Skip to main content

Tool Use (Function Calling)

Tool use is a language model's ability to call external functions in a structured way while forming an answer, for example a search, a calculation, or a database query.

Tool use, also called function calling, allows a language model to request a structured call to an external function.

This may be a search, calculation, database query or file operation. The tool name and parameters are returned in machine-readable form; the surrounding application controls actual execution.

How tool use works technically

The application provides the model with descriptions of available tools and their parameters. When the model selects a tool, it produces a structured call with the required arguments.

The application validates the call, decides whether to execute it and then returns the result to the model as additional context.

Why this gives a language model the ability to act

With tool use, an AI application can retrieve current data, use specialised functions and initiate actions in external systems. Repeated tool calls are a core component of agentic AI.

The connection to MCP

MCP standardises part of this integration. Regardless of the protocol, every tool needs clearly bounded permissions, logging and explicit approval for critical actions.

Further reading

Would you like to apply this in your organisation?

Discuss your project