TypeScript and ChatGPT: Building an AI Assistant from Scratch

Why TypeScript and ChatGPT Are the Perfect Pair for Modern Developers

Imagine: you're writing code in TypeScript, and suddenly you need to quickly generate a complex SQL query, rewrite outdated JavaScript, or even create a full-fledged chatbot for user support. Previously, this took hours, but now it takes minutes. All thanks to the fact that TypeScript and ChatGPT have become not just trendy tools, but a real workhorse for thousands of developers worldwide.

In this article, we won't give boring lectures. We'll just do it: write code that actually works. You'll learn how to make TypeScript and the OpenAI API work together, create your first AI assistant, and understand how to use ChatGPT in development to automate routine tasks. Let's go!

What You'll Need to Get Started

Before we dive into the code, make sure you have:

\n
• Node.js version 18+ (we'll use modern fetch)
\n
• npm or yarn for installing packages
\n
• An API key from OpenAI — get it at platform.openai.com (it's free if you're just starting out)
\n
• A basic understanding of TypeScript: types, interfaces, async/await
\n

We'll do everything else together. Even if you've never worked with neural networks — don't worry, I'll explain every step.

Step 1: Setting Up a TypeScript Project for AI Work

Create a new folder and initialize the project:

mkdir ai-chatbot\ncd ai-chatbot\nnpm init -y\nnpm install typescript @types/node ts-node --save-dev\nnpx tsc --init

Now let's install the client for OpenAI. Previously, everyone used the openai library, but in 2024-2025, OpenAI recommends its official SDK:

npm install openai dotenv

Create a .env file in the project root and add your key there:

OPENAI_API_KEY=your_secret_key_here

Important: Never commit .env to the repository! Add it to .gitignore.

Step 2: Writing Your First TypeScript Function to Communicate with ChatGPT

Create a file chat.ts. We'll start simple: send a message to the neural network and get a response. Here's how it looks:

import OpenAI from 'openai';\nimport dotenv from 'dotenv';\n\ndotenv.config();\n\nconst openai = new OpenAI({\n  apiKey: process.env.OPENAI_API_KEY,\n});\n\ninterface ChatResponse {\n  content: string | null;\n  error?: string;\n}\n\nasync function askChatGPT(prompt: string): Promise<ChatResponse> {\n  try {\n    const response = await openai.chat.completions.create({\n      model: 'gpt-4o-mini', // fast and cheap model\n      messages: [\n        {\n          role: 'system',\n          content: 'You are a helpful assistant for a TypeScript developer. Answer concisely with code examples.',\n        },\n        {\n          role: 'user',\n          content: prompt,\n        },\n      ],\n      max_tokens: 500,\n      temperature: 0.7,\n    });\n\n    return {\n      content: response.choices[0]?.message?.content || null,\n    };\n  } catch (error) {\n    console.error('Error when requesting OpenAI:', error);\n    return {\n      content: null,\n      error: 'Failed to get a response from the neural network',\n    };\n  }\n}\n\n// Example usage\nasync function main() {\n  const result = await askChatGPT('Write a TypeScript function that checks if a string is a palindrome');\n  \n  if (result.content) {\n    console.log('ChatGPT response:');\n    console.log(result.content);\n  } else {\n    console.error('Error:', result.error);\n  }\n}\n\nmain();

Run the script with the command npx ts-node chat.ts. If everything is set up correctly, you'll see the neural network's response with code and explanations in the console. ChatGPT integration