## Inspiration
Millions of people across Africa self-medicate without access to qualified pharmacists or trustworthy health advice. Clinics are far, pharmacies are overwhelmed, and even when care is available, it's rarely offered in local languages. Many rely on hearsay or unsafe combinations of herbal and modern medicine.
We were inspired to reimagine the role of a community pharmacist—by bringing their knowledge directly into people’s hands through WhatsApp or mobile, in languages they understand.
## What we built
AIpharmacist is a conversational AI tool that helps users explain their symptoms and instantly receive pharmacist-reviewed advice, dosage instructions, safe herbal alternatives, and interaction warnings—all translated to their preferred local dialect (Twi, Ewe, Hausa, etc).
It’s culturally adapted, simple to use, and works anonymously.
## How we built it
- Built on bolt.new, using a one-shot prompt to generate an intelligent assistant persona
- Integrated symptom parsing logic to simulate pharmacist triage
- Used multilingual NLP to translate medical responses into Ghanaian dialects
- Pulled from a verified Q&A base of pharmacist-reviewed guidance (structured knowledgebase)
- Layered logic for drug interaction warnings and herbal alternative mapping
- Focused on mobile-first experience and potential future WhatsApp API delivery
## What we learned
Building for people in local contexts is not just about simplifying—it’s about respecting culture, languages, and health beliefs. We learned how difficult it is to bridge the gap between herbal knowledge and science, and how important tone and translation are when communicating health information.
## Challenges we faced
- Translating complex medical information into simple, non-jargon responses
- Building a symptom-to-response flow that felt conversational and not robotic
- Ensuring advice did not cross into regulated diagnosis or treatment boundaries
- Mapping safe herbal remedies based on regionally accepted practices
- Balancing privacy with potential future user history tracking for continuity
🔧 Built with
- Bolt.new
- Langchain / Prompt engineering
- OpenAI GPT-4 (underlying language engine)
- Verified pharmacist content (structured in JSON)
- Multilingual translation layer (custom dictionaries + rule-based)
- Optional WhatsApp API delivery (future integration)
Built With
- bolt
- supabase
Log in or sign up for Devpost to join the conversation.