Tool Calling

An agentic pattern where LLMs invoke external functions or APIs sequentially, receiving results back through the context window to inform subsequent reasoning steps.

Tool calling is the dominant paradigm for AI agent execution, where the model decides which external function to invoke, receives the result, reasons about it, and then decides on the next action. Each tool invocation requires a full roundtrip through the model's context window, which can accumulate tokens and add latency for multi-step tasks. Alternative approaches like code generation aim to reduce these roundtrips by having the LLM write executable scripts upfront.

Also known as