Booklingo

Inspiration

Learning a new language through textbooks or flashcard apps often feels robotic and detached from real-world usage. On the flip side, jumping straight into authentic foreign literature is intimidating—you spend more time looking up words in a dictionary than enjoying the story, quickly leading to cognitive overload and burnout.

We created Booklingo to bridge this gap. We wanted to give language learners an immersive, joyful way to acquire new vocabulary and grammar naturally through books, adapting stories in real time to match the reader's exact proficiency level.

What it does

Booklingo is an interactive, AI-driven reading assistant and language-learning companion that turns any book or story into a personalized language lesson.

Adaptive Reading: Readers can upload an ebook or select from a curated library. Booklingo automatically adjusts text difficulty, replacing or explaining complex idioms and vocabulary based on the user's proficiency (A1 to C2).

Instant Smart Glossary & Contextual Definitions: Hovering over or clicking any word gives context-aware translations, grammatical breakdowns, and audio pronunciations without breaking the reading flow.

Interactive Storytelling & Quizzes: At the end of each chapter, Booklingo generates personalized comprehension quizzes, dynamic vocabulary flashcards (using spaced repetition), and interactive discussion prompts driven by AI characters from the story.

AI Reading Companion: Readers can ask an embedded AI tutor questions mid-chapter—like "Why is this verb conjugated this way?" or "What is the cultural significance of this scene?"

How we built it

Frontend: Built with React / Next.js and Tailwind CSS to create a clean, clutter-free reading environment, with TypeScript ensuring steady state management across interactive text components.

Backend: Powered by Node.js / Express (or Python FastAPI) to handle document processing, text tokenization, and API orchestration.

AI & Language Processing:

Integrated Gemini APIs for real-time text simplification, context-aware translation, character roleplay, and grammar explanations.

Used LangChain to structure prompts for chapter summarization and quiz generation.

Database & Storage: Supabase (PostgreSQL) for managing user profiles, vocabulary banks, and reading progress tracking.

Challenges we ran into

Maintaining Literary Quality: Standard translation or simplification models often strip out author tone or produce awkward phrasing. We had to carefully engineer prompts and parameters so the simplified text retained its narrative charm and flow.

Real-Time Latency: Parsing large chapters, identifying target vocabulary, and rendering interactive tooltips on the fly caused initial performance bottlenecks. We solved this by implementing smart caching and streaming AI responses for real-time interactivity.

Contextual Accuracy: Words often have multiple meanings depending on literary context. Ensuring the popup definitions reflected the exact contextual meaning in the story—rather than just the primary dictionary definition—required multi-stage prompt validation.

Accomplishments that we're proud of

Maintaining Literary Quality: Standard translation or simplification models often strip out author tone or produce awkward phrasing. We had to carefully engineer prompts and parameters so the simplified text retained its narrative charm and flow.

Real-Time Latency: Parsing large chapters, identifying target vocabulary, and rendering interactive tooltips on the fly caused initial performance bottlenecks. We solved this by implementing smart caching and streaming AI responses for real-time interactivity.

Contextual Accuracy: Words often have multiple meanings depending on literary context. Ensuring the popup definitions reflected the exact contextual meaning in the story—rather than just the primary dictionary definition—required multi-stage prompt validation.

What we learned

Prompt Fine-Tuning for Pedagogy: We learned how to write prompts that act like effective language teachers—explaining concepts simply without being patronizing or overly academic.

UX for AI Integration: AI features are only useful if they fit naturally into user workflows. Minimizing friction between reading and getting assistance was crucial to keeping the experience engaging.