Inspiration

I wanted to leverage the AI to create impactful applications that would be on the forefront of changing the world.

What it does

This application simply has an AI assistant that answers questions related to Sales, Products and Customer information of a bike shop named Cosmic Works. It makes getting information easier, intuitive and human like.

How we built it

I Used the RAG system pattern in utilising the ChatGPT along with vCore based CosmosDB for MongoDB resource created in Azure Cloud, which retrieves the information that is similar to the query and then leverages LLM (ChatGPT) to construct the conversational answer following the system prompt provided to guide the LLM in creating the answer.

Challenges we ran into

Every step of the way challenges were encountered, I had to not only learn new programming language like PowerShell but also had to learn the Cloud fundamentals of Azure Cloud and features it provides like vCore based CosmosDb for MongoDB and RegistryContainer to push up the docker images to understanding the Bicep automation script for deployment of resources on Azure.

Accomplishments that we're proud of

Successfully integrating the LLM into the application with CosmosDB for MongoDB while using the Vectorised data in building up the RAG pattern system to give information in conversational style

What we learned

I leaned how to leverage RAG pattern to implement LLM based application along with how I can utilise Azure Cloud and what challenges I would face and how to overcome those under a tight deadline.

What's next for Custom AI Copilot

I would want to build some more advanced Copilot applications where ChatGPT is able to take decision based on information provided and that decisions could help make lives of humans even more convenient.

Built With

Share this project:

Updates