Navigating Floods Without Data: Building Sentinel Voice Agent in 10 Days
It was during the peak monsoon season when I read a distress report from a family stranded on their rooftop. Power was flickering, rain was hammering against the walls, and cellular data was down to a crawling 2G edge. They had a phone with 14% battery, but opening an emergency app or downloading heavy government disaster PDFs was impossible. All they could do was place a direct phone call. That…
During heavy monsoon rains, a family was forced to take shelter on their rooftop. Power was unreliable, cellular service was slow, and they had only 14% battery life on their phone. The best they could do was make a direct phone call for help. This experience highlighted the need for a voice-based emergency system that could provide life-saving information instantly.
Inspired by this, I created Sentinel, an AI-powered dispatcher that uses speech-to-text, artificial intelligence and text-to-speech technologies to deliver critical details during disasters. The main problem was that existing emergency systems are fragmented across different departments and require users to interact with complicated user interfaces. With Sentinel, voice becomes the primary and most accessible method for receiving help.
Sentinel uses a streamlined pipeline to deliver results in real time. Incoming audio is converted to text using Deepgram Nova-3, which can understand Indian English and code mixed languages. The text is then analyzed by Google Gemini, which decides the next steps while following safety protocols. The response is then generated by Murf Falcon, which provides expressive speech in Indian English.
Sentinel also remembers previous interactions with callers, maintains privacy by requiring verbal consent to save data, and can provide live weather updates. This all happens in under a second, making it a reliable lifeline during floods and other emergencies.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
