Inspiration

Every year, natural disasters like floods, earthquakes, and cyclones impact millions of lives. Relief efforts are often slowed down by fragmented communication, delays in resource allocation, and lack of real-time coordination. Witnessing these challenges inspired us to build ReliefAI, an intelligent agentic AI platform designed to bridge the gap between victims, relief workers, and authorities — enabling faster, smarter, and more impactful disaster response.

What it does

ReliefAI acts as an AI-powered coordination hub during disasters. It:

Analyzes incoming data from social media, satellite feeds, and local reports.
Identifies affected areas, urgent needs, and potential hazards.
Connects victims with nearby relief resources.
Assists authorities in prioritizing response with clear, real-time insights.
Provides multilingual chatbot support for victims in distress.

In short, ReliefAI brings clarity in chaos by aligning relief supply with demand quickly and intelligently.

How we built it

Frontend: A lightweight web dashboard for responders and a chatbot interface for victims.
Backend: Python + FastAPI for APIs, integrated with IBM Watson Assistant for multilingual support.
AI/ML Models:
- NLP models (transformers) for processing social media/text reports.
- Computer vision models for analyzing disaster imagery.
- Prioritization algorithms to recommend optimal resource distribution.
Data: Open disaster datasets (UN OCHA, Kaggle crisis datasets, NASA satellite images).
Deployment: IBM Cloud & containerized microservices for scalability.

Challenges we ran into

Ensuring real-time accuracy of AI predictions during fast-changing disaster scenarios.
Handling noisy or incomplete data from social media feeds.
Designing a multilingual, low-bandwidth chatbot for areas with poor internet.
Coordinating multiple tools and APIs into one seamless system.

Accomplishments that we're proud of

Built an end-to-end AI pipeline that connects data → insights → action.
Developed a prototype chatbot that can guide disaster victims to nearby shelters.
Designed a system that is both scalable and globally adaptable for different disaster types.
Created an approach that has strong potential for social impact and AI for good.

What we learned

How to integrate agentic AI workflows with real-world humanitarian challenges.
The importance of bias-free, ethical AI in crisis situations.
Practical skills in combining IBM Cloud, NLP, and computer vision into a unified platform.
The power of collaboration under tight deadlines.