One explainable intelligence layer on top of every disaster feed on Earth. COVID outbreaks, floods, earthquakes, conflict — aggregated, scored, forecast and delivered before it's too late.
Four capabilities every humanitarian coordination cell needs — fused into one screen.
Earthquakes, cyclones, floods, wildfires, droughts, epidemics, displacement and conflict — ingested in real time from 7+ public feeds, 24/7.
Four ML models interpret the noise: XGBoost rates severity, DistilBERT reads reports, Prophet forecasts what comes next, HDBSCAN finds where it clusters.
AI you can audit. Every prediction shows its top contributing features in plain language — no black boxes in a command room.
Priority-ranked alerts hit the moment an event crosses the severity threshold — ranked the way an ops center ranks them.
Rates any in-progress event on a unified 1–5 risk scale using 7+ contextual features, backed by SHAP attribution.
Classifies incoming text reports into crisis types instantly — auto-tagging field notes and news wire copy.
Panels next weeks of crisis frequency per region with confidence bands so logistics can preposition resources.
Spots emerging crisis zones by density and severity — the difference between reacting and getting there first.
The moment an event crosses the severity threshold, an alert is generated with full context — geolocation, affected populations, displacement — and queued by urgency. Acknowledge, triage, act.
Renewed escalation displaces tens of thousands at the Turkish border crossing
Monsoon floods submerge low-lying districts, 5,000+ displaced
Cholera outbreak expands beyond containment zones
M8.1 earthquake triggers tsunami advisories
Free public data in. Explainable intelligence out. Everything in this project is open source, MIT-licensed, and runs on a single machine — or at planetary scale.