Inspiration
The inspiration for DriveWise AI came from the desire to transform insurance risk assessment using real-time driving and traffic data. Traditional insurance models rely on static, historical data, which often fails to capture the dynamic nature of road conditions and driver behavior. By integrating live traffic feeds, vehicle safety data, and AI-powered analytics, we aimed to create a smarter, fairer, and more transparent insurance platform. Our goal was to empower both insurers and drivers with actionable insights, reduce risk, and ultimately make roads safer for everyone.
What it does
DriveWise AI is a full-stack platform that provides:
Real-time risk scoring for drivers using live traffic and vehicle data Insurance portfolio analytics and management AI-powered conversational agent for insurance queries and driving advice Interactive dashboard for visualizing risk, user data, and portfolio performance Live streaming of traffic and user data, with controls to start/stop data ingestion
How we built it
Backend: Python FastAPI serves as the core API, handling insurance, risk scoring, user management, and AI chat endpoints. Real-time data is ingested from TomTom (traffic) and NHTSA (vehicle safety) APIs using async streaming tasks. Frontend: React with Tailwind CSS powers the dashboard, featuring live streaming controls, real-time user display, and an embedded AI chat widget. Data Pipeline: Custom connectors fetch and process external data, feeding into risk models and user analytics. ML/AI: BigQuery ML and Vertex AI (mocked for demo) provide risk modeling and conversational intelligence. Process Management: Async tasks and global state ensure smooth start/stop of live data, with robust error handling and logging. Deployment: Docker and setup scripts streamline environment setup and service orchestration.
Challenges we ran into
Managing multiple async processes for real-time data streaming and user generation Handling port conflicts and ensuring clean startup/shutdown of backend/frontend services Integrating external APIs (TomTom, NHTSA) with robust error handling and rate limiting Designing a scalable architecture for live data ingestion and analytics Building a seamless frontend experience with dynamic controls and real-time updates Mocking AI responses for demo while maintaining extensibility for real Vertex AI integration Accomplishments that we're proud of Successfully built a working prototype with real-time risk scoring and insurance analytics Integrated live traffic and vehicle data into actionable insights Developed a user-friendly dashboard with live streaming and AI chat Achieved smooth process management for starting/stopping data ingestion Created comprehensive documentation and a compelling project story
What we learned
The power of async programming for real-time data applications Best practices for API integration, error handling, and process management How to design clean, scalable architectures for full-stack AI platforms The importance of user experience in data-driven dashboards How to communicate complex technical solutions in a clear, compelling way
What's next for DriveWise AI
Integrate real Vertex AI for production-grade conversational intelligence Expand data sources to include telematics and driver behavior analytics Enhance risk models with advanced ML techniques and continuous learning Partner with insurers for pilot deployments and real-world validation Add mobile support and push notifications for driver safety alerts Explore regulatory compliance and data privacy enhancements
Log in or sign up for Devpost to join the conversation.