Tutorial: Create a recommendation app with .NET MAUI and ChatGPT

Article
04/12/2024

In this tutorial, you'll learn how to create a .NET MAUI app for Windows in Visual Studio that calls OpenAI's ChatGPT APIs to provide recommendations based on a location entered by the user. The app will have a simple UI that allows the user to enter a location and get recommendations for restaurants, hotels, and attractions.

In this tutorial, you learn how to:

Create a simple user interface for your .NET MAUI app
Reference and bootstrap the Azure.AI.OpenAI library
Use an API key to link your app to an OpenAI API account
Make calls to the OpenAI chat APIs to get recommendations

Prerequisites

An OpenAI account
An OpenAI API key
The .NET MAUI installation requirements
If you are new to .NET MAUI on Windows, you should start with the Build your first .NET MAUI app for Windows tutorial.

Set your environment variable

In order to use the OpenAI SDK, you'll need to set an environment variable with your API key. In this example, we'll use the OPENAI_API_KEY environment variable. Once you have your API key from the OpenAI developer dashboard, you can set the environment variable from the command line as follows:

setx OPENAI_API_KEY <your-api-key>

Note that this method works for development on Windows, but you'll want to use a more secure method for production apps and for mobile support. For example, you can store your API key in a secure key vault that a remote service can access on behalf of your app. See Best practices for OpenAI key safety for more information.

Create a new .NET MAUI project with the required UI elements

We're going to start by creating a new .NET MAUI project in Visual Studio. We'll use the .NET MAUI App template and add some UI elements to the MainPage to provide users with some recommendations based on a provided location. The UI will have buttons to get recommendations for restaurants, hotels, and attractions.

In Visual Studio, create a new .NET MAUI App project named ChatGptRecommendationApp.
Run the new project to make sure the app builds and runs successfully.
Open MainPage.xaml from the Solution Explorer.

Replace the contents of the VerticalStackLayout with the following XAML markup:

<Label
    Text="Local AI recommendations"
    SemanticProperties.HeadingLevel="Level1"
    FontSize="32"
    HorizontalOptions="Center" />

<Entry
    x:Name="LocationEntry"
    Placeholder="Enter your location"
    SemanticProperties.Hint="Enter the location for recommendations"
    HorizontalOptions="Center"/>

<Button
    x:Name="RestaurantBtn"
    Text="Get restaurant recommendations"
    SemanticProperties.Hint="Gets restaurant recommendations when you click"
    Clicked="OnRestaurantClicked"
    HorizontalOptions="Center" />

<Button
    x:Name="HotelBtn"
    Text="Get hotel recommendations"
    SemanticProperties.Hint="Gets hotel recommendations when you click"
    Clicked="OnHotelClicked"
    HorizontalOptions="Center" />

<Button
    x:Name="AttractionBtn"
    Text="Get attraction recommendations"
    SemanticProperties.Hint="Gets attraction recommendations when you click"
    Clicked="OnAttractionClicked"
    HorizontalOptions="Center" />

<Label x:Name="SmallLabel"
    Text="Click a button for recommendations!"
    SemanticProperties.HeadingLevel="Level2"
    FontSize="18"
    HorizontalOptions="Center" />

In order to build the project, you will need to add Clicked event handlers for each of the buttons. Add the following code to the MainPage.xaml.cs file and remove the existing event handler:

private async void OnRestaurantClicked(object sender, EventArgs e)
{
}

private async void OnHotelClicked(object sender, EventArgs e)
{
}

private async void OnAttractionClicked(object sender, EventArgs e)
{
}

The event handlers are all marked as async because we will be making asynchronous calls to the Azure.AI.OpenAI library. Now when you run the app, you should see the following UI:

Windows .NET MAUI app with the UI for providing recommendations.

Users can enter their location in the Entry control and click one of the buttons to get recommendations for restaurants, hotels, or attractions. The Label control at the bottom of the UI will display the results.

Next, let's add the Azure.AI.OpenAI library to the project and get it ready to make some API calls.

Reference and initialize the Azure.AI.OpenAI library

To call OpenAI's ChatGPT APIs, we're going to use an pre-release NuGet package from the Microsoft Azure team called Azure.AI.OpenAI. This library provides .NET APIs for making calls to both the OpenAI and Azure OpenAI APIs. We'll add the library to our project and bootstrap it with our API key.

