Skip to Content

Ever thought about Machine Learning in Finance for Wholesale Distribution?

Today, Wholesale Distribution companies take often a rule-based approach to apply payments to cash to achieve automation, reflecting customer and country specifics. However,

  • rule set-up increases implementation costs
  • rule effectiveness declines over time.

The new “Cash Application” of SAP is a choice for Wholesale Distributors where automation can pay out quickly. The SAP Cash Application uses the SAP Leonardo Machine Learning capabilities.

Historical clearing information is sent from an S/4HANA Finance system to the Cash Application. The Cash Application uses a Mashine Learning business service (API) running on the SAP Cloud Platform. The business service consists of an algorithm which describes a model to derive matching criteria.

As part of daily business in your S/4HANA Finance system, new incoming payment and open invoice information is passed to the Cash Application. Proposals are returned to S/4HANA and those that meet the configurable confidence threshold are cleared automatically. When there are multiple proposals for a payment, they are suggested for review to the AR accountant within the standard Fiori app they use today. The model can be adapted or trained on a regular basis to ensure that changing behavior of the AR accountant is captured without on-going maintenance.

The machine learning platform evaluates historical clearings to determine the best model and learns from the accountant behavior.With the flexible adoption, the system can capture more details of a customer- and country-specific behavior, without the costs of manually defined detailed rules.

The SAP Cash Application is a hybrid scenario where the machine learning capability is delivered by SAP Leonardo ML Foundation (in SAP Cloud Platform). It runs for S/4 HANA Enterprise Management 1702 Public Cloud and requires FI-AR. It is also available for the first time with S/4 HANA Enterprise Management 1709.

You must be Logged on to comment or reply to a post.