Better Reads | Devpost

Feed page
Timeline
Add book to library
Search book
AI summary
Edit book in library
Library page

Inspiration

Being avid readers, we like to track our reading habits and build lists of books we've read over time. Current solutions like Goodreads or other note-taking apps require quite a bit of data entry in order to maintain your reading list, and even more work to construct meaningful analysis from it. Moreover, some of the cover images on Goodreads are of really old editions which certainly don't look very appealing. This makes it tedious and visually outdated on other platforms - a problem we wanted to solve and a process we wanted to make more fun and interactive for fellow bookworms! Another solution we wanted to offer is digital annotations! Taking physical notes is not the most resilient method of storing memory in today's world.

What it does

BetterReads is a reading tracker. It's a modern reading habit tracker coupled with some social media features and of course, some agentic capabilties. You can maintain a list of all books that you read with minimal data entry and natural conversation. You can also view some analytics and trends in your reading, along with the ability to perform some customizations to the list views itself! Moreover, one feature that we wanted to support that other don't is annotations of pages! In order to do this, one uploads an image of the page (or a screenshot) and simple writes down the notes down. Using computer vision, we detect what words are written about and they're transcribed onto the platform.

How we built it

We decided to build a mono-repo webapp with next.js using typescript, react and trpc for end-to-end type safety. To prevent re-inventing the wheel, we used shadcn ui components and some integrations like openai and google-books. We also built a web crawler that finds top cover images for a book, and partially built a huggingface image inference bot that can pick the most modern and clear covers from the crawled results.