The decision logic can be assessed by the engine in batch processing mode or in real-time. Each transaction is immediately processed by the engine in real time. The engine downloads your dataset for batch processing, completes all transactions, and then uploads the updated dataset.
You must create and release your decision flow for both use cases. You must then deploy it in real-time to an endpoint that will carry out the released decision flow. You must decide which release will be used to evaluate your transactions when batch processing.
Credit scoring, anti-fraud transaction approval, and real-time product pricing are examples of uses for real-time processing.
The pre-scoring of customer portfolios is the main use of batch processing.