Definition
Tool calling is the ability of an AI model to request a defined function, such as searching a database or booking a meeting, which the application then validates and executes.
The model only requests the call. The application decides whether to run it, after checking the tool is allowed, the arguments are valid and the user has confirmed any external action.
Why it matters
Tool calling is what turns a model that can only talk into a system that can act: look up an order, create a ticket, schedule a meeting. It is the building block of every AI agent.
It is also where most of the risk sits. Because the application, not the model, executes the call, the application is where permissions, validation and confirmations belong. A model should never hold credentials directly.
An example
A user asks an assistant to book a meeting with a supplier next Tuesday. The model requests a call to a calendar tool with a date, time and attendee. The application checks the user may book on that calendar, confirms the slot is free, shows the user the invitation, and only sends it after they confirm.
Illustrative example, not a client project.
Common questions about Tool calling
Is tool calling the same as function calling?
Yes. Different model providers use different names for the same idea: the model returns a structured request to run a named function with arguments.
Can a model call any tool it wants?
Only the tools the application offers it, and a well-built application validates every request before running it.
Read more
Related terms