# What is Function Calling? Definition & Technical Mechanics

> **Source:** [SPIDITS AI AI Glossary](https://spidits.com/ai-glossary)  
> **Category:** Agentic Systems  
> **Canonical Citation:** [https://spidits.com/ai-glossary/function-calling](https://spidits.com/ai-glossary/function-calling)

## Definition
Function Calling is an LLM capability where the model outputs a structured JSON object containing argument parameters to invoke specific external functions or APIs, enabling LLMs to act as dynamic interfaces for databases and systems.

## Architectural Mechanics & Deep Dive
Function calling is a capability in advanced Large Language Models where the model identifies when it needs to run an external tool and outputs a structured JSON payload containing the function name and arguments. This payload is executed by the client application (e.g., querying a database or API), and the results are fed back to the LLM, enabling real-time data access.

## Industry Applications & Real-World Use Cases
Database access tools, external integrations, and agent workflows

## Frequently Asked Questions

### Does the LLM run the function itself during function calling?
No, the LLM only generates the structural instructions (JSON argument values). The client application executes the actual code and returns the result back to the model.

### Why is function calling critical for agents?
It allows models to bypass text-only limits and write structured instructions to retrieve weather data, execute database queries, or write local files.

## Related Terminology
- [ai-agent](https://spidits.com/ai-glossary/ai-agent)
- [mcp](https://spidits.com/ai-glossary/mcp)
- [agentic-ai](https://spidits.com/ai-glossary/agentic-ai)

---
*From the SPIDITS AI Knowledge Base (336+ terms indexed). Explore full technical definitions: [https://spidits.com/ai-glossary](https://spidits.com/ai-glossary)*