Exercise - Use nested functions for song suggestions

• 10 minutes

For this exercise, you combine your native functions with a prompt that asks the LLM to generate a recommended song for the user based on their recent plays. Let's get started!

Prepare your development environment

For these exercises, a starter project is available for you to use. Use the following steps to set up the starter project:

1. Open Visual Studio Code.

2. Under the Visual Studio Code Start section, select Clone Git Repository.

3. In the URL bar, enter https://github.com/MicrosoftLearning/MSLearn-Develop-AI-Agents-with-Azure-OpenAI-and-Semantic-Kernel-SDK.git

4. In the File Explorer, create a new folder in a location that is easy to find and remember, such as a folder in your Desktop.

5. Click the Select as Repository Destination button.

   You need to be signed in to GitHub to successfully clone the project.

6. Open the project in Visual Studio Code.

7. In Explorer, right-click the M04-combine-prompts-and-functions/M04-Project folder and click Open in Integrated Terminal.

8. Expand the M04-combine-prompts-and-functions/M04-Project folder.

   You should see a "Program.cs" file.

9. Open the Program.cs file and update the following variables with your Azure OpenAI Services deployment name, API key, endpoint.

   string yourDeploymentName = "";
   string yourEndpoint = "";
   string yourKey = "";
   

Now you're ready to begin the exercise. Good luck!

Provide personalized song recommendations

1. In your MusicLibraryPlugin.cs file, add the following function:

   [KernelFunction, Description("Get a list of music available to the user")]
   public static string GetMusicLibrary()
   {
       string dir = Directory.GetCurrentDirectory();
       string content = File.ReadAllText($"{dir}/data/musiclibrary.txt");
       return content;
   }
   

2. Update your 'Program.cs' file with the following code:

   var kernel = builder.Build();
   kernel.ImportPluginFromType<MusicLibraryPlugin>();
   
   string prompt = @"This is a list of music available to the user:
       {{MusicLibraryPlugin.GetMusicLibrary}} 
   
       This is a list of music the user has recently played:
       {{MusicLibraryPlugin.GetRecentPlays}}
   
       Based on their recently played music, suggest a song from
       the list to play next";
   
   var result = await kernel.InvokePromptAsync(prompt);
   Console.WriteLine(result);
   

   In this code, you combine your native functions with a semantic prompt. The native functions are able to retrieve user data that the large language model (LLM) couldn't access on its own, and the LLM is able to generate a song recommendation based on the text input.

3. To test your code, enter dotnet run in the terminal.

   You should see a response similar to the following output:

   Based on the user's recently played music, a suggested song to play next could be "Sabry Aalil" since the user seems to enjoy pop and Egyptian pop music.
   

   Note

   Your generated song recommendation may be different than the one shown here.

You successfully combined your native functions with a semantic prompt. You have the beginnings of a music recommendation agent! Try playing around with the prompts and input files to see what other recommendations you can generate.