DIGITAL KITCHEN
Back to glossaryGLOSSARY · Concepts

Function Calling

A specific implementation of tool use where the LLM emits structured JSON describing which function to call and with what arguments. The runtime then executes the function and feeds the result back to the LLM. The dominant pattern in 2026 frontier-model APIs.

How it works

All major LLM providers (Anthropic, OpenAI, Google) standardised on JSON-schema-defined function calling by 2025. The LLM is given function schemas, decides if and which to call, returns the call as structured JSON, the runtime executes, the result returns as a tool_result message.

Example

In the Anthropic SDK: client.messages.create({ tools: [{ name: "search", input_schema: {...} }] }). When the LLM wants to search, it returns content with type "tool_use" containing the function name and parsed arguments.

Related terms

Tool Use

Need to actually use Function Calling?

We build production AI systems that put these concepts to work. 30 minutes, we map your use case.