Inspiration

Most drivers struggle to understand what's wrong with their vehicles β€” especially when vague symptoms arise. Workshop visits can be expensive and time-consuming, and Googling symptoms often leads to confusing results.

ROSAK.MY was built to make car diagnostics as simple as chatting with a friend. No jargon. No complicated tools. Just type what you’re experiencing, and get fast, clear suggestions powered by AI.

What it does

🧾 AI-Powered Symptom Analysis Users type in a car issue (e.g., β€œmy car makes a clicking noise when I turn”), and ROSAK.MY suggests the most likely problems.

πŸ› οΈ Troubleshooting Guidance The platform offers a list of possible causes and next steps β€” whether it's safe to drive, whether to seek a mechanic, or if it's something you can fix yourself.

πŸ“± Minimalist, Mobile-First UI Built on the go β€” quick to load, easy to use.

🧠 Built-in Prompt Engineering Custom-designed AI prompts provide relevant and structured outputs for vehicle issues.

How we built it

Tool/Stack Role -Lovable.dev / Bolt.new -To brainstorm, wireframe, and scaffold the UI. -Cursor / Augment -For hands-on code editing, API integration, and logic refinement. -Perplexity AI (Sonar) -Natural language understanding and diagnostics generation. -Tailwind CSS -Minimalist, mobile-first styling. -Vercel -Seamless deployment and hosting.

We used AI tools during development to ideate and speed up implementation β€” but core logic, diagnostic design, and UX were custom-built.

Challenges we ran into

🧩 Translating vague human symptoms into meaningful diagnostics. ⚠️ Balancing AI creativity with safety and accuracy. πŸ“± Ensuring the experience feels fast and useful even on low-end mobile devices. πŸ‡²πŸ‡Ύ Making the platform culturally and linguistically accessible to Malaysians.

What's next for RosakMY - AI-Powered Vehicle Diagnostics

πŸ“Έ Upload photo/video support for richer diagnostics. πŸ“ Location-based mechanic recommendations. πŸ” Save/share diagnostic reports. πŸ‡²πŸ‡Ύ Full Bahasa Melayu experience. 🧠 AI model fine-tuning with actual user input.

Built With

Share this project:

Updates