The Intelligence Layer
EOSYN Core™
Runs proprietary GeoAI models over harmonised data to score risk, detect change, and flag anomalies with stated confidence.
Overview
What It Does
EOSYN Core applies proprietary GeoAI models to the data cubes Fabric produces. The outputs are measurements, not pictures. A map shows what is there. An instrument tells you how sure it is. Core's outputs are risk scores with confidence intervals, change detection with significance levels, and anomaly alerts ranked by severity.
Every Core output carries a confidence score, a statistical read on how certain the model is. That is what makes the output defensible in a boardroom or a regulatory review. A Head of Risk can justify a pricing decision. A Director of Asset Management can prioritise a maintenance backlog. A public institution can allocate scarce budget. All of it backed by a number, not a hunch. Each output also references the exact model version that produced it, so results are reproducible months later.
Clients with specialised needs can commission custom models: one trained on your historical claims data, another tuned to a specific corridor or asset class. Custom models go through the same bias testing, performance evaluation, and model card publication as everything else before they ship.
Capabilities
Key Capabilities
Change Detection
Identifies and classifies physical change over time
Risk Prediction
Forward-looking risk scores with confidence intervals
Anomaly Detection
Flags unusual patterns that warrant a closer look
Forecast Layers
Predictive modelling for future scenarios
Confidence Scoring
A stated statistical certainty on every output
Model Versioning
Every output traces to the exact model version behind it
Custom Model Development
Bespoke models built for a specific portfolio or region
Deliverables
Example Deliverables
| Deliverable | Description |
|---|---|
| Risk Scores | Quantified risk assessments with confidence intervals |
| Anomaly Alerts | Real-time notifications for detected anomalies |
| Forecast Layers | Predictive outputs for scenario planning |
| Detection Models | Trained models for specific change types |
| Model Performance Reports | Accuracy, precision, recall, and bias metrics |
The Stack
How It Connects
Data moves up the stack. Orbit sources raw signals. Fabric turns them into measurement-grade data cubes. Core scores them for risk and change. Atlas delivers the decision into your systems.