C# and Neural Networks: How to Connect ChatGPT to a Project

Why C# and AI Are the Perfect Pair?

When we talk about artificial intelligence and development, Python usually comes to mind. That's logical: TensorFlow, PyTorch, transformers... But what if your stack is C#, .NET, and you don't want to change your familiar toolkit? The good news: C# and neural networks combine perfectly, especially when it comes to using ready-made models via an API.

ChatGPT is not just a trendy toy. It's a powerful tool that can generate code, write documentation, analyze data, and even help with debugging. In this article, we'll break down how to connect ChatGPT to a C# project, write real code, and see what opportunities open up for .NET developers.

What You Need to Know Before Starting?

To work with the OpenAI API in C#, you will need:

\n
• An account on platform.openai.com (an API key is required)
\n
• Visual Studio 2022 or JetBrains Rider
\n
• .NET 6.0 or higher (.NET 8 is recommended)
\n
• The OpenAI library (NuGet package)
\n

Don't be scared: the whole process will take no more than 15 minutes. Let's go step-by-step through creating a console application that will communicate with ChatGPT.

Step 1. Create the Project and Install the Library

Open your terminal or Package Manager Console and run the command:

dotnet new console -n ChatGPTDemo\ncd ChatGPTDemo\ndotnet add package OpenAI --version 1.11.0\n

The OpenAI package for C# is an official client that supports all the latest models, including GPT-4 and GPT-4 Turbo. It's actively developed, so it's best to use the latest version.

Setting Up the API Key

Never store keys in your code. Use environment variables or a configuration file. For this example, let's create appsettings.json:

{\n  "OpenAiApiKey": "your_super_secret_key"\n}\n

Add the NuGet package Microsoft.Extensions.Configuration.Json to the project and load the key like this:

var config = new ConfigurationBuilder()\n    .AddJsonFile("appsettings.json")\n    .Build();\n\nstring apiKey = config["OpenAiApiKey"];\n

Now we are ready for the most interesting part.

Step 2. First Request to ChatGPT in C#

Let's write a simple method that sends a message to the model and gets a response. Here is the full code:

using OpenAI;\nusing OpenAI.Chat;\n\nvar client = new OpenAIClient(apiKey); // apiKey from config\n\nvar chatClient = client.GetChatClient("gpt-4o-mini");\n\nvar messages = new List<ChatMessage>\n{\n    new SystemChatMessage("You are a helpful assistant who responds in Russian."),\n    new UserChatMessage("Write an example of C# code for reading a JSON file.")\n};\n\nvar response = await chatClient.CompleteChatAsync(messages);\nvar reply = response.Value.Content[0].Text;\n\nConsole.WriteLine(reply);\n

What's happening here? We create a client, specify the gpt-4o-mini model (fast and cheap), set a system prompt (the assistant's role), and send the user's request. The response comes back as structured text.

This is the basics. From here, you can make it more complex: add conversation history, handle errors, make asynchronous calls.

Step 3. Creating a Smart Assistant for Working with Code

One of the most useful functions of ChatGPT in C# development is code review and documentation generation. Let's create a class that will analyze our code and suggest improvements.

public class CodeReviewer\n{\n    private readonly ChatClient _chatClient;\n\n    public CodeReviewer(string apiKey)\n    {\n        var client = new OpenAIClient(apiKey);\n        _chatClient = client.GetChatClient("gpt-4o-mini");\n    }\n\n    public async Task<string> ReviewCodeAsync(string code)\n    {\n