AI · Emergency Tech

RAPID-100.

An AI-powered emergency call triage system that transcribes calls in real time, classifies incidents by type and severity, and generates structured summaries — helping dispatchers respond faster when every second counts.

Real-timeLive transcription during active call
AIGemini-powered incident classification
<3sSummary generation latency
4Severity levels — low to critical

Seconds lost cost lives.

Emergency dispatchers handle high volumes of distress calls under extreme pressure. Critical details — location, incident type, number of people involved — are often buried in panicked, fragmented speech. Dispatchers must listen, process, and act simultaneously, with no AI assistance.

RAPID-100 removes that bottleneck. It listens alongside the dispatcher, transcribes in real time, and surfaces structured intelligence the moment it's needed.

AI / ML
Google Gemini APINLPSpeech-to-Text
Backend
PythonFastAPIWebSocket
Frontend
HTMLCSSJavaScript
Infra
TwilioREST APIReal-time Processing

From call to classified incident

01

Emergency Call Received

A distress call comes in to the RAPID-100 helpline. The system immediately begins real-time audio capture and transcription.

02

Live Transcription

Speech-to-text converts the caller's voice to text in real time, displayed on the dispatcher dashboard as the call progresses.

03

AI Triage Analysis

Gemini AI analyses the transcript continuously — identifying incident type, severity level, location cues, and key details.

04

Incident Classification

The system classifies the emergency into a type (medical, fire, accident, etc.) and assigns a severity score from low to critical.

05

Summary Generated

A structured incident summary is generated and displayed instantly — giving dispatchers everything they need to act without re-listening.

Built for dispatchers.

The dispatcher interface shows four things at a glance — call status, live transcription, incident type, and severity. No clutter, no delay. The AI summary panel updates as the call progresses, so by the time it ends, the dispatcher already has a complete picture.

RAPID-100
emergency helpline
Live
Transcription

"There's been an accident on the main road near the college... two cars... someone is not moving..."

AI Summary

Road accident. Possible injury. Location: near college main road. Minimum 2 vehicles. Casualties unknown.

Built by three.

  • Designed and built the real-time transcription pipeline using Speech-to-Text
  • Integrated Google Gemini API for incident classification and summary generation
  • Built the FastAPI backend with WebSocket support for live data streaming
  • Designed the dispatcher dashboard UI — focused on clarity under pressure
  • Implemented severity scoring logic (low / medium / high / critical)
  • Connected Twilio for call handling and audio capture

View the code.

GitHub Repository