Open the Package Manager Console from the Tools menu in Visual Studio.
Install the Azure.AI.OpenAI library by running the following command. The IncludePrerelease flag is required because the library is still in preview:
```
Install-Package Azure.AI.OpenAI -IncludePrerelease
```
Open MainPage.xaml.cs, the code-behind file for the MainPage. Add the following code to the top of the file to reference the Azure.AI.OpenAI library and create a variable to hold your OpenAI API key. Update the constructor to call the MainPage_Loaded method when the page is loaded, and add the MainPage_Loaded method to initialize the Azure OpenAI library:
```
private OpenAIClient _chatGptClient;
private string openAIKey = "<your-api-key-here>";
private string openAIEndpoint = null;

public MainPage()
{
    InitializeComponent();
    this.Loaded += MainPage_Loaded;
}

private void MainPage_Loaded(object sender, EventArgs e)
{
    _chatGptClient = !string.IsNullOrWhiteSpace(openAIEndpoint)
        ? new OpenAIClient(
            new Uri(openAIEndpoint),
            new AzureKeyCredential(openAIKey))
        : new OpenAIClient(openAIKey);
}
```
This will bootstrap the Azure OpenAI library with your API key and set some default options. You can create your API key on the OpenAI API settings page. The code is written so you could also set an endpoint if you are using the Azure OpenAI API. If you are using the OpenAI API, you will leave the openAIEndpoint variable set to null.
In order to compile the project, you will need to add the following using statements to the top of the MainPage.xaml.cs file:
```
using Azure.AI.OpenAI;
using Azure;
```

Now we're ready to put it all together. In the next section, we'll add some code to the three event handlers to make calls to the Azure.AI.OpenAI library and display the recommendation results.

Add ChatGPT API calls and test the app

It's time to add the code to our code-behind file that will use the Azure.AI.OpenAI library to make calls to the OpenAI ChatGPT API. We'll add the code to the three event handlers we created earlier. The code will get the user's location from the Entry control and pass it to the API to get recommendations. Then we'll display the results in the Label control at the bottom of the UI.

Create an async method named GetRecommendationAsync and call it from each of the event handlers:
```
private async void OnRestaurantClicked(object sender, EventArgs e)
{
    await GetRecommendationAsync("restaurant");
}

private async void OnHotelClicked(object sender, EventArgs e)
{
    await GetRecommendationAsync("hotel");
}

private async void OnAttractionClicked(object sender, EventArgs e)
{
    await GetRecommendationAsync("attraction");
}

private async Task GetRecommendationAsync(string recommendationType)
{
    if (string.IsNullOrWhiteSpace(LocationEntry.Text))
    {
        await DisplayAlert("Empty location", "Please enter a location (city or postal code)", "OK");
        return;
    }

    string prompt = $"What is a recommended {recommendationType} near {LocationEntry.Text}";

    // DeploymentName must match your custom deployment name (Azure OpenAI)
    // Or a default deployment name (such as OpenAI's GPT-3.5-turbo-0125) can be used
    ChatCompletionsOptions options = new()
    {
        DeploymentName = "gpt-3.5-turbo-0125",
        Messages =
        {
            new ChatRequestUserMessage(prompt)
        },
        ChoiceCount = 1,
        MaxTokens = 100,
    };

    var message = new ChatRequestUserMessage(prompt);
    options.Messages.Add(message);
    Response<ChatCompletions> response = await _chatGptClient.GetChatCompletionsAsync(options);
    SmallLabel.Text = response.Value.Choices[0].Message.Content;
}
```
This code first checks to make sure the user has entered a location in the Entry control. If not, it displays an alert and returns. If the user has entered a location, it calls the GetChatCompletionsAsync method on the Azure.AI.OpenAI library to make a call to OpenAI. The GetChatCompletionsAsync method takes a ChatCompletionOptions parameter. This parameter was created with the following options:
- DeploymentName - This is the name of the deployment (model) you want to use.
- Messages - This is a collection of ChatRequestUserMessage objects. We're only using one message in this example, but you can add more to the collection to provide more context to the API.
- ChoiceCount - This is the number of recommendations you want to get back from the API. We're only asking for one recommendation in this example.
- MaxTokens - This is the maximum number of tokens (words) you want the API to return in the response. We're asking for 100 tokens in this example.
The GetChatCompletionsAsync method returns a Response<ChatCompletions> object containing the response from the API. We'll get the Message.Content response text from the first Choice in the response.Value.Choices collection and display it in a control at the bottom of the UI.

Note

It's' a good idea to add error handling to the GetRecommendationAsync method to handle any exceptions that might be thrown by the API call. We've left that out of this example for simplicity.
The following using statements at the top of the MainPage.xaml.cs file are also needed for the code we added in this section:
```
using Azure.AI.OpenAI;
using Azure;
```
Run the app, enter a location, and test the recommendation buttons. You should see a response from the API in the Label control at the bottom of the UI:

That's it! You've successfully created a Windows .NET MAUI app that uses the OpenAI ChatGPT API to provide recommendations for restaurants, hotels, and attractions. Try changing the prompts to see if you can improve the results. You can also try changing the DeploymentName value in the ChatCompletionsOptions object in GetRecommendationAsync to see if you get better results from a different model.

Important

Remember to keep an eye on your API usage after your trial period has expired. You can also set monthly spending limits on your OpenAI account to avoid unexpected charges.

Next steps

Advance to the next article to learn how to...

Create a .NET MAUI app with C# Markup and the Community Toolkit