Relief Finder AI – Powered by Algolia MCP

#algoliachallenge #ai #mcp #devchallenge

Algolia MCP Server Challenge: Ultimate user Experience

This is a submission for the Algolia MCP Server Challenge.

Inspiration

Disasters like floods, earthquakes, and wildfires leave people vulnerable and disoriented. Getting real-time information about relief shelters, available resources, safety zones, and weather conditions can save lives. I wanted to create an AI-powered assistant that uses Algolia’s MCP tools to make this information searchable, intelligent, and fast.

What It Does

Relief Finder AI is a full-stack disaster response app that:

Lets users search for relief shelters with filters like food, water, and medical aid.
Uses Algolia MCP to intelligently select the right search index.
Offers a chat-based AI assistant to answer user questions naturally.
Displays real-time weather and safety scores for each shelter.
Fetches disaster alerts and shows them on an interactive map.
Fetches Shelter Reliefs and shows them on an interactive ui.

Demo

Source Code: GitHub Repository

Demo Video:

How We Built It

Frontend: React + Algolia InstantSearch + Leaflet + OpenWeather API
Backend: Django + Algolia MCP SDK + OpenRouter AI (AI models)
Data Sources:

Relief_Shelter index in Algolia for shelter info
disaster_alerts index for real-time threats
Weather from OpenWeatherMap API
AI assistant from OpenRouter

MCP Tools Used

searchSingleIndex – Used to search both relief shelters and disaster alerts from the appropriate Algolia index.
algolia_reindex – Used in the backend to import and reindex data dynamically into Algolia indices.
React InstantSearch – Used on the frontend to display and interact with search results using InstantSearch components.
Dynamic Prompt Generation – AI prompt is generated based on user input and current search context.
AI Tool Selection – The backend determines which MCP tool and Algolia index to use based on the user query using an AI model (e.g., DeepSeek/Mistrel).

Challenges We Ran Into

Building a tool-switching logic for AI to decide which index to use
Handling real-time weather and geolocation sync in React
Integrating MCP SDK cleanly with Django backend
Import data to my indexes through the Django backend

What I Learned

How to use Algolia MCP tools like searchSingleIndex and integrate them into a real-world application.
The process of setting up and reindexing Algolia indices from a Django backend using algolia_reindex.
How to build a React InstantSearch UI that connects seamlessly with Algolia for fast, filterable search experiences.
How to integrate AI models (DeepSeek/Mistrel) through OpenRouter and dynamically generate prompts based on user queries.
How to design a tool-selection logic so the AI assistant can choose the right Algolia index and return meaningful, context-aware responses.
How to combine multiple APIs (Algolia, OpenWeatherMap, OpenRouter) into one unified, intelligent disaster response system.