Build managed agents with the Gemini API (blog.google)

Google has announced the launch of Managed Agents within the Gemini API, allowing developers to create agents that can reason, execute code, and use tools in an isolated Linux environment with just a single API call. Powered by the new Antigravity agent built on Gemini 3.5 Flash, this feature simplifies the complexities of infrastructure management, enabling developers to focus on enhancing product experiences and agent behaviors. Initially introduced with the Deep Research managed agent for complex tasks, this update opens the door for creating custom agents tailored to specific use cases. The Managed Agents facilitate capabilities such as planning, web browsing for live data retrieval, and session continuity by preserving the environment state across interactions. Developers can easily extend the Antigravity agent using markdown files, eliminating the need for intricate orchestration code. This innovation not only accelerates development but also supports enterprises through the Gemini Enterprise Agent Platform. With access to robust agent infrastructure, the AI/ML community is poised to leverage these tools for innovative applications, reinforcing Google’s commitment to advancing AI capabilities.