Geo-Convergence
The globe split into 1° grid cells. When conflict, protest, military, and news signals cluster in the same cell within twenty-four hours, the system flags an emerging hotspot — before mainstream reporting catches up.
A Levenhall Product
Levenhall
INTELLIGENCE
Geopolitical intelligence for the corporate risk function.
The risk events that move share prices, disrupt supply chains, and reshape market access do not announce themselves in the news cycle. They surface in signal — weeks or months before the headlines.
Levenhall Intelligence is the operating picture we built for the corporate risk function. Enterprise risk officers, geopolitical risk leads, supply chain teams, security functions, and investment committees with international exposure all work the same way: one screen, every layer, on a globe that updates in real time.
Heavy machine learning runs in a Web Worker on your own device. Nothing leaves the browser. No round-trip to our servers.
Open it. The rest of this page explains what you'll see.
The Country Instability Index.
A composite 0–100 score across thirty-one countries that drive corporate risk exposure — markets where your supply chain runs, your assets sit, or your operating environment can change overnight.
(Baseline Risk × 0.4) + (Event Score × 0.6) + Hotspot Boost + News Urgency Boost + Focal Point Boost + Displacement Boost + Climate Stress
The Event Score itself is weighted: Conflict 30% Unrest 25% Information 25% Security 20%. Active wars score 70 or higher automatically. A million displaced adds 8. Each country carries a multiplier from 0.3× to 3× to compensate for media suppression and expected event frequency.
Output Low Normal Elevated High Critical
Strategic Risk Composite
One score per region. Geo-convergence at thirty percent, the CII at fifty, infrastructure cascade risk at twenty. The single number a board paper can carry.
Signal Aggregation
A unified bus that ingests conflict, unrest, economy, cyber, military, and climate streams, prioritizes by urgency, and feeds the rest of the dashboard.
On-Device ML
Four ONNX models in a Web Worker: MiniLM for embeddings, DistilBERT for sentiment, FLAN-T5 for summarization, BERT for named-entity recognition. Two-hundred-megabyte budget. Zero server roundtrips.
Infrastructure Cascade
Models dependencies between undersea cables, power grids, pipelines, and data centers. One disruption tells you what falls next — useful when your operations run through any of them.
§ The globe
34 layers on a WebGL globe. 31 markets you can drill into.
Eleven layers default on — conflicts, military assets, sanctions, hotspots, nuclear facilities, infrastructure outages. Twenty-three more a click away — undersea cables, pipelines, AIS tracking, cyber threats, displacement camps, financial centers, commodity hubs.
Core & Allied 🇺🇸 🇬🇧 🇫🇷 🇩🇪 🇵🇱 🇦🇺
Sanctioned & Adversarial 🇷🇺 🇨🇳 🇮🇷 🇰🇵
Active Conflict 🇺🇦 🇸🇾 🇾🇪 🇲🇲 🇸🇩 🇪🇹 🇨🇩 🇸🇴 🇦🇫 🇮🇶 🇱🇾 🇳🇬 🇱🇧
Strategic Emerging 🇮🇳 🇵🇰 🇸🇦 🇹🇷 🇦🇪 🇧🇷 🇻🇪 🇹🇼
Live Now
Open the dashboard.
Email-gated. Free to enter. Enterprise deployments and bespoke briefings available through Levenhall's intelligence practice.