the command line, Google Workspace, Microsoft 365, ServiceNow, and Slack.
Why Businesses Come First
Pichai said starting with enterprises lets Google tackle the “harder problems around security, scale, and performance” that come with powerful agents. That’s a fair point. An agent that can email people, read calendars, and query databases needs guardrails that a casual consumer rollout would put under real strain.
Early testers included sportswear brand On, Shopify, and PayPal. Existing Gemini Enterprise customers include BNP Paribas, Bradesco, Merck, Orange Spain, Santee Cooper, SOMPO, Ulta Beauty, and Wesfarmers.
The Cost Question
Google also announced new flexible spending options: multi-model orchestration, smart routing, and real-time spend caps. Agents that run long tasks across several models can burn through budgets fast, so these controls matter. Still, the report doesn’t include specific pricing, a firm consumer launch date, or a timeline for adding open source models.
What It Means
What stands out here is the Workspace account. Giving the agent its own identity, email, and audit trail means Google sees agents as accountable participants in the workplace, not features buried in a sidebar. If that model sticks, the next fight in enterprise AI won’t be about which chatbot is smartest. It’ll be about which agent companies trust enough to hire.
Full details are available in the original TechCrunch AI report.