Google Cloud has launched Gemini agent, an autonomous AI assistant designed to operate across web, mobile, and desktop platforms. The tool integrates with enterprise software including Google Workspace, Microsoft 365, Slack, Salesforce, and ServiceNow, as well as databases like BigQuery, Databricks, and Snowflake.
The system utilizes a multi-model orchestration approach, selecting specific models such as Gemini Flash or Argon based on task complexity. It also supports third-party models, including Anthropic’s Claude. Users can create and share custom reusable skills for multi-step workflows within a company registry.
To manage enterprise spending, Google introduced flexible pricing options and automated workload routing to optimize costs. The Cloud Billing Console allows administrators to set hard budget limits on AI usage, with the ability to track and audit costs at the departmental or project level.
The agent maintains persistent memory and context across communication channels. It can generate sub-agents with dedicated identities and storage, while grounding its responses in institutional data such as product portfolios and departmental norms.
