Google expands Managed Agents in Gemini API

Google expanded Managed Agents in the Gemini API on July 7, 2026, introducing four core capabilities designed to support developers building reliable, long-running agent systems Google.

New Developer Capabilities

The update adds background execution, allowing agents to perform tasks without blocking user interactions; remote MCP server integration, enabling direct connections to Model Context Protocol servers for private databases and internal APIs; custom function calling for agent-specific workflows; and credential refresh to maintain authenticated access across multiple interactions Google.

Google's official statement emphasizes that remote MCP integration eliminates the need for custom proxy middleware. "Instead of writing custom proxy middleware to access private databases or internal APIs, you can now connect managed agents directly to remote Model Context Protocol (MCP) servers," the company said Google.

Production-Ready Architecture

These features address a core friction point for enterprise adoption: orchestration complexity in agent systems. Background execution enables agents to perform time-consuming tasks—data processing, API polling, file uploads—without degrading response time. Credential refresh ensures agents maintain access to protected resources across sessions, critical for enterprise deployments where authentication tokens expire.

The MCP server integration is particularly significant for enterprise developers. Rather than maintaining middleware layers that translate between agent frameworks and legacy systems, developers can now route agent requests directly to MCP-compliant endpoints. This reduces infrastructure overhead and attack surface area in agent orchestration.

Developer Access

Developers can access Managed Agents through the Gemini API documentation and AI Studio interface Google. The capabilities are available as part of Google's broader Gemini Enterprise Agent Platform, which provides governance and deployment tools for teams building scalable agent systems.

This release reflects growing developer demand for stateful, long-running agents that integrate with existing enterprise infrastructure. The timing aligns with increased enterprise adoption of agentic systems and competitive pressure from other API providers expanding agent orchestration features.