The entire application is available as a web wrapper for hackathon use at https://everaid.app under “Try EverAid,” with a full instructional page containing descriptions, guides, and videos explaining how everything works.

Inspiration

Being in an emergency situation without any phone signal and nowhere to get help made me realize how critical it is to have something ready for any scenario. I wanted a tool that could guide me step-by-step through what to do, even when completely offline.

What it does

  • Mobile-first AI emergency toolkit for any situation — works both online and offline.
  • Comes with 80+ (scalable to thousands) preloaded emergency packs across 4 core categories: Health, Survive, Fix, and Speak.
  • AI-powered custom scenarios with multilingual auto-detection and natural chat for guidance.
  • Offline-ready: all packs cam be downloaded locally, AI fallback logic when networks fail.
  • Step-by-step actionable instructions with urgency levels (EMERGENCY, WARNING, INFO) and optional timers and checklists.
  • Creator tools to build, save, and share custom emergency packs with communities or organizations.
  • Open-source and privacy-respecting — self-host or run locally for schools, rescue teams, or remote areas.

How I built it

  • Frontend: React Native + Expo (mobile), React (web), PWA support.
  • Backend: Supabase Edge Functions (Hono) + PostgreSQL database.
  • AI: Together AI (gpt-oss-20b) generating structured JSON instructions.
  • Offline-ready: Service Worker caching + local AI fallback for critical content.
  • Design: Tailwind CSS + custom high-contrast UI components.

Challenges we ran into

  • Creating short, clear, and reliable AI instructions under stress.
  • Implementing a mega-fast AI response
  • Designing a UI that is fast, low-light-friendly, and works with minimal taps.

Accomplishments that we're proud of

  • Fully functional cross-platform any emergency assistant (iOS, Android, Web).
  • Creating and managing packs
  • Bringing solution to help humanity
  • Fast real-time responses with smooth animations
  • Multilingual support with automatic language detection.

What we learned

  • Designing for crisis situations requires a different UX approach.
  • Offline-ready systems must be planned from the very beginning.
  • AI content needs to balance flexibility with absolute clarity for the user.

What's next for EverAid

  • Expand language support.
  • Add voice-guided, hands-free instructions.
  • Build an open, community-driven library of emergency packs.

Built With

Share this project:

Updates