Using SOTA technology to manage time-pressing issues
Real-Time: 911 emergency calls are caught in real time by converting the audio stream into exploitable text, then analyzed with state-of-the-art NLP algorithms. Key relevant information is extracted, and the urgency of the situation is evaluated for an accurate intervention.
Analysis: Using graph optimization algorithms, available emergency units are dispatched in the most efficient way possible based on the need and urgency extracted from calls. Time to intervention is key, and the way first responders are dispatched makes a life-saving difference.
Feedback: AsTeR is more than an AI-powered API, it is also a platform on which dispatchers have an intuitive overview of all the communications. Relying on centralized data, it becomes easy for emergency centers and units to share information, for an optimal help from anyone when the time comes.
Tech for good, open-source, listen to our story
Whenever a natural disaster hits, everyone affected or panicking will tend to call 911 for support. Some of those calls have a higher priority than others, but the current responders are not able to know it before taking the call. CalAster has been designed to answer this problem. Built as a modular platform, powered by an AI-enabled API, CalAster catches incoming audio streams from saturated centers to extract key information and prioritize those calls before the responders answer those.
This project was founded in Berkeley, California, as part of the Call for Code 2019. In California, earthquakes and wildfires are more common than in other places in the US. Emergency centers are overwhelmed every time such event occurs. By switching to an automatic mode, people needing help still have the opportunity to expose their emergency, before being redirected to a responder that ended his previous call.