Function CallingTool Use
Letting an AI invoke real code and APIs mid-reasoning.
Function calling (also called tool use) is a model capability where the LLM emits a structured "call this function with these arguments" signal mid-response — rather than just generating prose. The application runs the function, returns the result, and the model incorporates it into its next prediction.
This is what makes agents possible: the model can call a search API, read results, run a calculation, check the output, and weave everything into a coherent response — all in one turn.
Most frontier models support this: Anthropic calls it "tool use," OpenAI calls it "function calling." Both expose it via API as a special message type.
In plain terms
Like a consultant who mid-meeting says "hold on, let me pull the live data" — checks real numbers, then continues the conversation with actual facts.