ResQAI – Disaster Prediction, Relief, and Recovery Assistant

An AI-powered multilingual assistant that predicts climate disasters, guides citizens in real time, and automates recovery using voice, vision, and low-code tech.

Comment

Inspiration

Climate disasters like floods, wildfires, and heatwaves are increasing due to climate change — and it's the most vulnerable communities that suffer the most. In rural and semi-urban areas, millions of farmers, workers, and citizens face unpredictable weather with little or no warning. Many don’t have smartphones with advanced apps, and they often rely on unreliable media or word of mouth for critical updates.

We were especially inspired by:

The lack of multilingual, accessible early-warning systems

The incredible potential of AI to democratize access to lifesaving insights

The urgent need to blend vision, voice, and real-time data into a single tool that even offline users can rely on

ResQAI was born out of a vision to create a voice-activated, AI-powered assistant that understands local languages, works offline, and delivers hyperlocal climate risk intelligence.

What it does

ResQAI is a multilingual, voice- and vision-powered AI assistant that helps individuals and communities prepare for, respond to, and recover from natural disasters. Key features include:

🔮 AI-powered prediction engine using satellite + historical weather data

🗣️ Voice assistant that answers queries like “Can I work today?” based on live risk analysis

📡 Mass alerts to users in affected zones when disasters are likely

📷 Vision module to detect damage or wildfires from user-uploaded images

📱 Offline-first disaster guide with first-aid, evacuation plans, and FAQs

🧾 AI-generated reports for NGOs and governments post-disaster

🌍 Local language support and inclusive design

It empowers farmers, civil workers, and everyday users with personalized, real-time, and accessible climate intelligence — even in low-connectivity regions.

How we built it

We used a combination of no-code and AI tools to rapidly build a powerful, production-ready platform:

Component Technology App Builder Bolt.new (UI + voice + workflow logic) Voice Assistant Google Cloud Speech-to-Text + Agentic AI framework Data & Backend Supabase (user storage, feedback, alerts), Firebase Weather & Risk Prediction OpenWeather API, simulated ISRO datasets, rule-based ML models Vision Detection Gemini Vision API + Teachable Machine Offline Support PWA with service workers in Bolt.new Report Generator Gemini Pro + Markdown-to-PDF via LangChain Deployment Netlify + Entri Domain for challenges

We designed the app mobile-first with a clean UI, added multilingual voice flows, and embedded all critical tools behind intuitive actions like “Ask for today’s risk” or “Generate report.”

Challenges we ran into

⚙️ Simulating live prediction models in a no-code environment required creative workarounds using logic blocks and pre-trained thresholds

🌐 Building offline-first functionality that felt seamless took time — caching critical data and fallback flows was tricky

🗣️ Multilingual voice support (e.g. “Kya aaj kaam safe hai?”) was tough to implement fluently

🧠 Making AI responses interpretable and actionable for non-technical users was a design challenge

🔁 Rate limits in Bolt.new paused development flow — required careful prompt management

📡 Real-time alerts and user-wide notification architecture in Bolt + Supabase had limitations — we had to innovate using alert databases and push logic

Despite these challenges, we pushed forward by blending simplicity with scalability.

Accomplishments that we're proud of

💬 Built a multilingual voice-first AI that can serve users in Hindi, English, and other languages

🔔 Developed a real-time mass alert engine that can notify all users in a danger zone — not just individuals

🛰️ Combined historical disaster data with live inputs to simulate prediction models in a no-code framework

🧾 Created AI-generated PDF reports with maps, images, and summaries — useful for governments and NGOs

🧠 Delivered vision-AI support where users can upload images of damage for instant classification

💡 Built the entire system using accessible tools (Bolt.new, Supabase, Gemini) in just days — with zero-code backend logic

What we learned

Agentic AI workflows can create highly interactive, context-aware systems — especially when combined with voice

No-code platforms like Bolt.new can support real-world production apps when used strategically

Designing for inclusion (voice, language, offline) is not a nice-to-have — it’s essential

AI becomes truly valuable when grounded in real use cases with local context, not just abstract capabilities

Building for extreme conditions (e.g. low-connectivity rural areas) requires thoughtful fallbacks and data handling

What's next for ResQAI – Disaster Prediction, Relief, and Recovery Assistant

ResQAI has the potential to become a global climate resilience platform. Next steps include:

☁️ Integrate real-time OpenWeather and satellite feeds via APIs

📞 Add WhatsApp bot and IVR/voice-call alert support

🧠 Evolve from rules-based logic to lightweight ML models with online learning

🗺️ Add map-based risk visualization and “safe zones” navigation

🧪 Launch pilot tests in 2–3 Indian states with rural farming communities

🏢 Partner with government disaster management cells and agri-tech companies

📈 Add user analytics and admin dashboards for NGOs/authorities

🌍 Translate app into 8+ regional languages to reach millions

We see ResQAI as more than a hackathon project — it’s a scalable, life-saving assistant for a climate-vulnerable world.

Built With

  • 13
  • action)
  • agentic
  • agentic-ai-framework
  • ai
  • alert
  • alerts
  • alerts)-openweathermap-api-(live-weather-data:-rainfall
  • api
  • apifirebase
  • app
  • assistant
  • bigquery
  • bolt-blocks-?-authentication-&-users-supabase-auth
  • bolt.new
  • climate
  • cloud
  • communities)
  • cyclone
  • data
  • database
  • detection
  • disaster
  • emergency
  • entri-(ionos-custom-domain-challenge)-?-databases-&-storage-supabase-(user-auth
  • firebase
  • firebase-auth-(fallback)
  • firebase-realtime-db-(iot-+-alerts)-?-live-weather-api-openweathermap-api-(live-rainfall
  • flood
  • framework
  • gemini
  • gemini-vision
  • generative
  • generator
  • geofencing
  • geolocation-api-?-report-generation-gemini-pro-+-markdown-to-pdf-pipeline-using-langchain-???-multilingual-support-google-translate-api
  • google
  • gps
  • hackathon
  • humidity
  • impact
  • india
  • innovation
  • iot
  • isro
  • isro-datasets-??-voice-&-agentic-ai-google-cloud-speech-to-text-api
  • langchain
  • langchain-(for-smart-response-chaining-+-report-generation)-?-vision-ai-google-vision-api
  • local-caching-?-alert-system-push-notifications
  • logic
  • low-code
  • machine
  • ml
  • modeling
  • msg91
  • multilingual
  • mvp
  • no-code
  • offline-first
  • prediction
  • predictive
  • pro
  • progressive
  • real-time
  • real-time-updates)-firebase-realtime-database-(iot-location-tracking
  • realtime
  • realtime-db)
  • report
  • resilience
  • response
  • risk
  • rule-based-pattern-matching
  • rural
  • satellite
  • sdg
  • service-workers
  • sms-api-(twilio-/-msg91-for-simulation)
  • social
  • speech-to-text
  • supabase
  • sustainable
  • system
  • teachable
  • teachable-machine-(image-classification)-?-no-code-app-builder-bolt.new-(frontend-ui
  • temperature)-?-offline-ready-features-progressive-web-app-(pwa)-architecture
  • translate
  • translate-api-?-genai-&-text-generation-gemini-pro
  • triggers)-?-platforms-&-deployment-netlify-(for-pwa-deployment)
  • twilio
  • vision
  • voice
  • voice-based
  • weather
  • web
  • wind
  • workflows
Share this project:

Updates