Google

Google Updates Gemini API for Advanced Agentic Workflows


Executive Summary

Google has announced significant updates to the Gemini API tooling to simplify the development of complex, agentic AI applications. Developers can now combine built-in tools like Google Search and Google Maps with their own custom functions in a single API call, reducing latency and architectural complexity. The update also introduces context circulation for multi-step tasks and expands Google Maps grounding to the Gemini 3 model family, enabling more sophisticated and location-aware AI agents.

Key Takeaways

* Combined Tool Calls: Developers can now use both built-in tools (e.g., Google Search) and custom-defined functions within the same API request, streamlining orchestration.

* Context Circulation: The model can preserve the output from one tool call and use it as input for another tool in a subsequent step, enabling more complex reasoning workflows.

* Google Maps Grounding for Gemini 3: The Gemini 3 family of models now supports grounding with Google Maps, allowing them to access rich spatial data, business information, and commute times for location-aware responses.

* Tool Response IDs: Unique identifiers are now assigned to each tool call to improve debugging and the management of asynchronous or parallel function executions.

* New Interactions API: Google recommends using its new `Interactions API` for these workflows to take advantage of server-side state management.

Strategic Importance

This update lowers the barrier for developers to build sophisticated AI agents, making the Gemini API a more competitive platform for creating complex applications that can reason and interact with multiple data sources.

Original article