NetSage

Sponsor/Special Prizes

• Frontier Challenge
• Samba Nova Challenge

Inspiration

Inspired by the decrease in productivity when having an inadequate internet connection, by the difficulty of finding the perfect plan for you, and the difficulties faced trying to understand technical specifications for the average person, we built NetSage.

What it does

NetSage utilizes a RAG LLM to smartly generate product recommendations from Frontier's product catalog depending on the customers exact requirements. The LLM bases its decisions off Frontier's dataset of 800,000 customers, ensuring accurate recommendations for customers.

How we built it

We utilized a Retrieval-augmented generation (RAG) natural language processing system with a large language model to create the artificial intelligence behind NetSage. Our dataset included the large 800k customer database provided by Frontier for this challenge, as well as the other datasets of products and more. The backend was made in Python, utilizing Jupyter Notebook to learn and implement our Artificial Intelligence model. We then created a web app interface using Tailwind, React, Typescript, Streamlit, and an authentication system using Firebase.

Challenges we ran into

By far the hardest challenge we overcame was creating the RAG artificial intelligence model. Thanks for the persistence of our team members, we were able to learn about RAG models and build one from the ground up within 24 hours, an accomplishment we strived hard to achieve. Apart from that, we faced challenges in implementing our modal into the front end interface for a seamless user interface, but that too, we were able to overcome, thanks to our team.

Accomplishments that we're proud of

Our persistence, resilience, and in the end the final product are all accomplishments we are very proud of. Being able to explore, learn, and create a complete application, utilizing a technology we had to learn from scratch, within 24 hours is an enormous feat. We accomplished more than what we had set out to do, and that is an accomplishment we'll value forever.

What we learned

Throughout this project, we gained valuable insights into artificial intelligence, focusing on how Large Language Models (LLMs) operate with vast datasets. We explored the concept of Retrieval-Augmented Generation (RAG) models, which enhance LLMs by incorporating a retrieval mechanism. This allows the AI to supplement its responses with current and specific information, leading to more accurate, data-backed answers. By combining the generative capabilities of LLMs with the precision of information retrieval, we can create AI systems that are not only proficient in generating text but also equipped to provide up-to-date and contextually relevant responses.

What's next for NetSage

There are many future expansions for NetSage, after getting some rest we all need, we plan on adding more features such as automatic data monitoring from the customers router, augmented reality to help customers position products such as routers, and more to aid the customer is picking the perfect plan for them!

Built With

• cloudflare
• firebase
• jupyter
• next.js
• python
• sambanova
• tailwindcss