About UNOCHA
At the United Nations Office for the Coordination of Humanitarian Affairs (OCHA), we bring the world together to tackle humanitarian emergencies and save the lives of people caught in crises.Activities by UNOCHA
EPIC – Exposure & Population Impact Console
Global Not Applicable
EPIC (Exposure & Population Impact Console) is a high‑performance, AI‑enabled geospatial analytics platform designed to automatically detect natural hazard events—such as earthquakes, floods, and cyclones—and rapidly estimate their potential impact on human populations.
The system continuously ingests authoritative hazard data from external providers, triggers automated analysis workflows, and combines hazard footprints with high‑resolution population datasets to calculate exposure and impact metrics. Results are disseminated through APIs, interactive maps, and automatically generated situation reports to support rapid humanitarian analysis, preparedness, and early decision‑making.
EPIC is designed for operational use in time‑critical contexts, emphasizing speed, scalability, and reproducibility while remaining modular, cloud‑deployable, and open for integration with downstream analytical and coordination platforms.
Ask ReliefWeb Q&A Chatbot
Global United StatesThe Ask ReliefWeb Q&A Chatbot allows a site visitor to ask questions about the (single) report page that they are viewing. On average, a single report had an attached file of approximately 17 pages. The Q&A Chatbot retrieves answers from that material.
Activity Type AI Tools/SolutionsOCHA RAG (Retrieval‑Augmented Generation) – AI‑Powered Situation Analysis Platform
Global AfghanistanEthiopiaIranJamaicaPakistanPapua New GuineaPhilippinesSri LankaUkraine
Objective
To strengthen humanitarian decision‑making by accelerating and improving the quality, consistency, and scalability of situation analysis and other Humanitarian Programme Cycle (HPC) analytical products (e.g. Situation Analyses, Flash Appeals, HNRPs) using AI‑assisted retrieval and synthesis grounded in verified sources.
What OCHA RAG Does
Applies retrieval‑augmented generation (RAG) to large volumes of humanitarian documents (UN, NGO, government, media, assessments).
Ground all AI‑generated text directly in cited source material to reduce hallucination risk.
Produces structured, analyst‑reviewed draft analyses, with human‑in‑the‑loop validation at all stages.
Supports rapid triangulation of complex, unstructured information across crises.
Current Status (Latest)
Operational pilot in use since June 2025, with continuous refinement through 2025–2026.
Applied across multiple sudden‑onset emergencies and protracted crises, supporting the production of situation analyses and related HPC outputs.
Currently operates as a locally managed analyst tool (not yet a corporate‑wide production system).
Technical specifications for secure Azure deployment and multi‑country scaling have been developed.
Innovation Aspect
Moves beyond generic AI chat tools toward a domain‑specific, governance‑aware humanitarian AI system.
Transition underway from document‑centric RAG to Graph‑based RAG using humanitarian ontologies, common operational datasets (CODs), and analytical frameworks such as JIAF.
Designed explicitly for human‑controlled analytical drafting, not automated decision‑making.
Key Results to Date
Faster analytical turnaround during emergencies.
Improved consistency and analytical framing across products.
Demonstrated feasibility of AI‑assisted situation analysis under strict humanitarian governance and quality controls. Over 20 countries supported
Next Steps
Secure funding to scale OCHA RAG into a field‑ready, secure, multi‑user platform.
Expand pilot deployments to country offices.
Strengthen governance, security review, and institutional integration.
Automated tagging of ReliefWeb content
Global United StatesLeverage a Large Language Model to automatically apply certain categories to content being posted to ReliefWeb - https://reliefweb.int/. The intention is to reduce human efforts in areas that where AI can provide a suitable solution thereby allowing our human capacity to focus on more impactful activities. We are using LLM without any fine-tuning as we have found the results to be more than sufficient ‘as is’.
Activity Type AI Tools/Solutions