CYPH3

Inspiration

Rwanda’s rapid digital transformation has increased the use of mobile money, online government services, and social platforms—but it has also created new opportunities for phishing, fraud, and social engineering attacks. Many people receive scam messages daily via SMS, WhatsApp, email, or fake websites, yet lack the tools or technical knowledge to identify them. Sentinel Rwanda AI was inspired by the need for a locally aware, intelligent cybersecurity assistant that understands Rwandan digital ecosystems and protects users in real time.

What it does

Sentinel Rwanda AI is a web-based phishing detection platform that analyzes text messages, emails, web links, and screenshots to determine whether they are safe or dangerous. It provides:

AI-powered phishing and scam detection

Risk scoring and clear verdicts

Context-aware explanations tailored to Rwandan scams (MoMo, banks, Irembo, job scams, etc........)

Multilingual support for English, Kinyarwanda, French, and other languages

Simple prevention advice users can immediately act on

How we built it

We built Sentinel Rwanda AI as a Streamlit web application with a modern cybersecurity-inspired interface. The core analysis engine uses the Google Gemini API to perform advanced language understanding and threat analysis. Key technologies include:

Python & Streamlit for the web interface

Google Gemini API for AI-powered scam detection

Custom prompt engineering for phishing and social engineering analysis

Dynamic UI animations and visual indicators to enhance user experience

Modular design for future API and mobile integration

Challenges we ran into

Configuring and integrating the Gemini API securely and reliably

Designing an interface that remains visually rich without impacting performance

Handling different types of inputs (text vs images) consistently

Making the explanations simple and understandable for non-technical users

Ensuring the system remains culturally and locally relevant to Rwanda

Accomplishments that we're proud of

Successfully integrating Gemini AI for real-time cybersecurity analysis

Building a professional, animated, and user-friendly interface

Supporting multiple languages to increase accessibility

Creating a system that explains why something is a scam—not just labeling it

Delivering a functional cybersecurity tool suitable for real-world use

What we learned

How to apply large language models to real cybersecurity problems

Effective prompt design for fraud and social engineering detection

Secure API integration and error handling

UI/UX design for security-focused applications

The importance of local context in cybersecurity solutions

What's next for Sentinel Rwanda AI

Mobile app integration (Android / USSD support)

Browser extension for real-time website scanning

SMS and WhatsApp bot integration

Image OCR enhancement for better screenshot analysis

Advanced threat intelligence dashboards

Collaboration with local institutions for wider deployment

Built With

• apis
• css
• fetch
• gemini-api
• google-ai-studio
• html5
• javascript
• next.js
• node.js
• react
• tailwind
• typescript