CRISP

An AI-driven agent, CRISP provides real-time hurricane support for Florida and Texas.Accessible via WhatsApp Message, ensuring usability during low connectivity.

Comment

CRISP - Crisis Information Support Platform

Inspiration

Natural disasters can leave people stranded without crucial information. We wanted to build a system that delivers real-time, actionable data to affected individuals through accessible communication channels like SMS.

What it does

CRISP is an agentic framework that processes FEMA and other disaster data to provide timely, life-saving information to people in disaster zones. Users can request and receive critical updates via SMS and WhatsApp, ensuring access to vital resources when internet connectivity is limited. We developed this on LangGraph

How we built it

  • AI Architecture: Used LangGraph to make an AI agent which is converstation aware.
  • Deployment: Designed an GCP-based architecture with CI/CD pipelines for seamless deployment.
  • Frontend: Built a Bootstrap-based UI with light/dark mode to improve accessibility.

Challenges we ran into

  • Handling real-time disaster data efficiently.
  • Ensuring SMS delivery works reliably under various network conditions.

Accomplishments that we're proud of

  • Successfully integrating FEMA data with a scalable backend.
  • Deploying a fully automated, cloud-based solution.
  • Building a user-friendly interface with dynamic light/dark mode.

What we learned

  • How to make a safe AI agent that can provide good and helpful responses.
  • Make sure that the data is realtime and reliable

What's next for CRISP

  • Expanding data sources beyond FEMA for better disaster response.
  • Enhancing AI-driven predictions for real-time alerts.
  • Adding support for more languages to reach a global audience.
  • Developing a mobile app for broader accessibility.

Built With

Share this project:

Updates