Using its Cognitive AI architecture, Droma observes diverse telemetry, anomalies and recurring incident patterns, analyzing how often similar incidents have occurred in the past. These insights are displayed on a Prediction Queue for easy monitoring and prioritization.

The system uses time between failures as the primary input for its predictive model and applies probabilistic inference to calculate the probability of a particular failure occurring at a given time horizon. Specifically, it determines how likely an incident is to happen given the previously observed time between failures and the time elapsed since the last occurrence.

By continuously tuning the model to maximize accuracy and minimize errors, Droma Predictive AI generates reliable predictions within defined time windows, such as the next 3, 6, or 12 hours, as well as longer-term forecasts like the next 24 to 48 hours.