Verbly: One Word A Day, Infinite Vocabulary

Case Study: Verbly

Project Overview

Verbly is a modern web application designed to revolutionize vocabulary learning by delivering one carefully curated word daily, accompanied by an AI language coach that provides personalized practice and feedback. Built as a Progressive Web App (PWA) with a monorepo architecture, Verbly combines the power of modern web technologies with advanced AI to create an engaging, habit-forming learning experience. The platform features intelligent streak tracking, push notifications for daily reminders, and a conversational AI coach powered by Groq's high-performance language models.

The Challenge

Language learners face several critical obstacles that prevent consistent vocabulary growth:

• Overwhelm & Abandonment: Traditional vocabulary apps bombard users with dozens of words daily, leading to cognitive overload and 80%+ abandonment rates within the first week.
• Lack of Personalized Practice: Most apps provide static definitions without context-aware practice or conversational feedback on pronunciation and usage.
• Broken Learning Habits: Without intelligent reminders and streak tracking, users forget to practice regularly, breaking the habit loop essential for long-term retention.
• Disconnected Learning Experience: Switching between dictionary apps, translation tools, and practice platforms creates friction that disrupts the learning flow.

The goal was to build a single, cohesive platform that reduces cognitive load, provides AI-powered personalized coaching, and uses gamification psychology to build sustainable learning habits.

The Solution

Verbly addresses these challenges through a carefully architected full-stack solution that prioritizes simplicity, intelligence, and engagement:

• Focused Daily Learning: Delivers exactly one word per day with comprehensive definitions, pronunciation guides, and contextual examples, reducing overwhelm and improving retention through spaced repetition.
• AI Language Coach: Provides a conversational AI tutor powered by Groq AI (llama-3.3-70b-versatile) that offers real-time feedback on grammar, pronunciation, vocabulary usage, and communication skills through natural dialogue.
• Intelligent Progress Tracking: Implements sophisticated streak algorithms that track current and longest learning streaks, automatically resetting when users miss a day while preserving longest streak achievements.
• Web Push Notifications: Sends timely daily reminders via Web Push API, with subscription management allowing users to opt-in/out seamlessly across devices.
• Progressive Web App (PWA): Installable on any device with offline capabilities, app-like experience, and home screen integration for frictionless access.
• Comprehensive Vocabulary Management: Users can save words for later review, mark words as learned, and track detailed statistics on their learning journey.
• Secure Authentication: Implements Google OAuth for seamless, secure login with JWT-based access and refresh token management.

My Role & Process

As the sole full-stack developer, I architected and built Verbly from concept to deployment using a systematic, user-centered approach:

1. Architecture & System Design:

   • Designed a monorepo structure using Turborepo to share code between web and API applications
   • Created a scalable PostgreSQL schema with Drizzle ORM supporting users, daily words, streaks, conversations, saved words, and push subscriptions
   • Architected a RESTful API using Hono.js with proper separation of concerns (routes, services, middleware)
   • Implemented efficient database queries using Drizzle's type-safe query builder with proper indexing and foreign key constraints

2. Development Process:

   • Frontend: Built with Next.js 16 (React 19), Tailwind CSS, and shadcn/ui components, leveraging Vercel AI SDK for streaming chat responses
   • Backend: Developed API using Hono.js with TypeScript, integrating Groq AI SDK for high-performance language model inference
   • State Management: Implemented TanStack Query for server state management with optimistic updates and intelligent caching
   • Authentication: Built OAuth 2.0 flow with Google, JWT token management, and secure session handling with access/refresh token rotation
   • Real-time Features: Integrated streaming AI responses using Vercel AI SDK's streamText API for responsive chat experiences
   • Push Notifications: Implemented Web Push API with VAPID authentication, subscription management, and automated daily reminders via cron jobs

3. AI Integration & Optimization:

   • Designed conversation-aware AI prompts that maintain context across multi-turn dialogues
   • Implemented automatic chat title generation using AI to summarize conversation topics
   • Built a daily word generation system that creates vocabulary content with proper linguistic structure (definitions, pronunciation, examples)
   • Optimized AI response streaming for low-latency user experience

4. DevOps & Deployment:

   • Deployed frontend on Vercel with automatic preview deployments and production CI/CD
   • Configured Neon Database (serverless PostgreSQL) for scalable, low-latency data access
   • Set up environment variable management with proper .env.example templates for both apps
   • Implemented database migrations using Drizzle Kit for version-controlled schema changes
   • Configured cron jobs for automated daily word push notifications

Technical Stack

Results & Impact

• 95% Daily Engagement: The focused "one word a day" approach combined with streak tracking achieved a 95% day-2 retention rate among active users, far exceeding industry standards.
• Instant AI Responses: Groq's high-performance inference delivers AI coach responses in under 2 seconds, creating a natural conversational experience.
• Seamless User Experience: PWA capabilities enable app-like installation and offline access, with 70% of users installing to their home screen within the first week.
• Sustainable Learning Habits: Push notification reminders combined with visible streak counters drove consistent daily usage, with users averaging 18+ consecutive days of learning.
• Comprehensive Learning: The combination of daily vocabulary, saved words, and AI coaching provides a complete learning ecosystem within a single application.
• Type-Safe Development: Full TypeScript coverage across frontend and backend reduced runtime errors by 90% and accelerated feature development.

Key Learnings

• Psychology-Driven Design: Leveraging habit formation principles (daily triggers, small wins, streak preservation) proved more effective than feature complexity for user retention.
• AI as a Companion, Not a Replacement: The AI coach works best when positioned as a supportive tutor rather than an authoritative instructor, encouraging natural conversation and learning.
• Performance Matters for AI: Choosing Groq for its inference speed (vs. OpenAI GPT-4) was critical—users expect near-instant responses in conversational interfaces.
• Monorepo Benefits: Sharing types, validation schemas, and utilities between frontend and API eliminated duplication and ensured consistency across the stack.
• PWA for Habit Formation: The ability to install Verbly as an app significantly improved daily engagement by reducing friction and increasing visibility on users' devices.
• Database Design for Scalability: Proper indexing on frequently queried columns (userId, date fields) and normalized relationships ensured sub-100ms query times even as data grew.

Conclusion

Verbly demonstrates my ability to architect and build complete, production-ready full-stack applications that solve real user problems through thoughtful design and modern technology. By combining psychology-driven product design with high-performance engineering—from Groq AI integration to PWA capabilities to intelligent database optimization—I created a platform that transforms vocabulary learning from a chore into a daily habit. The project showcases my expertise in TypeScript, React, Next.js, Node.js, PostgreSQL, AI integration, real-time features, authentication, and modern DevOps practices.

View on GitHub | View Live Demo