Our idea for this competition was to provide an interactive chat-interface to the articles on mslearn. The mslearn database is a vast resource with information on many different Microsoft products and tools. It is navigable through tags and collections like modules and learning paths, but it can still be tricky to find just the article you need in a particular situation since they overlap on certain topics. We wanted to make it easier to filter through the modules in the mslearn library, and we wanted to extend the functionality from just finding and reading up on certain topics, to being able to automatically find all modules related to some topic, so that it's easier to compare the different tools, products and methods. We do this by creating an interactive chat-like experience where you explain your problem/situation to an AI and get recommendations based on the knowledge in mslearn.
For example you could:
Post your resume to get suggestions of tools and products that you might be interested in
Describe a workflow and get suggestions for tools and products that might help you get the work done
The solution is built in three steps:
An ETL-pipeline built in a Fabric notebook that produces a knowledge database in a lakehouse on Onelake.
An Azure AI Search index that is built and populated via a Fabric notebook.
A simple web app GUI where the user can ask questions to a LLM on Azure OpenAI. It has vector search capabilities and can provide direct links to cited articles on mslearn.
Project name
Ask-MSlearn
Description
Our idea for this competition was to provide an interactive chat-interface to the articles on mslearn. The mslearn database is a vast resource with information on many different Microsoft products and tools. It is navigable through tags and collections like modules and learning paths, but it can still be tricky to find just the article you need in a particular situation since they overlap on certain topics. We wanted to make it easier to filter through the modules in the mslearn library, and we wanted to extend the functionality from just finding and reading up on certain topics, to being able to automatically find all modules related to some topic, so that it's easier to compare the different tools, products and methods. We do this by creating an interactive chat-like experience where you explain your problem/situation to an AI and get recommendations based on the knowledge in mslearn. For example you could:
The solution is built in three steps:
Project Repository URL
https://github.com/RikardMartin/mslearn-smart-chat
Project video
https://github.com/RikardMartin/mslearn-smart-chat/blob/main/docs/Ask-MSlearn-video.mp4
Team members
RikardMartin, mrmarten