CrisisCompanion

CrisisCompanion – Your AI ally for urgent situations, providing guidance, clarity, and support when every second matters.

Comment

Inspiration

In critical situations, whether emergencies or high-stress environments, human responders can get overwhelmed. We wanted to build a solution that combines AI empathy with actionable guidance, ensuring that help and clarity are available immediately when it’s needed most. CrisisCompanion aims to reduce response time, prioritize critical situations, and provide calm, precise assistance.

What it does

CrisisCompanion acts as an AI ally during urgent scenarios. Users can submit text or audio reports of emergencies, and the AI:

  • Provides empathetic, immediate responses

  • Summarizes the incident for administrators

  • Assesses severity and provides actionable recommendations

  • Handles follow-up interactions to guide users further

For administrators, the platform:

  • Ranks and prioritizes incoming incidents using the Snowflake API

  • Summarizes and categorizes bulk incident reports to reduce cognitive load

  • Offers a clear dashboard for monitoring, managing, and resolving incidents

How we built it

• Frontend: React with TypeScript, TailwindCSS for styling, and a smooth, responsive UI
• Audio input: Custom React component for recording voice reports
• AI integration: Google Gemini API for empathetic responses and automated recommendations
• Data management: Snowflake API to rank, summarize, and prioritize incident requests efficiently
• Backend: FastAPI for handling follow-up conversations and incident data persistence
• Deployment: Vite for fast frontend bundling

Challenges we ran into

  • Parsing AI responses consistently in JSON while maintaining natural, empathetic messaging

  • Handling location and audio inputs across different devices and browsers

  • Managing bulk incident prioritization in real-time while maintaining a responsive dashboard

  • Balancing user privacy with meaningful data collection for AI analysis

Accomplishments that we're proud of

  • Built a fully functional AI-assisted incident reporting system in a hackathon timeframe

  • Implemented Snowflake API to intelligently prioritize and summarize incidents, saving human administrators' time

  • Designed a responsive, modern UI that supports both text and audio inputs seamlessly

  • Created an end-to-end system where AI assists both the user and the admin in crisis scenarios

What we learned

  • Real-time AI assistance can significantly reduce cognitive load for humans in emergency workflows

  • Handling both text and audio inputs requires careful UI/UX consideration

  • Data prioritization and summarization are critical for scaling emergency response systems

What's next for CrisisCompanion

  • Expand AI capabilities to handle multiple languages and emergency types

  • Integrate real-time alerts for high-severity incidents

  • Add analytics and reporting features for admins to better understand incident trends

  • Explore mobile app deployment for faster access in the field

Built With

Share this project:

Updates