Mentorak

Inspiration

As a self-learner and developer, I often found that testing my understanding of a topic was harder than learning it. I wanted a fast and simple way to generate practice quizzes without digging through long PDFs or copying content manually. That’s where Mentorak started.

Inspired by ChatGPT’s conversational UI and powered by Kiro, I decided to build a topic-to-quiz generator that anyone could use. No uploads. No prep. Just type and learn.

What it does

Mentorak is an AI-powered quiz generator that instantly transforms any topic into a 10-question multiple-choice quiz.

-Just type a topic like "Photosynthesis", "Newton’s Laws", or "World War II" -Get 10 structured MCQs in seconds, each with 4 answer choices -At the end of the quiz, you’ll see your score, and Mentorak shows you the correct and incorrect answers.

How we built it

This project was built entirely with Kiro. I began by writing specs that described the entire flow topic input, quiz generation, and result scoring. From those specs, Kiro generated the initial layout, routes, and key logic for quiz sessions As the product evolved, I updated the specs and Kiro kept the codebase clean, type-safe, and organized.

Challenges we ran into

Consistent JSON parsing from the quiz generation logic was a challenge — sometimes the model returned malformed arrays, breaking the app flow. Creating a clean UI/UX for quiz feedback showing correct/wrong answers clearly while maintaining a smooth experience required multiple iterations.

Protecting routes and session state with Firebase Auth, while keeping the app responsive and fast, took some debugging work.

Accomplishments that we're proud of

🌟 "Learn Anything" with AI-Generated Quizzes, The core achievement of Mentorak is enabling users to type any topic and instantly get a personalized quiz. No uploads, no templates, just fast, focused learning.

Created a quiz results view with instant visual feedback on right and wrong answers, plus score tracking and history for review.

What we learned

Specs are powerful, once we got comfortable writing Kiro specs, our development speed increased dramatically. We no longer feared UI refactors.

Building with AI (OpenAI for quizzes, Kiro for development) requires clear structure and fallback logic, especially when parsing or rendering generated content.

UI/UX decisions matter even for developer-facing hackathon apps, i saw how even simple transitions or button placements improved the user journey.

What's next for Mentorak

🎙 Voice-powered quizzes — users will be able to speak answers instead of clicking 🤖 AI chat tutor — review wrong answers with a conversational assistant that explains the concepts 📄 Bring back PDF input — but only after ensuring 100% stable parsing and quiz mapping 🧪 Adaptive difficulty — Mentorak will adjust question difficulty based on your past performance 👥 Multi-user quiz battles — compete with friends or classmates in real-time

Built With

• kiro
• next.js
• openai
• tailwind
• typescript