AI-built Japanese Flashcard WebApp

Timeline: Aug 4th, 2025 (1 day)
Role: Developer, UX Designer
Team: Solo

How to learn Japanese Vocabularies comprehensively?

Skills
Vibe-coding
UX Thinking
AI-assisted Prototyping
Tools
Claude Code
Cursor
Browser DevTools
GitHub
Team
Solo (I treated it as a hackathon project)
My Role
Developer
UX Designer

Final Product

Link to WebApp: https://reenazys.github.io/JP-vocab-helper/

Main Features Demo

I. Three study modes:
Japanese → English translation
Kanji → Hiragana conversion
Reverse (English → Japanese)


II. Flagging system to mark and review difficult words


III. Custom vocabulary management: add, edit, delete words


IV. Bulk vocabulary import for faster setup

Background

Problem context: Duolingo Experience

While studying Japanese on Duolingo, I found that the platform limited my ability to review vocabulary flexibly.

Beyond personal learning, my secondary goal was to experiment with AI-assisted development as a way to prototype faster, while still owning the product vision and UX direction. I chose Claude Code and Cursor for my vibe-coding tool.

Development & Adjustments

Accessibility

After implementing most of the core features, I turned my attention to accessibility.

I realized that moving between input fields and advancing to the next flashcard would feel much smoother if users didn’t need to rely on a mouse click every time.

Keyboard navigation becomes the next priority.

Technical challenge

Japanese IME input (used to type hiragana, katakana, and kanji) created unique challenges, often interfering with standard keyboard shortcuts.

Reflection

This was my first experiment with “vibe coding” using Claude AI, and it taught me a lot about pairing AI-assisted development with my own UX direction.

Bottom Line: Frequent testing, clear feedback, and rounds of iterations were essential to achieve a polished user experience.

Other takeaways

Even with AI assistance, product vision and UX decisions remain human-driven. AI accelerated the build, but thoughtful design choices—like accessibility, navigation flow, and feedback mechanisms—still required deliberate iteration. This project reinforced my belief that AI can speed up development, but design ownership is what makes a product usable and meaningful.