Google lets Gemini API combine Search, Maps and custom functions in a single request

Google has expanded the Gemini API: built-in tools such as Google Search and Google Maps can now be combined with custom functions in a single request. This removes some of the manual scaffolding around agent workflows and moves the API closer to a model that coordinates search, code calls and transitions between steps on its own.

What changed

The update, announced by Google in March 2026, gives the Gemini API more cohesive tool handling. Previously, developers often had to organize search separately, set up a separate route to maps or external data, and then stitch the responses together by hand in the application. Now part of that logic can live inside a single call: the model has access to Google's built-in services and can also invoke custom functions when a scenario requires actions beyond the standard toolset.

The practical walkthrough shows five demos that gradually raise the complexity. It starts with a basic tool combination, then moves to multi-step agent chains, where passing context between stages, identifying parallel calls and correctly continuing the conversation after an external function responds all matter. For developers this is an important shift: orchestration is less spread across the backend and moves more into the model session itself. That makes the chain's behavior noticeably more transparent for debugging and testing.

How the call works

The key idea is that Gemini can now decide in a single pass when it needs a web search, when it needs geodata from Maps, and when it needs an application's custom function. If a question requires several actions, the model can hold shared context, compare results from different tools and continue the chain without fully restarting the scenario. The walkthrough puts particular emphasis on parallel tool IDs and context circulation: they help avoid mixing up tool responses and carry the needed data into the next step.

• Google Search pulls in fresh information on the topic of the query
• Google Maps adds addresses, geographic context and place data
• Custom functions plug in the application's internal business logic
• Multi-step chains let you build scenarios from several sequential actions

This setup suits scenarios where no single source can provide the answer. For example, an assistant can first find fresh information via Search, then check a location on the map via Maps, and then call an internal function for booking, calculation or availability checks. Previously, such a sequence more often had to be split across several requests, with state maintained manually between them. For services that take real actions, this noticeably simplifies the architecture.

Why it matters to developers

The main benefit is less glue between the model and product logic. Instead of writing a separate orchestrator for each combination of tools, a team can describe its functions, give the model access to the services it needs and build more natural agent workflows on top. This is especially useful for assistants that must do more than answer in text and actually carry out tasks: look up data, pick a place on the map, check parameters, pass the result to an internal service and only then compose the final answer for the user.

Another advantage is more predictable scaling of complex chains. When Google's built-in tools and a company's own functions are connected in one process, there are fewer intermediate layers where context is usually lost or call logic breaks. For teams, that means faster prototypes and less auxiliary code around them. For products, it means a chance to move more quickly from a chatbot with suggestions to an agent that can actually see a task through to a result.

What it means

Google is steering the Gemini API toward full-fledged agent interfaces, where the model does not just generate text but manages a set of tools within one session. If this approach proves stable in real production scenarios, developers will be able to build useful AI assistants with less manual orchestration and ship them faster.