Scott and Mark learn AI

This is the code repo used in the Scott and Mark learn AI session at Microsoft Build 2024. Check the session recording here: https://www.youtube.com/watch?v=KKWPSkYN3vw

Config Azure OpenAI endpoints

The chatbot uses Azure OpenAI endpoints. To use the app, config your model, Azure OpenAI endpoints and keys in Resources\appsettings.json first.

Demo desktop setup

To use this, you need to set up your Desktop first.

1. Create a non OneDrive folder for this demo as root folder.
2. Create a Desktop folder under the root folder.
3. Run git init in the root folder so that we can restore the Desktop folder if things go wrong.
4. Right click Desktop in the left tree view from your Explorer window, go to Properties, Location tab, set the Desktop path to the desktop folder in the root folder.
5. Move the files and folders to the new Desktop folder.
6. Run the app, first click the Backup button to back up the state that you might want to restore later. You can also do this via git commands directly.

Run Aspire dashboard

1. Install Docker Desktop
2. Run this command

   docker run --rm -it -p 18888:18888 -p 4317:18889 -d --name aspire-dashboard mcr.microsoft.com/dotnet/aspire-dashboard:8.0.0
   

3. Open Aspire dashboard from http://localhost:18888/.
4. Login token can be found at the beginning of the container log like this Login to the dashboard at http://0.0.0.0:18888/login?t=04cc5ebb75c2b22e39173b3dab2a50af. Copy the token after login? to login.

Project structure

Main files to interact with:

• MauiProgram.cs - initialize services, setup kernel model entpoints, logging
• ChatManager.cs - manage metaprompt to GPT
• OrganizeDesktopPlugin.cs - plugin functions for GPT to call
• ChatHistoryViewModel.cs - view model that manages chat history and other stuff for UI

Local model server setup

Check news on Announcing the AI Toolkit for Visual Studio Code.

See news from Microsoft for the to-be-released local model server.