DishDive

Inspiration

The pace of modern urban life is accelerating rapidly, making people's pace of life becoming increasingly fast. Meanwhile, we've noticed that the food recommendations available in the market are becoming increasingly complex, adding to the time and decision-making costs of selecting a meal. We built our website DishDive to solve the issue we all have, choosing what to eat. We hope that DishDive can provide the most concise and appropriate choices, allowing users to quickly decide on their next meal. At the same time, instead of just eating out or choosing to eat some junk food, DishDive will help you pick healthier and more sustainable options for food.

What it does

DishDive asks you to fill out some data about your food preferences, what are some foods you do or don’t like. You can then add some filters about what you’re feeling today, maybe feeling like eating something sweet? or salty? How about some meat? Dishdive will then use gpt 3.5 to analyze your data and to suggest a unique and creative food option for you to try and enjoy. Don’t like it? No problem, you can just click to recommend a new dish within seconds until you find one that suits your cravings.

How we built it

We started with user habits, eliminating the need for a complicated registration process. Upon the first login, we use MongoDB to store user id, with the respective filters and food preferences as keys. And we continuously record the user's eating habits thereafter to update the data base for the sake of more precise prediction. We then utilize the GPT 3.5 AI model to provide concise and quick suggestions on the backend, and offer like/dislike suggestion feedback to enhance the prompt to give better suggestions.

Challenges we ran into

This was our first attempt at using MongoDB to store user data and obtaining results with the OpenAI API. The most challenging part was establishing a connection between the frontend and the backend. We spent a considerable amount of time ensuring the backend could work successfully on the frontend. Additionally, merging everything together was also a significant challenge.

Accomplishments that we're proud of

We take pride in creating an engaging food recommendation website, allowing AI to assist in every aspect of life and also help in reducing food waste. Our website offers a variety of personalized options and can tailor meals based on user habits, while single meal suggestions and flexible adjustments can minimize decision-making costs. Such a website is unique in the market and can be a valuable choice for food enthusiasts and those who don't want to be bothered with dietary decisions. Along the way, we've overcome numerous technical challenges, enjoyed smooth team communication, and collaborated happily to complete this fascinating project.

What we learned

We learned how to use MongoDB for data storage and how to call the OpenAI API and prompts to obtain the desired results. Additionally, a significant takeaway was understanding and applying the proper linkage between the frontend and backend.

What's next for DishDive

Moving forward, we will be adding more food preferences to provide a wider range of personalized options. We will also introduce recipe suggestions and refine the record of users' past dietary preferences, enabling the AI to offer more appropriate and tailored food recommendations that align with users' expectations. Moreover, we'll introduce brand new meal options for users, making our food suggestions even more delightful and surprising!