Inspiration

The inspiration behind SafeHaven came from witnessing the chaos and confusion that surrounds natural disasters. In recent years, we've seen hurricanes devastate coastal cities, wildfires consume entire communities, and floods displace thousands—often with little to no warning. Despite the availability of weather apps and alert systems, there’s a critical gap: people receive warnings, but they don’t always know what to do next or where to go. During Hurricane Harvey in 2017, for example, tens of thousands of people were stranded without clear information on where to find shelter or medical help. In California, wildfires have forced entire neighborhoods to evacuate within minutes—leaving people scrambling for safety.

We realized that in moments of crisis, people don’t just need alerts—they need guidance. They need a digital companion that not only informs them of the threat but actively helps them navigate to safety.

SafeHaven was born from that need: a smart, real-time emergency assistant that combines disaster alerts with actionable, location-based recommendations. It's our way of turning technology into a lifeline—because when every second counts, clarity and direction can save lives.

What it does

What SafeHaven Does SafeHaven is an intelligent, real-time emergency response assistant designed to guide users to safety during natural disasters. It automatically detects your current location and checks whether your area is impacted by severe weather conditions such as hurricanes, floods, wildfires, or earthquakes. Using live data from trusted weather and disaster sources, SafeHaven determines the level of threat in your vicinity. Powered by Databricks and AI-driven analytics, the system then recommends the nearest safe zones—prioritizing hotels and hospitals with strong infrastructure and accessibility. It intelligently forecasts the availability of these facilities using historical occupancy data and regional trends, helping you make faster, smarter decisions when it matters most. Whether you're a local resident or a traveler caught in an unfamiliar place, SafeHaven ensures you don't just get alerted—you get guided to safety with clarity, speed, and confidence.

How we built it

SafeHaven is built on a robust data-driven architecture that combines real-time intelligence with predictive analytics to deliver life-saving recommendations. We aggregate live and historical weather data from trusted meteorological sources to monitor and forecast severe weather events. For hospitality and medical infrastructure insights, we utilize Bright Data to collect and structure information on nearby hotels and hospitals, including availability trends, amenities, and capacity indicators. At the core of our system is a powerful AI agent, built using Databricks and integrated with vector search capabilities. This agent analyzes your current location, cross-references it with disaster zones, and performs similarity searches across a vectorized knowledge base to identify the safest and most accessible nearby facilities. By fusing real-time data with historical patterns, SafeHaven doesn’t just react to emergencies—it anticipates user needs and delivers personalized, context-aware guidance when every second counts.

Challenges we ran into

One of the key challenges we encountered was designing and orchestrating multiple specialized agents—one focused on weather intelligence and another on identifying nearby hotels and hospitals. While separating these responsibilities allowed for modular development, integrating their outputs into a seamless, unified user experience proved complex. Coordinating communication between agents, ensuring consistent context sharing, and managing real-time data synchronization required careful architectural planning. We had to overcome issues related to data latency, decision overlap, and response prioritization to ensure the system delivered accurate, timely, and actionable recommendations without confusion or delay.

Accomplishments that we're proud of

Building meaningful connections and forming lasting friendships throughout the journey.
Gaining valuable exposure to diverse work styles and team dynamics, which enriched our collaboration and broadened our perspective.
Embracing cross-functional teamwork and learning how to adapt and thrive in a multicultural, interdisciplinary environment. ## What we learned We gained a deeper understanding of how AI agents operate, particularly in the context of vector search and semantic retrieval. This experience enhanced our knowledge of how to structure, index, and query high-dimensional data to deliver context-aware, intelligent responses—an essential capability for building responsive and scalable AI-driven applications. ## What's next for SafeHaven: Guiding You to Safety With the MVP successfully developed, SafeHaven is just getting started. We envision a powerful roadmap ahead, with opportunities to integrate advanced features such as real-time occupancy tracking, multilingual support, offline emergency access, personalized evacuation routes, and integration with public safety networks. Our goal is to evolve SafeHaven into a comprehensive disaster-response platform—one that not only alerts users but actively supports them before, during, and after a crisis.