MedBud AI

Inspiration

Indian homes often keep old medicine strips, half-used bottles, and duplicate medicines. People self-medicate from memory or Google, leading to expired medicine use, unsafe self-medication, and delayed doctor consultation. We wanted to turn every Indian home medicine box into a safety-aware AI assistant.

What it does

MedBud AI scans medicine labels using Gemini 2.5 Flash vision AI, builds a home inventory, tracks expiry and low stock, and runs a conservative symptom-safety workflow. It detects red flags (chest pain, breathing issues, pregnancy, severe allergy), checks for duplicate ingredients, and recommends when to consult a doctor. Supports Hinglish for natural conversations.

How we built it

Next.js 14 App Router with TypeScript for the frontend Gemini 2.5 Flash REST API for vision-based medicine extraction and safety consultation Local browser storage for persistent inventory Precision-minimal UI inspired by Vercel/Linear aesthetics Mobile-first responsive design with premium black/white/gray styling

Challenges we ran into

Handling image upload and base64 conversion for Gemini API Designing a safety-first system that doesn't cross into medical advice Creating a dense, professional UI that works for Indian users Balancing AI confidence scores with manual fallback entry Netlify build ESLint configuration for production deployment Accomplishments that we're proud of Working end-to-end medicine scanning with real confidence scores Comprehensive safety reasoning that shows exactly what was checked Clear demo flow with sample data for hackathon judges Structured triage system (self-care/doctor/urgent) with red flag detection Premium, polished UI that looks professional and trustworthy Complete report generation for doctor consultation

What we learned

Vision AI can reliably extract medicine data from imperfect images Safety guardrails are crucial for health-related AI applications Dense, information-rich UIs require careful typography and spacing Indian users need Hinglish support and culturally relevant context Sample data dramatically improves demo experience for judges

What's next for MedBud AI

WhatsApp reminders for expiry and restock Pharmacy refill links integration Family profiles for multiple household members Doctor-shareable PDF reports Drug interaction database Voice input for elderly users Real dosage calculator (with proper medical oversight)