Diagnosis method
Nearby Hospitals
Landing page
User Dashboard
Ebola example page

🏥 Sani AI — Smart Healthcare Gateway for East Africa

Empowering communities with instant, localized medical guidance while equipping health authorities with real-time epidemiological insights.

💡 Inspiration

In Africa, and particularly in Uganda, a critical healthcare challenge persists: most people only go to the hospital when they are severely ill. This delayed approach to seeking medical care is driven by:

Prohibitive medical and transit costs
Long travel distances to regional clinics
Exhaustive waiting times to receive basic test results

Compounding this issue is a severe shortage of accessible, reliable medical information. Citizens frequently struggle to find immediate guidance on managing chronic conditions like sickle cell disease and HIV, lack instant instructions during emergencies like snakebites, and face uncertainty during outbreaks like Ebola. Furthermore, finding a hospital with the right specialists in a specific region remains difficult.

The Data Gap for Health Authorities

Health authorities and planners lack real-time data to track spreading illnesses. Without immediate, localized insights, health systems cannot see where outbreaks are escalating, making proactive planning and resource allocation nearly impossible.

The Solution: Sani AI

Sani AI was created to solve these exact problems. By turning any smartphone into a personal healthcare gateway, the app allows users to input their symptoms and location to get an immediate, localized health assessment.

Crucially, Sani AI aggregates this regional symptom data in real time. By tracking user inputs across different areas, the platform detects early warning signals and tracks patterns when diseases like malaria or Ebola start breaking records. This bridges the information gap for patients while empowering health organizations with the data needed for early intervention and proper planning.

🚀 What It Does

SaniAI is an intelligent, emergency-first digital health assistant designed specifically for East African communities:

🚨 Instant Symptom Triage: Users describe how they feel and receive immediate, color-coded medical triage (Red, Yellow, Green) alongside step-by-step first-aid advice tailored to regional hazards.
🗣️ Speak Your Native Language: Users can speak naturally to log symptoms. Powered by native audio AI, it transcribes and intelligently refines voice inputs in English, Swahili, and Luganda, making healthcare accessible regardless of literacy levels.
📶 Full PWA Capabilities (Offline First): Now a fully-equipped Progressive Web App (PWA), SaniAI can be added directly to any phone's home screen. It caches essential clinical guidelines and first-aid resources, allowing it to boot instantly and provide offline access when internet connectivity drops.
📊 Dynamic Epidemiological Surveillance: Aggregates anonymous, regional symptom data in real-time, displaying localized health trends to flag disease spikes before they turn into major outbreaks.
🔐 Secure Health Diary: Users can optionally sign in securely using Google to log chronic illness patterns and maintain a medical history to share with doctors.

🛠️ How We Built It

The architecture of Sani AI balances cutting-edge AI capability with optimized performance for low-bandwidth environments:

Component	Technology Used	Implementation Details
The Smart Core	`@google/genai` (Gemini 1.5 Flash)	Drives localized diagnostic reasoning, enforces strict safety blocks on prescription drug output, and handles real-time translation pipelines.
Search Grounding	Google Search Grounding	Combined Gemini with live web-search to fetch current WHO and local ministry statistics on ongoing outbreaks (such as Ebola) for the interactive dashboard.
Audio Intelligence	Multimodal Audio Parsing	Implements direct server-side audio parsing, allowing users to upload voice recordings directly without relying on heavy external APIs.
Backend & Storage	Firebase Auth & Cloud Firestore	Secures historical user records, chronic disease logs, and symptom tracking with strict security rules.
Frontend & PWA	React (TypeScript), Vite, Tailwind CSS	Custom Service Worker built for complete home-screen installability, offline resiliency, and instant load speeds.

⚠️ Challenges We Ran Into

Language Nuances & Audio Quality: Transcribing and refining raw speech with regional accents in Swahili and Luganda required careful prompt engineering to prevent hallucinations while cleaning up colloquial phrases.
Connectivity Barriers: Building for regions with unstable 3G networks meant a heavy focus on performance. We solved this by creating a lightweight service worker to cache critical assets to render the app shell offline, keeping the core client payload incredibly small.
Balancing Safety and Value: Enforcing strict safety guardrails so the AI provides helpful triage support without ever making a formal medical prescription or replacing physical doctors.

🎉 Accomplishments That We're Proud Of

📱 PWA Installation: Turning SaniAI into an installable, native-feeling Progressive Web App that consumes virtually zero storage space and loads instantly.
⚡ Emergency-First Flow: A layout that requires zero login barriers to access life-saving triage or snakebite protocols when seconds matter most.
🌍 Localization: Successfully implementing multilingual voice support so users can seek healthcare guidance in their most comfortable, mother-tongue language.

🧠 What We Learned

Design is Accessibility: User interfaces in medical emergencies must be ultra-simple, legible, and visual. Eliminating the sign-up requirement was the single best decision we made for user experience.
Power of Local Context: AI models are incredibly capable, but they are infinitely more effective when grounded in local realities—knowing when to worry about malaria after the rainy season or recognizing native snake species (like the Puff Adder).

🔮 What's Next for SANI AI

💬 SMS Triage Integration: Launching an offline USSD/SMS fallback integration so users without smartphones or mobile data can text their symptoms to a local number and receive Gemini-driven SMS instructions.
🏥 Clinician Dashboard: Partnering with regional clinics in Uganda to provide them with the aggregated, anonymized epidemiological trends to distribute medical resources proactively.