Google Agentspace: Closer look at Google’s new AI platform

Google’s Agentspace, announced at Cloud Next 2025, is the company’s latest step into enterprise AI. Instead of being just another chatbot, it’s pitched as a secure hub where companies can build and manage AI agents that handle real work across different tools.

The big appeal is search. Agentspace can connect to systems like Google Drive, Jira, Salesforce, and Slack, then answer questions across them in plain language. Ask about the status of a project, and it pulls information together rather than just surfacing documents. Early users say it feels more like a helpful assistant than a search box.

Google also includes a few starter agents, like a research assistant and an enterprise version of NotebookLM for conversational data analysis. More interesting, though, is the option for teams to build their own. Non-technical staff get a drag-and-drop builder, while developers can go deeper with Vertex AI. That flexibility could make it easier for companies to experiment without waiting on IT.

Google is also trying something bold with Agent2Agent, a protocol that lets agents interact. In theory, this means a marketing agent could kick off a task with a sales agent, or an internal tool could pass work to an external one. It’s a clever idea, but still untested at scale.

On the practical side, Agentspace benefits from Google Cloud’s security and compliance standards, which is critical for industries like finance and healthcare. Banco BV, a Brazilian bank, has already piloted it to let employees query sensitive data safely.

There are caveats. Pricing starts around $25 per user per month, which adds up quickly. Real-world deployments are limited, so it’s too early to know how well it scales. And for organizations already using Microsoft Copilot or ServiceNow’s Now Assist, there may be overlap.

Overall, Agentspace feels less like a direct rival to Copilot and more like an attempt to create a new category: a platform for managing many agents across a company. The vision is ambitious, and if it works, it could reshape how enterprises adopt AI. For now, it looks promising but unproven – best suited to small pilots in areas like knowledge management or research before any company-wide rollout.

Strengths

• Integration with Google’s ecosystem: For enterprises already invested in Google Workspace and Vertex AI, Agentspace feels like a natural extension.
• Balance of accessibility and power: Business users can configure simple agents, while IT and data science teams can push further.
• Governance-first approach: Unlike many AI “chatbot” tools, this is built with enterprise controls in mind from the outset.
• Future-facing multi-agent vision: Agent2Agent interoperability suggests Google is thinking about scale and long-term complexity.