DECISION INTELLIGENCE FOR THE FIRST 72 HOURS
Move aid before
roads disappear.
ReliefGrid turns fragmented disaster data into a transparent, equity-aware delivery plan—then stress-tests it against the routes most likely to fail.
< 80msto re-plan locally
240×route shock simulations
100%auditable allocations
MISSION CONTROL / DEMO DATA
Allocation command console
OPTIMIZER READY
ACTIVE RESPONSE
01 / NETWORK PLAN
Live response graph
Hub Need zone Shipment
Line weight = allocated kits · opacity = route reliability
02 / UNCERTAINTY
Plan resilience
240 SCENARIOS
—chance of ≥50%
coverage
coverage
P10MEDIANP90
———
Every plan is replayed against deterministic, seeded route shocks. The range exposes brittleness before a dispatcher commits.
03 / EXPLAINABLE PLAN
Zone-level allocation
| NEED ZONE | VULNERABILITY | WATER | MEDICAL | SHELTER | COVERAGE |
|---|
WHY THIS PLAN
OPTIMIZATION, NOT A BLACK BOX
Two objectives.
One accountable plan.
ReliefGrid separates the moral constraint from the efficiency objective. First it protects a configurable minimum service level for vulnerable communities; then it maximizes expected impact with the remaining stock.
- 01
Protect the floor
Reserve service for every zone, weighted by vulnerability. No community becomes mathematically invisible.
- 02
Optimize the remainder
Min-cost flow balances urgency, human impact, time, route capacity, and failure probability.
- 03
Stress-test the decision
Monte Carlo route shocks report the downside, median, and upside—using a reproducible seed.
DESIGNED FOR TRUST
A dispatcher can challenge every assumption.
Offline-first
Zero cloud dependency. The optimization engine runs entirely in the browser.
Human override
Equity and risk controls remain visible, reversible, and owned by responders.
Audit trail
Every allocation maps to a route, constraint, and declared objective.