Concept·Patterns & Practices·Added 1 month ago

Function Calling

Also known as: tool calling, function use

A mechanism where a model can decide to invoke an external function or API instead of generating text. The model outputs a structured request (which function to call and with what arguments); your code runs the function and feeds the result back to the model.

Function calling bridges the gap between a model that generates text and a system that does things in the world. You tell the model what functions are available, what arguments each one takes, and what it returns. The model then decides, mid-response, to pause and invoke a function rather than answer from its training data alone.

In practice, function calling is how agents use tools. When a model decides to search the web, query a database, or call your internal API, it does so via function calling. The model produces a structured request; your application executes it; the result comes back into the model's context. This loop can repeat many times in a single agent run.

Function calling and tool use are often used interchangeably, though tool use is sometimes the broader term. The key concept is the same: the model is not just generating text, it's making structured requests to external capabilities, which is the foundation of any real agentic workflow.

This definition is AI-generated and refreshed weekly. It may contain inaccuracies. Use your own judgment, especially for production decisions.

Related terms