-
-
Urban Sentinel SustAInxt
Inspiration We live in a world where food can be delivered in minutes, taxis arrive instantly, and information travels at lightning speed yet during emergencies, help is often delayed. This contradiction inspired SustAInxt.
Aligned with the UN Sustainable Development Goals (especially SDG-11: Sustainable Cities & Communities and SDG-12: Responsible Consumption & Resilient Infrastructure), we asked a simple but powerful question:
If AI can optimize convenience, why can’t it optimize survival?
SustAInxt was born to bridge this gap using Gemini 3 to ensure that emergency response is as fast, intelligent, and accessible as modern digital services, while remaining sustainable and inclusive.
What it does SustAInxt is an AI-powered smart city emergency platform that enables real-time incident detection, analysis, and response.
Citizens upload images or videos of incidents (fire, flood, accident, disaster).
AI instantly analyzes the situation, identifies severity, and extracts location.
The system provides clear safety instructions, visualizes the incident on a live map, and alerts authorities automatically.
Information is translated into multiple languages to ensure inclusivity.
Verified incidents are shared with the community for awareness and prevention.
All of this happens in real time, minimizing response delays and reducing impact.
How we built it SustAInxt is entirely achieved using Gemini 3 models as the core intelligence layer.
🔧 Technical Stack Gemini 3 Vision – Incident detection, classification, and severity analysis from images/videos
Gemini 3 Multimodal Reasoning – Contextual understanding, recommendations, and decision logic
Gemini 3 Language Models – Multilingual translation and instruction generation
Next.js + React – Frontend UI for uploads, map view, and live updates
OpenStreetMap – Tactical GIS visualization
Twilio (Voice/SMS) – Emergency alert dispatch
Cloud-based deployment (Vercel) – Scalable and sustainable hosting
Challenges we ran into Designing a fully automated decision flow without human intervention
Handling real-time media analysis with accuracy and confidence scoring
Ensuring multilingual emergency clarity, not just translation
Balancing technical depth with user simplicity
Deploying sensitive alert workflows securely and reliably
Each challenge pushed us to deeply understand Gemini 3’s multimodal and reasoning capabilities.
Accomplishments that we’re proud of Built a complete end-to-end AI emergency response pipeline
Achieved vision, language, reasoning, and decision-making using only Gemini 3
Integrated real-time GIS mapping + AI voice dispatch
Designed a system aligned with global sustainability goals
Created a solution that is scalable, inclusive, and hackathon-ready
Most importantly, we built something that can save time — and potentially lives.
What we enhanced for this finale Round:
- Added multilingial (both text + Voice) for regional and other languages for better reachability.
- Fine-tuned Model reasoning capability.
- Added additional options to configure the Twilio numbers.
What we learned Multimodal AI can move beyond prediction into real-world action
Sustainability is not just environmental — it’s about resilient systems
Gemini 3 is powerful enough to act as a full-stack intelligence engine
Clear UX is critical when people are under stress
Responsible AI design is essential in emergency systems
What’s next for SustAInxt Predictive analytics to forecast emergencies before they occur
Integration with IoT sensors and city infrastructure
AI-based resource optimization for emergency services
Expansion to disaster management at national scale
Deeper alignment with SDG-11 & SDG-12 for global impact
Log in or sign up for Devpost to join the conversation.