Function Calling | HyperKit.ai

Definition

Function calling allows LLMs to output structured requests to execute external functions. Instead of generating text, the model outputs a function name and arguments in a structured format that your code can execute.

This enables LLMs to interact with external systems, APIs, databases, and tools in a controlled, reliable way.

Key Concepts

Function schema: JSON definition of available functions
Structured output: Model outputs function calls, not prose
Execution loop: Model requests → you execute → return results
Tool choice: Control when model can/must use functions

Examples

Flow

Function Calling Pipeline

User: "What's the weather in Tokyo?"

Step 1: Define functions for the model
functions = [{
    "name": "get_weather",
    "description": "Get current weather for a city",
    "parameters": {
        "type": "object",
        "properties": {
            "city": {"type": "string"},
            "units": {"type": "string", "enum": ["celsius", "fahrenheit"]}
        },
        "required": ["city"]
    }
}]

Step 2: Model outputs function call (not text!)
{
    "function_call": {
        "name": "get_weather",
        "arguments": "{\"city\": \"Tokyo\", \"units\": \"celsius\"}"
    }
}

Step 3: Your code executes the function
result = get_weather("Tokyo", "celsius")
# Returns: {"temp": 22, "condition": "cloudy"}

Step 4: Send result back to model
Model generates: "The weather in Tokyo is 22°C and cloudy."

Implementation

OpenAI Function Calling

from openai import OpenAI
import json

client = OpenAI()

# Define tools
tools = [{
    "type": "function",
    "function": {
        "name": "search_products",
        "description": "Search for products in the catalog",
        "parameters": {
            "type": "object",
            "properties": {
                "query": {"type": "string"},
                "max_price": {"type": "number"},
                "category": {"type": "string"}
            },
            "required": ["query"]
        }
    }
}]

# Call with tools
response = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "Find laptops under $1000"}],
    tools=tools,
    tool_choice="auto"  # or "none" or {"type": "function", "function": {"name": "..."}}
)

# Handle function call
if response.choices[0].message.tool_calls:
    call = response.choices[0].message.tool_calls[0]
    args = json.loads(call.function.arguments)
    # Execute: search_products(**args)

Interactive Exercise

✎

Design a Function Schema

Design a function schema for a "send_email" function that an LLM could call:

Pro Tips

Write clear descriptions—the model uses them to decide when to call
Use enums for constrained parameters (status, category, etc.)
Always validate arguments before execution
Consider parallel function calls for efficiency

Definition

Key Concepts

Examples

Interactive Exercise

Related Terms