SafeCall | Devpost

🌟 Inspiration

There are moments when calling for help isn’t an option. Sometimes a wrong move could make things worse. Whether it’s feeling uneasy in a rideshare, being followed on the street, or navigating an abusive situation, victims often need a way out that doesn’t put them in more danger. Having quick and easy access to calling services can be a lifeline, offering both reassurance and immediate help when it’s needed most. We wanted to create a solution that feels natural, provides reassurance, and most importantly, helps people safely remove themselves from harm. SafeCall is designed to give users a believable excuse to leave, a discreet way to seek help, and a sense of control in situations where options feel limited.

🛠️ What it does

SafeCall is an AI-powered safety tool designed to assist users in high-risk situations. It connects them to a real-time voice AI that simulates a natural conversation. Unlike static prerecorded messages, SafeCall adapts dynamically, allowing for fluid, lifelike interactions that can continue as long as needed. In addition, SafeCall securely logs call details, including time, severity level, and last known location and puts it all onto a dashboard for emergency response analysis.

🧱 How we built it

The app is separated into two parts: the frontend and backend.

The frontend is built with React.js, Next.js, and Tailwind CSS, and is deployed on Vercel for scalability and performance. It leverages React components to construct an intuitive dashboard, displaying call logs in a structured format. Users can easily initiate and schedule calls, as well as review their previous interactions through the different pages.

The backend is made with FastAPI, designed for high-speed performance and seamless API communication. We use Twilio to handle call routing, enabling real-time voice interactions with AI-generated responses. OpenAI's real-time API powers the dynamic conversation, ensuring natural and adaptive speech that makes the call feel realistic.

⚔️ Challenges we ran into

Combining Twilio and the OpenAI real-time voice API to facilitate calling.
Integrating the backend real-time call data with the frontend dashboard.
Building the prompts for seamless natural conversation.

🎖️Our Accomplishments

We successfully integrated the frontend with the backend, and we are incredibly happy that we are able to call and utilize the realtimeAPI and get a responding voice AI with minimal latency, which is crucial for this app. We are also proud of successfully managing and navigating our database, which took some time to work with.