-
The user can prompt our LLM recommend a book based on the user's prompt.
-
When asked for a book about horses, this book is returned and a reason for the suggestion is also given.
-
The books the user has in their 'Currently Reading' bookshelf.
-
The user has several bookshelves based on how they have interacted with the books.
-
The top search bar allows the user to search the database for books.
-
The home screen recommends the user popular books.
-
The user can click on a book to read the description, see the themes of the book, and add the book to a 'bookshelf'.
-
Our login page, which includes various log in methods.
-
The founders with the mascot!
There is a secret Easter Egg on our website. Try and find it!!
Inspiration
Slightly Better Reads is our pride and joy — a labour of love born from our shared passion for books. As devoted readers, we felt the existing book recommendation tools left much to be desired, so we took it upon ourselves to create something better. With the emergence of powerful LLMs, we decided to innovate a novel way to suggest new books to a reader by creating a tool that really understands the user.
How we built it
The tool takes in a prompt from the user and sends an HTTP request to the backend. Here, we use Chat-GPT 4o to recommend books to the user based on the prompt and the books they have read, are reading, dropped etc. All this information is hosted using Firestore. The LLM outputs a list of books that are sent to OpenLibrary's Works API to gather key information about the book in a way that is easy to display. This is finally returned to the front end, where it is displayed using React and Tailwind framework. Each book recommendation also comes with a short reason why the user in particular would enjoy the book generated by the LLM.
Challenges we ran into
A large issue we had to deal with was the inconsistent amount of information held about each book on the API that we used. In particular, the API sometimes held up to 100 genres for a single book, so we had to develop algorithms to filter through it and select the best ones. Additionally, our lack of experience with technologies like React, Firebase and APIs meant that we had to learn a lot within a short time frame.
Accomplishments that we're proud of
Despite all the challenges we ran into, we are super proud of the speed at which we were able to create the final product. We are also proud of the perseverence of our team members to push through the many, many errors we ran into and still produce a complete product.
What's next for Slightly Better Reads
In the future, we hope to add new features into the AI chat part of the application so that the LLM does not recommend books right away, but prompts the reader further to gain a better understanding of the user's needs.
Note for use: Please do not test the website too strenuously, as there is limited credit for the LLM we are using.
Our GitHub link: https://github.com/MrCaspion/Copy-of-book-repo
Log in or sign up for Devpost to join the conversation.