AI based Real Time Order Anomaly detection to detect any fraud or an unusual order. It will help sellers to isolate the anomalous orders so that same can be reviewed before releasing for execution. This will support sellers to avoid any fraud orders and for better utilisation of inventory for normal orders.

The AI model in Ordazzle is based on a fast & robust ML algorithm which explicitly isolates anomalies rather than profiles normal instances. Most anomaly detection methods involve building out a 'distribution' of normal based behaviours and then training a 'model' or threshold, to identify periods that do not conform to this 'normality' definition. While the AI model we have used take a fundamentally different approach; optimised specifically to isolate and identify anomalies. Anomalous data points are further away from the regular observations in the feature space.

The highlight of this AI mode is that may flag multiple false alarms but will not have any false negative i.e. will not miss a single anomaly. We intent to use same ML algorithm for detecting any unusual pattern in order processing to flag an alarm for possible SLA breach of order ship.