CLARKUBOT

Inspiration

ClarkuBot was born from the common frustration of finding specific information across Clark University's numerous resources. We aimed to create a centralized, intelligent tool to simplify access to knowledge for students, faculty, and staff.

What it does

ClarkuBot is an intelligent virtual assistant that answers your questions about Clark University by learning from uploaded documents(at the moment). Users can ask questions in natural language and receive concise, context-aware responses, eliminating the need to sift through multiple sources and erase through multiple pages

How we built it

We utilized Python as our core language, Streamlit for a user-friendly web interface, and the Google Gemini API to provide natural language understanding and generation. Our system includes document processing to make uploaded content accessible to the AI and session management for conversational context. we have also managed to keep it scalable by centralizing the data processing

Challenges we ran into

Key challenges included finding and scrapping consistent data for the AI to effectively learn and integrate clarks information and work with the Gemini API, it was also a good challenge to figure out the part of maintaining conversational context and handling ambiguous user queries, it was also a tough task to optimizing for both response speed and accuracy of the responses.

Accomplishments that we're proud of

We successfully developed a functional prototype featuring document uploading and seamless integration with the Gemini API, along with the ability to conduct context-aware conversations. We believe ClarkuBot demonstrates a tangible solution to navigating complex information challenges at Clark University.

What we learned

Through this project, we gained a ton of valuable experience about the powers, usage and limitations of large language models, along with that we also learned the importance of effective prompt engineering, and also the rapid development capabilities of Streamlit, the nuances of data preparation for AI, and the fast-paced nature of hackathon development.

What's next for CLARKUBOT

Our future plans involves expanding support for more file types like PDFs etc, and also exploring integration with official Clark University data sources (with proper authorization), implementing advanced search functionalities, considering user authentication for personalization, and ultimately deploying ClarkuBot for broader use within the university community and also for the prospective Clarkies.