Resilience planning failed for years because the wrong models were applied: historical cat models for non-stationary climate, country risk scores for supply-chain nuance, ESG ratings without mechanism. Intelligence models for resilience are a distinct category—models whose training objective is to infer how systems fail and recover under stress.

The intelligence model stack

Earthian's resilience intelligence stack maps to hazard and failure modes:

Lucid Climate-0 — physical climate intelligence

Forward physical hazard inference at asset level: heat, flood, wind, drought propagation into operational and valuation impact. Resilience planning starts with where physical stress concentrates—not regional averages.

NatCat Lighthouse-0 — catastrophe intelligence

Natural catastrophe pathway reasoning: secondary peril amplification, accumulation zones, cascading infrastructure failure. Resilience for insurers, real assets, and governments depends on nat cat intelligence that updates with non-stationary peril.

Geopolitics Axiom-0 — geopolitical intelligence

Sanctions, conflict, trade disruption, and sovereign stress propagation through holdings and supply networks. Geopolitical resilience requires pathway models—not headline sentiment.

Technology Tenet-0 — technology and cyber intelligence

Vendor dependency, attack surface dynamics, obsolescence, and digital infrastructure failure modes. Operational resilience is increasingly technology resilience.

Policy Evergreen-0 — regulatory and transition intelligence

Policy shock, transition cost, and disclosure-driven repricing. Regulatory resilience is financial resilience when capital and labels depend on compliance trajectory.

Ichnos-0 — cyber risk intelligence

Specialized cyber inference for institutions where digital failure modes dominate tail risk.

Why categorized models beat one general model

Resilience failures are domain-specific in mechanism. A general LLM may narrate resilience convincingly while missing the compound link between grid stress, data-center exposure, and credit spread widening.

Intelligence models encode ontology and propagation logic per domain—then coordinate on Earthian Hub for portfolio-level resilience views.

Deployment patterns

• Pre-event: identify vulnerabilities, size mitigations, set inferred tolerance bands
• In-event: refresh pathways as shocks unfold; redirect live sessions with risk teams
• Post-event: replay inference for governance; feed outcomes into model calibration

Bottom line

Intelligence models for resilience are Earthian's core product thesis: categorized foundation models that institutions deploy to see shocks before they become losses—and to recover with evidence, not guesswork.