Gallery
Description
MENDUS — From Public AED Data to Real-World Access
What Is MENDUS?
Just because an Automated External Defibrillator (AED) is registered in public data doesn’t necessarily mean people can access it during an emergency.
For example, an AED may be installed inside a building with restricted entry, or its registered operating hours may differ from the hours when the public can actually enter.
MENDUS is a community-driven platform designed to identify and help close the gap between registered AED information and real-world accessibility.
It brings together public AED data, citizen reports, volunteer field observations, coordinator reviews, and AI-assisted follow-up actions into one connected workflow.
How Does MENDUS Work?
MENDUS supports three user roles:
- Citizen: Search for AEDs and report potential accessibility issues without logging in.
- Supporter: Review citizen reports and submit field observations for approved missions.
- Coordinator: Generate AI-assisted investigation plans, approve missions, review field observations, and manage facility follow-up actions.
MENDUS integrates Korean public AED data through a REST API and uses two AI agents:
- MENDUS_AI: Generates investigation drafts and suggests field verification steps based on citizen reports and registered AED information.
- MENDUS_Facility_Agent: Drafts cooperation requests to facility operators based on reviewed information.
The core principle is Human-in-the-Loop. AI assists with drafting and workflow preparation, but people remain responsible for final decisions and approvals.
Try the MENDUS Demo
Application URL: https://mend-hack26.outsystems.app/MENDUS/
MENDUS supports both Korean and English interfaces.
The demo lets you explore how a single AED accessibility concern moves from a citizen report to facility follow-up actions.
7-Step Demo Scenario
STEP 1. Citizen Report
A citizen searches for an AED without logging in and submits a report about a potential accessibility issue.
STEP 2. Mission Request
A Supporter logs in, reviews the citizen report, and determines whether further field investigation is needed.
STEP 3. AI Mission Plan
A Coordinator uses AI to generate a draft investigation plan, reviews the proposed verification steps, and approves the mission.
STEP 4. Field Observation
A Supporter opens the approved mission and submits observations about the AED’s location, accessibility, and operating hours.
STEP 5. Coordinator Review
A Coordinator reviews the submitted field observations and determines whether additional verification or facility follow-up is necessary.
STEP 6. AI Follow-up
AI generates a draft cooperation request, including a subject and message body, for the facility operator. The Coordinator reviews and saves the draft.
STEP 7. Facility Reply
The Coordinator records and manages facility follow-up activities and responses.
The full workflow can be explored using the same AED and citizen report throughout all seven steps.
For the demo, Supporters are assigned to three areas: Yeonsu-gu and Michuhol-gu in Incheon, and Gangnam-gu in Seoul.
Technology and Key Features
MENDUS was developed using the following technologies and design principles:
- OutSystems Developer Cloud (ODC): Web application development and workflow implementation
- Public Data REST API: Integration of Korean AED registration and operating-hours information
- Two AI Agents: Draft generation for investigation plans and facility cooperation requests
- Structured AI Output and Server-Side Validation: Improving the reliability of AI-generated results
- Role-Based Access Control: Separation of permissions for Citizens, Supporters, and Coordinators
- Human-in-the-Loop: Human review and approval for critical decisions
- Korean and English Support: Language-switching functionality
The AI agents generate drafts using the information provided within the workflow. An external semantic-search knowledge base is not currently connected.
Facility cooperation requests are not sent automatically. Actual communication and responses are managed separately by the responsible Coordinator. AI features may occasionally be unavailable depending on the model service status.
Building MENDUS with AI Mentor
As I’m still learning OutSystems, I relied extensively on AI Mentor throughout the development process.
With AI Mentor’s assistance, I was able to overcome technical challenges and experience building an application aimed at addressing a real-world social problem.
Why MENDUS Matters
Public data tells us where AEDs are registered. MENDUS helps communities verify whether those AEDs are actually accessible and take steps toward improving access.
In the future, this approach could be extended to other public services where real-world accessibility matters, such as cooling shelters during heatwaves.
Thank you for taking the time to explore MENDUS!
I hope this project helps demonstrate how public information can be transformed into meaningful action and positive change within our communities.
MENDUS — MEND + US. Together, we can close the accessibility gap.
From Public Data to Real-World Access.