Google Announces One Gemini Agent to Rule Them All, Gives It Its Own Email Address

At a Google Cloud event on Thursday, Google announced a unified Gemini agent for work, and the detail that deserves a moment of silence is this: the agent gets its own Workspace account, including an email address. Gemini is now a coworker. Its onboarding presumably does not include the mandatory training video.

The Register’s headline for the news is “There can be only one,” which is a good summary of the strategy and an accurate summary of the 1986 film. Google Cloud CEO Thomas Kurian called Gemini “your new single, universal agent for work,” per The Register, and said users should give it “objectives, not just instructions.”

The facts, per TechCrunch’s Sarah Perez. Google is launching the agent for businesses first, with consumers later. Alphabet CEO Sundar Pichai said Gemini has over 1 billion monthly active users and that nearly 90% of Fortune 100 businesses use Gemini Enterprise at work. The agent can plan work, use custom skills and tools, and connect to a company’s internal systems, including Google Workspace, Microsoft 365, Slack, Jira, Confluence, Git, BigQuery, Databricks, Postgres, Snowflake, and any Model Context Protocol server. A “tasks inbox” shows its reasoning, delegation to subagents, loaded skills, code, and progress.

And then the org-chart parts. The agent knows team membership, time zones, approvers, and calendar details. You can tag it, email it, share with it, or add it to a group chat. Its actions appear in an audit trail attributed to the agent rather than a person. In human terms: it has a seat, it has a calendar, and when something goes wrong, the paper trail says Gemini did it. This is, for the first time, an employee who is genuinely excited about being on the hook.

Several details read like HR policy. The Register reports that each agent has an identity separate from the user’s, that company policies apply to agents the same way they apply to people, and that a budget is set before a task starts and covers its full cost. TechCrunch lists “real-time spend caps” among the new cost controls. We’ve all worked with someone who needed a hard cap; here it is, built into the product.

By default the agent chooses the best model for each task. Users can pick manually, including third-party models, starting with Anthropic’s Claude, per TechCrunch. So Google’s “single universal agent” will, when asked, delegate some work to a competitor’s model. It is a very modern kind of vertical integration: you own the front desk, and the person behind the front desk may be from the other company.

Early testers named by TechCrunch include On, Shopify, and PayPal. Named Gemini Enterprise customers include BNP Paribas, Bradesco, Merck, Orange Spain, Santee Cooper, SOMPO, Ulta Beauty, and Wesfarmers. The Register adds a customer example from Kurian: NTT Docomo, which reportedly saved 450,000 hours annually using data agents.

The Register’s Thomas Claburn is not entirely sold, and he supplies the context. He questions the vendor-reported statistics and savings, and cites a dataset putting the median lifespan of 308 discontinued Google products and services at about 4.1 years. He notes that Google introduced flexible billing and cost controls for agents two months ago, that the Wall Street Journal reported AI spending is becoming nearly impossible for businesses to budget, and that in July Coworker AI observed Google had two Gemini Enterprise products that could be confused. The article says Google now presents just one.

That last point is the quiet headline. “There can be only one” is a strategy, but it is also a cleanup: after a summer of overlapping Gemini offerings, the company arrived at the radical position that a product should have one name. The agent, for its part, does not appear to mind.

Pichai also cited the company’s Gemini 4 Argon scoring 77.9% on the DeepSWE v1.1 coding benchmark, which he said beat Claude Opus 5.5, Claude Fable 5.1, and GPT-6 Astra, per The Register. This is a vendor-reported figure delivered by the vendor’s CEO, at the vendor’s event, and it is presented here with all the weight that arrangement deserves.

The pattern is clear. A year ago, the question was whether AI agents could do the work. This week, Google’s pitch has moved to everything around the work: identity, permissions, audit logs, budgets, who gets to tag whom. TechCrunch places the launch in a growing field alongside Meta’s Muse, Instinct and other messaging-based agents, and ChatGPT’s Dots. The agent race, it turns out, is increasingly about the administrative layer, which is also the layer where actual companies spend actual years of their lives.

So Gemini now has an email address. Somewhere, a human in IT is typing the first support ticket in the history of the position: Gemini cc’d the entire company again. The audit trail, at least, will know exactly whom to blame.

Sources: