Inspiration

Duke University, known for its diverse community, welcomes students from all corners of the globe. Yet, amidst the plethora of options, one common challenge unites us all: the quest for satisfying dining options. For vegetarians like us, navigating these choices amidst busy schedules and limited options is a daily struggle. This struggle isn't unique; it resonates deeply with many students who share diverse dietary preferences. Inspired by our peers' experiences, we set out to revolutionize Duke's dining scene with a smart, all-in-one solution that solves all your dining problems, be it finding the perfect meal, tracking your spending, or managing your calorie intake.

What it does

Introducing Blue Dine, your personal dining companion at Duke. Using AI, Blue Dine tailors dining options at Duke's Brodhead Center to your dietary preferences, budget, and nutritional goals that align with your needs and desires. But that's not all. Blue Dine isn't just about finding food; it's about empowering you to make informed choices. Our AI-powered chatbot engages with you, answering questions, offering recommendations, and guiding you through the culinary landscape of Duke. And because we know that budgeting is just as crucial as nutrition, Blue Dine also helps you track your spending wisely, ensuring that your wallet stays as nourished as your body. Whether you're a freshman exploring campus for the first time, a seasoned student seeking healthier meal options, a gym enthusiast looking for a calorie-conscious meal, or a dedicated vegan in need of culinary inspiration, Blue Dine has something for everyone.

How we built it

Our app Blue Dine is built on the following key features:

Flutter Framework: Our app is based out of Flutter. We chose Flutter for its versatility and ability to create natively compiled applications for mobile, web, and desktop from a single codebase. This choice allowed us to ensure a consistent user experience across all platforms while speeding up the development process.

Benefit: Fast development and beautiful UI with a single codebase for iOS and Android.

Pinecone for AI-Powered Search: Utilized Pinecone's vector database to build an RAG for our AI-powered search engine. This enables the app to understand natural language queries and return the most relevant dining options based on various criteria such as dietary preferences, meal type, and price range.

Benefit: Delivers highly accurate search results tailored to user preferences, improving the overall app usability.

PocketBase and FastAPI for Backend services : Integrated PocketBase and FastAPI to manage the backend efficiently with the restaurants, items and user data . This combination provided a robust and scalable solution for handling user data, preferences, and interactions within the app.

Benefit: Offers a lightweight yet powerful backend solution, ensuring quick data retrieval and secure storage.

Duke's Public API for Restaurant Info: Accessed Duke University's public API to fetch detailed information about campus dining locations, including operating hours, menu options, and location details.

Benefit: Ensures users have access to up-to-date and accurate information directly from Duke's dining services.

Gemini-Powered Chatbot for Interactive User Queries: Integrated a Gemini-powered chatbot to assist users in navigating the app, answering queries, and providing personalized recommendations based on AI understanding of user preferences and dietary restrictions. This is build on our RAG specially curated with out dataset in Pinecone.

Benefit: Enhances user engagement and provides immediate assistance, making the dining discovery process more interactive.

Personalized Finance Tracking with Hackathon-Provided Dataset: Leveraged a dataset provided by the hackathon organizers to develop a finance tracking feature. This allows users to monitor their dining expenses, view monthly spend trends, and set budget goals directly within the app.

Benefit: Empowers users with financial insights and budgeting tools tailored to their dining habits, promoting better spending decisions.

Challenges we ran into

Navigating the integration of diverse APIs (Pinecone, Duke's public API, and Gemini) posed a significant challenge, as did ensuring accurate and relevant search results. Developing a user-friendly finance feature from limited hackathon data and creating a comprehensive dataset for Duke's dining options by scraping the Duke Dining website also tested our problem-solving skills.

Accomplishments that we're proud of

We're proud of creating a unified platform that addresses a broad spectrum of dining concerns for Duke students. Our AI search functionality, AI chatbot, nutritional tracking, and intuitive finance analysis stand out as key features that enhance the campus dining experience. The seamless integration of diverse technologies into a user-friendly app is our biggest achievement.

We're also proud of the dataset we generated by scarping the duke dining website consisting of 300+ items just from restaurants in Brodhead Center of Duke with details including description, calories, ingredients, price, restaurant name and more.

What we learned

This project deepened our understanding of AI and vector databases, improved our skills in Flutter and API integration especially with Pinecone and highlighted the importance of user-centered design in developing practical solutions to real-world problems.

What's next for Blue Dine

Looking ahead, we plan to expand Blue Dine's dataset to include more dining locations in Duke, refine our AI search algorithm for even more personalised results, and enhance our finance feature to support comprehensive budget management. Additionally, we aim to incorporate user feedback to continuously improve functionality and user experience.

Built With

Share this project:

Updates