Function Calling

Function calling, also known as tool use, is the mechanism that lets a language model interact with external code and services. The developer describes a set of available functions, and the model decides when one is needed and returns a structured JSON object naming the function and its arguments. Crucially, the model does not execute anything itself — it only signals that a call should be made, and the surrounding application runs the actual function.

How it works

Each tool is described using a JSON schema that specifies the function name and the shape of its arguments. The schema is passed to the model alongside the prompt. When the model determines a tool is appropriate, it produces a JSON payload conforming to that schema, which the application parses and uses to call the real function — querying a database, hitting an API, controlling hardware. The function's result is then fed back to the model so it can continue reasoning. Strict modes can enforce that the emitted JSON exactly matches the supplied schema, removing a common source of parsing errors.

Agents and sovereignty

Function calling is the foundation of AI agents: a model that can call tools can take multi-step actions rather than only producing text. For a sovereign operator, this means a locally hosted model can be wired to local tools and private data sources, so an assistant can act on a homelab or mining setup without any tool inputs or outputs traversing a third-party service. Several open-weight models support function calling, making fully self-hosted agents realistic.

D-Central documents function calling as the action layer of a local AI stack. It pairs naturally with retrieval tools built on semantic search and reranking, letting an agent look things up before it acts.