-
-
Auto-generated SOS message in the user’s local language, ready to broadcast or call emergency services with one tap.
-
Emergency scan screen where users capture the scene. CrisisSight analyzes danger levels and prepares guidance instantly.
-
Immediate survival checklist showing calm, step-by-step actions tailored to the detected disaster scenario.
-
Tactical map view highlighting nearby hospitals, fire stations, and shelters to help users navigate to safety quickly.
Inspiration
What it does
CrisisSight — Guidance When It Matters Most
Inspiration
When disasters strike, people don’t panic because they lack courage — they panic because they lack clear guidance. In floods, fires, or building collapses, most people only have their phone and a few seconds to decide what to do. CrisisSight was built for that exact moment.
What We Built
CrisisSight is an AI-powered disaster assistance tool that turns photos or short videos from disaster sites into clear, actionable guidance. Users upload an image or video, and CrisisSight analyzes the situation to identify the disaster type, assess urgency, and suggest safe next steps.
Instead of overwhelming users with information, it provides a short checklist and a ready-to-send SOS message. It also explains why certain actions are recommended, helping users trust the guidance during stressful situations.
What We Learned
We learned that in high-stress scenarios, clarity is more important than complexity. Responsible AI design, explainability, and knowing when to communicate uncertainty are essential in safety-focused systems.
Challenges
The biggest challenge was balancing helpfulness with safety. We carefully avoided medical diagnoses, hallucinated details, and overconfident recommendations, especially when visual information was unclear.
Impact
CrisisSight supports people while emergency help is on the way by helping them make safer decisions, reduce confusion, and feel less alone during moments that matter most.
How we built it
Challenges we ran into
Accomplishments that we're proud of
What we learned
What's next for Crisis Sight
Built With
- actionable-guidance.-users-upload-an-image-or-video
- and-crisissight-analyzes-the-situation-to-identify-the-disaster-type
- and-knowing-when-to-communicate-uncertainty-are-essential-in-safety-focused-systems.-###-challenges-the-biggest-challenge-was-balancing-helpfulness-with-safety.-we-carefully-avoided-medical-diagnoses
- and-overconfident-recommendations
- and-suggest-safe-next-steps.-instead-of-overwhelming-users-with-information
- api
- assess-urgency
- clarity-is-more-important-than-complexity.-responsible-ai-design
- css3
- especially-when-visual-information-was-unclear.-###-impact-crisissight-supports-people-while-emergency-help-is-on-the-way-by-helping-them-make-safer-decisions
- explainability
- fires
- gemini
- hallucinated-details
- helping-users-trust-the-guidance-during-stressful-situations.-###-what-we-learned-we-learned-that-in-high-stress-scenarios
- html5
- it-provides-a-short-checklist-and-a-ready-to-send-sos-message.-it-also-explains-why-certain-actions-are-recommended
- javascript
- most-people-only-have-their-phone-and-a-few-seconds-to-decide-what-to-do.-crisissight-was-built-for-that-exact-moment.-###-what-we-built-crisissight-is-an-ai-powered-disaster-assistance-tool-that-turns-photos-or-short-videos-from-disaster-sites-into-clear
- node.js
- or-building-collapses
- people-don?t-panic-because-they-lack-courage-?-they-panic-because-they-lack-clear-guidance.-in-floods
- react
- reduce-confusion
- typescript
Log in or sign up for Devpost to join the conversation.