Cassandra AI

• 

  architecture of Cassandra AI

• 

Inspiration

What it does Inspiration Emergency response delays cause thousands of preventable critical complications every year. Most dispatch systems operate purely on geographical proximity, routing patients to the nearest hospital regardless of whether its ER is overwhelmed or its ICU beds are full. Inspired by the myth of Cassandra—who foresaw crises that others ignored—we built Cassandra AI to give emergency response teams real-time, predictive insight into regional hospital capacities.

Inspiration

Emergency response delays cause thousands of preventable critical complications every year. Most dispatch systems operate purely on geographical proximity, routing patients to the nearest hospital regardless of whether its ER is overwhelmed or its ICU beds are full. Inspired by the myth of Cassandra—who foresaw crises that others ignored—we built Cassandra AI to give emergency response teams real-time, predictive insight into regional hospital capacities.

What it does

Cassandra AI is a multi-agent routing and resource allocation system for ambulances:

• Triage Scoring: Evaluates live patient telemetry (heart rate, blood pressure, trauma severity) to generate an immediate condition priority score.
• Capacity & Load Balancing: Continuously monitors regional hospital network data to identify ICU bed availability, specialist readiness, and ER queue length.
• Proactive Routing & Reservation: Automatically reroutes units to optimal care facilities rather than clogged nearby ERs, pre-reserving required medical assets before arrival.

How we built it

• Frontend Dashboard: Built with React and Tailwind CSS for a fast, responsive dark-mode dispatch interface.
• Multi-Agent Logic: Orchestrated autonomous agent workflows using the Strands Agents SDK to handle parallel evaluation of triage severity, hospital capacity, and route optimization.
• Development Environment: Rapidly prototyped and deployed via StackBlitz.

Challenges we faced

• State Synchronization: Handling fast dynamic updates between live patient telemetry inputs and multi-agent reasoning drawers without lag.
• UI/UX Balancing: Designing a dark dashboard layout that presents dense healthcare metrics cleanly without overwhelming dispatch operators.

Accomplishments that we're proud of

• Successfully creating a functioning multi-agent workflow that calculates optimal dispatch routes based on live resource capacity.
• Building a clean, end-to-end interface that simplifies complex medical triage into actionable decisions.

What we learned

• How to structure multi-agent task delegation effectively using Strands Agents SDK.
• How critical capacity-aware routing is compared to traditional distance-only dispatch algorithms.

What's next for Cassandra AI

• Integrating live traffic and mapping APIs for dynamic route duration prediction.
• Expanding multi-agent capabilities to support direct hospital EHR communication.

Built With

• amazon-web-services
• html
• javascript
• react
• strands-agents-sdk
• tailwind-css