Best use of AI
The main goal of our submission is to have a Gen AI based chat assistant aka Copilot. For this purpose, we developed a Copilot for PLAY! Summit website, based upon Azure OpenAI Studio and Azure Cognitive Search, and integrated it back to the PLAY! Summit repository. The Copilot can help improve the customer experience for finding useful information about the sessions and they can ask questions like:
- When is the opening note session and what is it about?
- Recommend me some sessions about Outdoor Adventure
- I cannot decide between a cycling and outdoor survival session, help me choose!
1- Setup Azure Cognitive Search and Azure OpenAI with your own data.
2- User sends an input via chat interface like “When is the opening key note session”
3- This information is passed to Azure Cognitive Search
4- The Azure Cognitive Search returns the response from the index.
5- The response plus the chat input (prompt) is then passed to Azure OpenAI model
6- The Azure OpenAI model returns the AI generated response.
7- This response is then sent back to the user.
This the overall architecture diagram of the Copilot
Video link => docs/ PlaySummit-Video.mp4
1- Requires Azure Subscription for Azure OpenAI
2- Requires Azure Subscription for Azure Cognitive Search
3- Docker for Windows with Windows containers enabled
4- Windows PowerShell 5.1
5- LTS version of Node.js
6- .NET Core 3.1 SDK or higher
7- .NET Framework 4.8 SDK
8- Visual Studio 2022
Additionally, you will need to setup index schema for Azure Cognitive Search and upload data to the index using REST API commands as mentioned here.
Follow the document present at Sitecore Demo Edge
Need to change the Azur related configuration which mentioned in the Website\src\rendering\src\pages\copilot\index.tsx
1- Install the environment using provided instructions
2- Replace Azure Keys and Endpoints with your own (Keys are omitted from the repo and will only be provided to the judges for evalution purposes)
3- Chat and organize your sessions!
We are using free subscription so need to wait for 30 seconds for another query in the chat.