What is Tool Calling?
Tool Calling: A model capability where the LLM returns a structured request to run a defined function, like an API call, which your code executes and returns.
Tool calling (also called function calling) lets a model use external capabilities. You describe the available tools to the model with a name, a description, and a JSON schema for the parameters. When the model decides a tool would help, it returns a structured call instead of plain text. Your code runs the tool, sends the result back, and the model continues with that information.
The model never executes anything itself. Your application decides whether to run each call, with which credentials, and what to return. That is where safety lives: validating arguments, enforcing permissions, rate limiting, and requiring approval for sensitive actions. Tool descriptions are effectively part of the prompt, so clear names and descriptions make a large difference to how reliably the model picks the right tool.
Why it matters when you build or buy AI
Tool calling is what turns a chatbot into something that does work. It is also where real damage happens if you are careless: a tool called "update_order" with no limits lets the model change anything. Design tools narrowly (refund up to a set amount, only for this customer's orders), make write actions idempotent, and keep humans on anything irreversible.
Example
Given "cancel my order 1042," the model calls get_order(order_id: 1042), sees it has not shipped, then calls cancel_order(order_id: 1042), which the app executes only after checking the requester owns that order.
Related terms
More in Agents & Automation
Book 20 minutes, or send a brief
A former CTO replies. You leave with a written plan and a fixed estimate.
Prefer LinkedIn? Connect with Harish.