Bookly

• 

  Wizard of Oz with insightful AI summary

• 

  Favorites page

• 

  Daily book recommendation

Inspiration

We wanted to create a tool that helps people connect with books in a meaningful way—especially those who might feel overwhelmed by large catalogs or lack the time to read through long descriptions or reviews. The idea was to leverage AI to summarize content and use genre-based recommendations to help users discover new books they’re likely to enjoy, without needing to do deep manual research.

What it does

Bookly is a book discovery and recommendation platform that:

Displays a curated catalog of popular books

Uses AI to generate summaries based on book titles (and descriptions when available)

Allows users to filter books by genre and receive tailored recommendations

Lets users favorite books they’re interested in reading later

Our goal is to enhance how users interact with books by providing intelligent summaries and actionable insights from minimal input.

How we built it

Frontend: Built using Vue.js and Vite for a fast and responsive single-page app experience.

Backend: Node.js with Express.js to manage routes, API calls, and communication between the frontend and database.

Database: We used MongoDB Atlas, leveraging its robust full-text search and cutting-edge vector search capabilities to support intelligent querying and future recommendation scalability.

AI Integration: Summaries are generated through Hugging Face’s inference API. We tuned our prompts and model selection to generate summaries even when only the title was available.

DevOps: The project supports local environment setup with dynamic port detection and automated .env configuration for both server and client using a custom setup script.

Challenges we ran into

Many books lacked detailed metadata, especially descriptions, so generating useful summaries required careful model prompting and tuning. Also, some Hugging Face models had inconsistent performance or latency, so we had to adapt by retrying requests and caching intelligently. Additionally, MongoDB URI configuration issues caused unexpected connection errors, particularly with duplicated URI options like retryWrites. Lastly, making AI inference seamless while preserving frontend responsiveness was a challenge, especially balancing UI loading states with meaningful content.

Accomplishments that we're proud of

We built a clean, full-stack experience that integrates generative AI, real-time search, and modern UI design.

Managed to generate insightful AI summaries even from incomplete datasets.

Learned and successfully applied MongoDB’s powerful search and vector features to support future advanced querying and recommendation features.

Delivered an app that not only looks great but also enhances the way users discover and relate to book data.

What we learned

We learned how to use Hugging Face APIs to craft summaries from sparse input data

How to apply MongoDB Atlas features—especially text and vector search—to power smarter interactions with content

Improved our knowledge of secure environment configuration and full-stack project orchestration

Developed better error-handling strategies around real-world third-party services (e.g., AI inference APIs, database connections)

What's next for Bookly

Add book ratings, user reviews, and reading lists

Integrate external links for purchasing or accessing eBooks and audiobooks

Built With

• css
• express.js
• html5
• huggingface
• javascript
• mongodb
• mongoose
• node.js
• openlibrary
• render
• shell
• vercel
• vite
• vue.js