CareCoord

Inspiration

Hospitals are complex, fast-moving environments where coordination delays can directly impact patient outcomes. We were inspired by the idea of using autonomous agents to reduce the cognitive load on healthcare staff—automating routine but critical workflows like scheduling labs, managing bed availability, and tracking patient status. CareCoord was built to demonstrate how intelligent agents can seamlessly collaborate behind the scenes to make hospital operations more efficient, transparent, and responsive.

What it does

CareCoord is a real-time hospital coordination platform powered by autonomous agents.

It has four main tabs:

Patients — A live list of all active hospital patients pulled from MongoDB. Each card shows name, room number, status (Admitted / Pending Discharge / Discharged), and priority (High / Medium / Low). High priority patients are visually highlighted. Tapping a patient opens a detail view where staff can trigger actions like Request Lab or Transfer Bed.

Beds — A real-time view of hospital bed occupancy. Each bed shows availability (occupied vs available) and which patient is assigned.

Lab Results — Displays lab test results with patient name, test type, status (Normal / Abnormal), and timestamps.

Activity — A live feed of all agent actions, updating every 5 seconds. Events are color-coded by agent type and include human-readable timestamps ("2 minutes ago"). This creates a transparent, real-time view into system activity.

How we built it

CareCoord combines a modern mobile frontend with a distributed backend powered by cloud services and autonomous agents:

Frontend: Built in Swift, designed for clarity and quick interaction in high-pressure environments Backend: AWS Lambda + API Gateway for handling requests and routing Database: MongoDB for storing patients, beds, lab results, appointments, and activity logs Agents: Fetch.ai agents (Scheduling, Lab Results, Bed Management) running locally and exposed via ngrok Communication Flow: Swift App → API Gateway → AWS Lambda → ngrok → Flask Agents → MongoDB → UI updates

The Scheduling Agent is fully integrated end-to-end. It processes lab requests, determines priority-based scheduling, assigns departments, books appointments, and logs all actions.

Challenges we ran into

One of the biggest challenges was connecting the multiple systems in real time. Connecting a mobile frontend, cloud infrastructure, and locally running agents required careful handling of networking, routing, and data consistency.

Another challenge was agent integration. While all agents were built and functional individually, wiring them into a unified pipeline required setting up consistent APIs (via Flask) and ensuring reliable message passing.

We also had to balance speed vs. realism. For the hackathon, we seeded some data (like lab results) and reused test logs to create a convincing demo while still maintaining a real, working architecture.

Accomplishments that we're proud of

Built a fully working end-to-end pipeline from mobile UI → backend → autonomous agent → database → live UI update Successfully implemented a priority-aware scheduling system that dynamically assigns appointments Created a real-time activity feed that makes backend processes visible and understandable Designed a system where multiple agents collaborate, demonstrating a scalable architecture Delivered a polished, interactive demo with real data—not just mockups

What we learned

We learned how to design and implement agent-based systems and how they differ from traditional request-response architectures.

We also gained experience with:

Integrating cloud services (AWS Lambda, API Gateway) with local development environments Structuring real-time data pipelines. Designing systems with user-centered thinking, ensuring the UI reflects what matters most to hospital staff

What's next for CareCoord

Our next steps focus on making CareCoord production-ready and expanding agent capabilities:

Fully integrate the Lab Results Agent and Bed Management Agent with live endpoints Build a complete booking flow UI (test selection → confirmation → feedback) Redesign the interface to match a more polished, scalable design system Deploy agents to the cloud instead of local machines for reliability and scalability Expand agent intelligence to handle more complex hospital workflows

Built With

• agentverse
• ai
• amazon-web-services
• api
• atlas
• fetch.ai
• frontend:
• gateway
• lambda
• macos
• mongodb
• python
• swift
• swiftui