eShopSupport

A sample .NET application showcasing common use cases and development practices for build AI solutions in .NET (Generative AI, specifically). This sample demonstrates a customer support application for an e-commerce website using a services-based architecture with .NET Aspire. It includes support for the following AI use cases:

• Text classification, applying labels based on content
• Sentiment analysis based on message content
• Summarization of large sets of text
• Synthetic data generation, creating test content for the sample
• Chat bot interactions with chat history and suggested responses

This sample also demonstrates the following development practices:

• Developing a solution locally, using small local models
• Evaluating the quality of AI responses using grounded Q&A data
• Leveraging Python projects as part of a .NET Aspire solution
• Deploying the application, including small local models, to the Cloud (coming soon)

Architecture

Getting Started

Prerequisites

• A device with an Nvidia GPU (see workaround for running on the CPU)
• Clone the eShopSupport repository: https://github.com/dotnet/eshopsupport
• Install & start Docker Desktop
• Install Python 3.12.5

Windows with Visual Studio

• Install Visual Studio 2022 version 17.10 or newer
  • Select the following workloads:
    • ASP.NET and web development workload.
    • Python Development workload.
    • .NET Aspire SDK component in Individual components.

Mac, Linux, & Windows without Visual Studio

• Install the latest .NET 8 SDK

• Install the .NET Aspire workload with the following commands:

  dotnet workload update
  dotnet workload install aspire
  dotnet restore eShopSupport.sln

• (Optionally) Install Visual Studio Code with the C# Dev Kit extension

Install Python requirements

From the Terminal, at the root of the cloned repo, run:

pip install -r src/PythonInference/requirements.txt

Note: If the above command doesn't work on Windows, use the following command:

py -m pip install -r src/PythonInference/requirements.txt

Running the solution

Warning

Remember to ensure that Docker is started.

• (Windows only) Run the application from Visual Studio:

  • Open the eShopSupport.sln file in Visual Studio
  • Ensure that AppHost is your startup project
  • Hit Ctrl-F5 to launch .NET Aspire

• Or run the application from your terminal:

  dotnet run --project src/AppHost

  then look for lines like this in the console output in order to find the URL to open the Aspire dashboard:

  Login to the dashboard at: http://localhost:17191/login?t=uniquelogincodeforyou

You may need to install ASP.NET Core HTTPS development certificates first, and then close all browser tabs. Learn more at https://aka.ms/aspnet/https-trust-dev-cert

Contributing

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact [email protected] with any additional questions or comments.

Sample data

The sample data is defined in seeddata. All products/descriptions/brands, manuals, customers, and support tickets names are fictional and were generated using GPT-35-Turbo using the included DataGenerator project.