Google just gave Gemini a Workspace account. Not a metaphor — an actual email address, calendar awareness, and the ability to tag into group chats like any other employee. That detail, buried midway through Thursday’s Google Cloud event, says more about where AI is heading than almost anything else the company shared.
As TechCrunch reported, Google is rolling out a unified agentic version of Gemini that can plan and execute multi-step work, connect to internal systems, delegate to subagents, and keep a written audit trail of everything it does. The goal is to give it objectives, not just prompts. Thomas Kurian, CEO of Google Cloud, framed it that way explicitly: the agent plans the work, selects the right tools, and connects to whatever systems the business already uses.
The integration list is broad. Gemini can pull from Google Workspace, Microsoft 365, Slack, Jira, Confluence, Git, BigQuery, Databricks, Postgres, Snowflake, and any Model Context Protocol server inside or outside the company network. Users interact with it by tagging it, emailing it, or adding it to a chat. It writes its own audit trail attributed to the agent, not to any individual person, which matters a lot in regulated industries.
On the model side, Google isn’t locking users into Gemini only. By default the agent picks the best model for the task, but users can override that. And notably, Anthropic’s Claude models are available from day one in the model picker, with open source and other third-party models to follow. That’s a meaningful signal. Google is betting that enterprise buyers care more about flexibility than exclusivity, and that being the orchestration layer matters more than being the only model in the room.
Still, the business case here rests on Gemini’s existing footprint. Sundar Pichai noted at the event that Gemini now has over one billion monthly active users, and that nearly 90% of Fortune 100 companies are already using Gemini Enterprise. That installed base is a real advantage when you’re trying to sell agentic AI into organizations that are already familiar with the product. Early testers included On, Shopify, and PayPal, with large enterprise customers like BNP Paribas, Merck, and Ulta Beauty already in the Gemini Enterprise column.
The decision to start with businesses before consumers is also worth paying attention to. Google is framing it as a way to work through the harder problems around security, scale, and performance in a more controlled environment. But it’s also the right commercial move. Enterprise buyers tolerate more friction, pay more per seat, and provide the kind of structured feedback that helps a company like Google refine agent behavior before it goes anywhere near consumer use cases.
The competitive context is real. OpenAI’s ChatGPT has been building toward similar agent functionality with Dots and operator integrations. Meta has Muse. Messaging-native agents like Instinct are targeting specific workflows. Microsoft, which has deep Copilot integrations across 365, is the most direct competitor here given how much enterprise infrastructure runs on its stack. Google’s answer seems to be: meet users where they already are, across both Google and Microsoft environments, and let the agent move freely between them.
For developers and IT teams evaluating this, the key questions will be around control and cost. Google said it’s introducing flexible spending options including multi-model orchestration, smart routing, and real-time spend caps. That suggests they’ve heard the feedback from early enterprise AI deployments where costs scaled faster than expected. The tasks inbox, which shows Gemini’s thinking process and subagent delegation in real time, is also a meaningful transparency feature. Knowing what the agent is doing, and why, is something most organizations will require before they hand over any real operational responsibility.
- Connects to Google Workspace, Microsoft 365, Slack, Jira, Confluence, Git, BigQuery, Databricks, Postgres, Snowflake, and MCP servers
- Supports Anthropic Claude models from launch, with open source and other private models coming
- Has its own Workspace account with an email address, calendar context, and org-level awareness
- Accessible via iOS, Android, Windows, Mac, CLI, Google Workspace, Microsoft 365, ServiceNow, and Slack
- Writes audit trails attributed to the agent, not to individual users
There’s no public pricing attached yet beyond the mention of flexible enterprise spending controls. Availability is rolling out to Gemini Enterprise customers first. If you’re already in that ecosystem, it’s worth requesting access early. If you’re evaluating enterprise AI platforms and haven’t looked at Gemini in the last six months, this release changes the comparison.
