Google unveils enterprise AI agent for workflows across Workspace and third-party apps

Google has unveiled an enterprise AI agent that plans and executes multi-step workflows across business applications, according to Adgully. Announced at a Google Cloud event, it coordinates subagents and accesses internal databases, corporate memory, and specialized tools. Integrations span Google Workspace, Microsoft 365, Slack, Jira, Confluence, Git, BigQuery, Databricks, Postgres, Snowflake, and Model Context Protocol servers. The report names Shopify, PayPal, On, BNP Paribas, and Merck as early enterprise adopters, without providing deployment scope or performance results.

The market significance is Google's move from conversational assistance toward delegated execution across competing software ecosystems. The agent operates with a dedicated Workspace account, email address, and audit trail, giving automated work a distinct organizational identity. It can select Gemini or third-party models such as Anthropic's Claude, while administrators control model routing and spending. This positions Google as an orchestration and distribution layer for enterprise agents, where access to workplace systems and administrative controls may matter alongside model capability. The report does not establish whether these workflows deliver reliable production outcomes.

For builders and investors, the concrete question is whether independent orchestration products offer capabilities beyond those bundled into an enterprise platform. Google's task inbox lets employees monitor code execution, tool use, and delegation, making oversight part of the product surface. Buyers should test end-to-end completion rates, permission boundaries, and total task costs before treating the named adopters as evidence of repeatable returns. The reported enterprise-first approach puts those operating requirements ahead of a planned consumer rollout.