README.md

Lesson 1 - Why Semantic Kernel? - Building our first App?

What will you learn in this lession?

• Why is an SDK like Semantic Kernel Needed?
• What is required to build a basic Chat Bot with Semantic Kernel?

Why is this important?

It's important to understand why a framework is needed for AI based solutions and what are the fundemental building blocks needed for a basic Semantic Kernel Chat Bot. Once you understand this basics, you are start to leverage the more complex features.

Why Semantic Kernel?

What is needed to build our first App?

5 Simple steps are needed

1. Create a Kernel Builder

         var builder = Kernel.CreateBuilder();
   

2. Load you AI Endpoint Details

        var openAiDeployment = ConfigurationManager.AppSettings.Get("AzureOpenAIModel");
        var openAiUri = ConfigurationManager.AppSettings.Get("AzureOpenAIEndpoint");
        var openAiApiKey = ConfigurationManager.AppSettings.Get("AzureOpenAIKey");
   

3. Add the Chat Completion Service

       builder.Services.AddAzureOpenAIChatCompletion(
            deploymentName: openAiDeployment!,
            endpoint: openAiUri!,
            apiKey: openAiApiKey!);
   

4. Construct the Kernel, ChatHistory, get instance of the ChatCompletion Service

        var kernel = builder.Build();
        ChatHistory history = [];
        var chatCompletionService = kernel.GetRequiredService<IChatCompletionService>();
   

5. Send a prompt or ChatHistory and get a response from LLM

       var prompt = "Why is the Sky blue?";
       var result = await chatCompletionService.GetChatMessageContentAsync(prompt);
       Console.WriteLine(result);
   

Let's build your first .NET Core SK Console App

1. Create a new .NET Core Console App in Visual Studio and name it SK-Lesson-1

2. Add the Semantic Kernel Package to the project.

   • Click on Dependecies->Manage NuGet Packages and inssue the package below
   Packages (click to expand)

    Microsoft.SemanticKernel 1.6.3 or better
   

3. Come up with some way to store and read the Model Name, AI Key and AI Endpoint. Here is an example of using the System.Configuration class and an App.Config file.

   • Create an App.Config file in the root of the project using the following format and replace the values to point to your AI Endpoint

      <?xml version="1.0" encoding="utf-8" ?>
      <configuration>
          <appSettings>
    	      <add key="AzureOpenAIEndpoint" value="AzureOpenAI-Endpoint-URI" />
              <add key="AzureOpenAIKey" value="AzureOpenAI KEY" />  
              <add key="AzureOpenAIModel" value="AzureOpenAI Model Name" />
          </appSettings>
      </configuration>
   

   • Add a reference to the System.Configuration assembly to the program.cs file

4. Using the steps outlined in the 5 Simple steps build and test your first app.

   • If needed you can take a look at the soltion to see a fishined project, but ideally you should put hands on keyboard and work through any issues you come across.

Finish your homework before the next lesson

🔼 Home | Next Lesson 2 ▶