Function Calling

A model capability that lets an LLM emit a structured call to a developer-defined function - the mechanism underneath tool use and agents.

Function calling lets a model output a structured, schema-conformant request to invoke a named function with arguments, instead of only free text. The application runs the function and returns the result for the model to reason over. It is the low-level mechanism that turns a text generator into something that can act, and it underlies both tool calls and the Model Context Protocol.

Because a function call can read data or trigger a real action, it is the highest-leverage and highest-risk thing a model does. Each call is a natural place to authorize against policy, scope to the caller's purpose, and record - so a model can only invoke what it is entitled to, with the arguments and the outcome captured as evidence.

All terms

Govern AI like infrastructure.

Talk to our team about deploying DataStrict across your enterprise stack.