About the project

Operational intelligence with visible uncertainty

OpsIntel explores how behavioral baselines and multi-source evidence can help operators investigate physical-operation disruptions. The facility, entities, events, incidents, and evaluation truth are deterministic synthetic data.

What is implemented

The simulation, anomaly scoring, bounded incident correlation, NetworkX evidence graphs, ranked hypotheses, counter-evidence, operator feedback, and evaluation are working code. No paid model API is required.

Limitations and responsible use

Correlation is not causal proof. Scores depend on simplified synthetic baselines, and recommendations are review prompts rather than autonomous physical-control instructions. Anomaly severity and root-cause confidence are intentionally separate.

Independent proof-of-work prototype demonstrating Operational AI concepts using synthetic data.