KnowYourMedicine

AI generated image or not check
Snap your medicine
Medicine details and translate
Yoga
Food plan

Inspiration

I watched my mother struggle to understand her prescribed medication. The label was unclear, information was in English (not her native language), and she had no idea what foods or exercises could help her recovery. 768 million adults worldwide cannot read their medicine labels. Medication errors cause over 100,000 deaths annually in the US alone. I built this for her. And for millions like her.

What it does

One photo. Complete health clarity.

📸 Upload medicine photo 🔍 AI verifies authenticity (detects counterfeit/AI-generated images) 💊 Extracts complete medicine details (name, dosage, side effects, interactions) 🧘 Provides personalized yoga recommendations 🥗 Suggests foods to eat/avoid with the medication 🌍 Translates everything into your preferred language

All in under 1min.

How we built it

The Frankenstein Architecture: I stitched together five incompatible technologies into one unified system. Tech Stack:

AI: Google Gemini 2.0 Flash Backend: 5 Supabase Edge Functions (microservices)

verify-image - OCR + authenticity detection analyze-medicine - Medical information extraction get-yoga-practices - Context-aware yoga recommendations get-food-suggestions - Medication-specific dietary guidance translate - Multi-language support

Frontend: Responsive web interface Deployment: Netlify

Why "Frankenstein"? I united technologies that rarely communicate:

Computer Vision ↔ Medical Databases Healthcare APIs ↔ Wellness Systems Clinical Information ↔ Holistic Practices Medical Data ↔ Language Translation

Each Edge Function is a specialized AI agent. Together, they work like a coordinated medical team.

Challenges we ran into

OCR Accuracy: Medicine labels have small fonts, curved surfaces, reflections, and poor lighting. Solution: Advanced prompt engineering with Gemini 2.0's vision capabilities. Medical Reliability: AI hallucination is dangerous in healthcare. Solution: Strict fact-checking prompts, confidence scores, and clear limitations disclosure. Safe Holistic Recommendations: Yoga/food suggestions must avoid contraindications. Solution: Medicine-aware prompts with safety caveats and "consult your doctor" disclaimers. Edge Function Orchestration: Coordinating 5 serverless functions with proper error handling. Solution: Clear data pipeline with structured JSON and fallback mechanisms.

Accomplishments that we're proud of

✅ Successfully orchestrated 5 AI agents into seamless UX ✅ Real-time processing in under 10 seconds ✅ Bridged modern medicine with holistic wellness ✅ Made healthcare accessible across 100+ languages ✅ Solved a real problem - my mom now confidently manages her medications ✅ Created "Holistic Health AI" - a new category merging clinical and wellness tech

What we learned

Technical: Gemini 2.0 Flash handles multi-modal tasks beautifully. Prompt engineering is as important as coding. Serverless edge functions scale naturally for AI workloads. Product: Users want simplicity - "take photo, get answer." Trust is everything in healthcare - transparency features are critical. Holistic health is underserved and deeply valued. Personal: Build for people you love. Healthcare information should be a right, not a privilege. AI should augment human care, not replace it.