How to use OpenAI function calling to build structured AI workflows

What function calling is and why it matters

Function calling allows developers to describe custom tools (functions) to OpenAI models using JSON Schema formats. Instead of returning unstructured prose text, the LLM outputs valid JSON parameters specifying which function to invoke.

Basic function calling setup

Define tool definitions in your OpenAI API request payload: ``typescript const tools = [ { type: "function", function: { name: "getUserOrders", description: "Retrieve recent orders for a client by user ID", parameters: { type: "object", properties: { userId: { type: "string", description: "The unique customer identifier" }, limit: { type: "number", description: "Number of orders to return" } }, required: ["userId"] } } } ]; ``

Error handling and fallback logic

LLMs may occasionally hallucinate invalid arguments or fail JSON schema validation. Always validate function arguments using Zod schemas before executing database or payment API operations.