back to projects
ReliefCopilot project visual

health ai / disaster response

ReliefCopilot

an offline AI assistant that turns messy disaster field notes into action plans, ICS-201 briefings, multilingual messages, and evidence-backed guidance.

featuredopen-source

overview

ReliefCopilot helps responders work when the internet is down. It uses local LLMs, retrieval over humanitarian standards, and safety guardrails to generate structured response plans from chaotic field notes.

why it matters

ReliefCopilot focuses on useful AI under constraint: low connectivity, noisy field notes, multilingual communication, and decisions that need evidence-backed structure quickly.

what i built

  • converted messy disaster-response notes into structured action plans and ICS-201 style briefings.
  • used local LLM workflows and retrieval over humanitarian guidance for low-connectivity environments.
  • generated multilingual communication drafts for affected communities and field teams.
  • kept outputs grounded in evidence-backed response guidance instead of generic chatbot answers.