CIRO (Crisis Intelligence & Response Orchestration) is an autonomous, multi-agent AI system developed specifically for the Capital Development Authority (CDA), Islamabad. By leveraging advanced agentic workflows and structured data processing, CIRO transforms urban emergency management, making real-time, high-stakes decisions to minimize response latency across the capital.
To ensure operational security, CIRO is protected by a multi-layered authentication portal. Only authorized CDA personnel can access the live intelligence dashboard, maintaining the integrity of emergency operations.
CIRO operates on a highly structured multi-agent workflow that handles crisis data from initial report to final execution:
- Ingestion (Intake Agent): Real-time citizen complaints (via WhatsApp) and city signals are captured. The Intake Agent performs initial NLP-based triage to classify and validate the crisis against our core JSON data schemas.
- Commander (Core Agent): The central intelligence hub. It analyzes severity, context, and historical precedents using Gemini models to determine the optimal emergency response strategy.
- Dispatch (Action Agent): The system automatically executes the final command, triggering resource allocation, calculating routing protocols, and updating the live intelligence feed in Firebase so the mobile app reflects status changes instantly.
The CIRO Command Center provides a real-time view of Islamabad’s urban landscape. Below, the system illustrates a G-10 sector flash flood scenario, showing active rerouting strategies and emergency zones.
The CIRO dashboard features a tactical map powered by Folium. Our dispatch logic interacts with the map by:
- Dynamically updating coordinate markers based on incoming event data.
- Visualizing path optimizations for emergency vehicles navigating Islamabad’s road network.
- Injecting real-time route overlays directly into the map component to guide field personnel.
CIRO features a powerful "Execute Communication" module. When a crisis is confirmed, the AI does not just plan the response; it performs the outreach autonomously:
- AI-Generated Content: Upon clicking the "Execute Communication" button, the Commander Agent triggers specialized sub-agents to draft high-impact content.
- Multi-Channel Orchestration: The AI writes a platform-specific Twitter post for public awareness, an emergency SMS for field responders, and a professional email for CDA administration, all tailored to the specific nature of the event.
To experience the CIRO autonomous intake agent in real-time, you can report a simulated crisis via WhatsApp.
🔒 Security Note: To prevent spam and API abuse, our live Twilio testing number and Sandbox code are not public. Please find the phone number and join code provided directly in our official AI Seekho submission form.
- Initiate the Connection: Add the provided Twilio number to your phone.
- Join the Sandbox: Send the exact join code (e.g.,
join [word]-[word]) provided in our submission form to authenticate your number with our Cloud Run backend. - Report the Crisis: Once you receive the "You are all set!" confirmation from Twilio, act as a citizen. The AI Intake Agent is fully trained to synthesize unstructured Roman Urdu and English.
Try sending these exact test signals:
- "e-9 k pass toofan aya hai"
- "F-7 k paass bari aag lagai hai"
- Observe the AI Orchestration: * On WhatsApp: The Intake Agent will immediately process the severity and reply contextually.
- On the CIRO Dashboard: Open the mobile application and check the Citizen Complaint Registry. You will see your exact report instantly parsed, categorized, and synced via Firebase!
CIRO relies on a sophisticated architecture for handling city-wide data:
- Signal Injection: Through the "Inject Signal" interface, we can simulate complex scenarios (e.g., Flash Floods, Highway Pileups). Incoming signals are mapped against our JSON schemas, allowing agents to instantly parse and convert raw alerts into standardized formats required for agentic reasoning.
- Citizen Complaint Registry: Integrated with Firebase, this registry logs all incoming emergency reports for historical analysis and immediate task management.
- Backend: FastAPI (Python) for robust, high-performance API handling.
- AI Orchestration: Google Antigravity SDK and Vertex AI for advanced agentic decision-making.
- Frontend: Flutter for a cross-platform, high-performance mobile experience.
- Database: Firebase Firestore for real-time synchronization of the Crisis Registry.
- Infrastructure: Deployed on Google Cloud Run for seamless, scalable, and secure deployment.