Machine learning (ML) is a core part of Ai functionality in SAP S/4HANA. However, for any machine learning model to work, there must be an understanding of what the business needs and how it intends to achieve those goals. Embedded AI can collect data, but it must follow a standard for that data to be usable. The ML models implemented in SAP follow the cross-industry standard process for data mining (CRISP-DM). Once the business knows what it wants to do with its data, the AI agent can collect it, develop a model, train itself on it, evaluate its results based on existing information, and improve iteratively.
The Difference Between Embedded AI and Advanced AI
Many of the solutions that SAP offers to customers come with their own AI. This AI uses a database of standard functions such as regression calculations, classification, clustering, and time series. These essential embedded AI functions are part of the scope of many of SAP’s products, like SAP SuccessFactors, for example. While these crucial functions are helpful and have proven their worth, advanced AI needs a bit more to impact real-world problems. As such, complex algorithms comprising several smaller computations may be required. Complicated machine learning tasks such as image recognition or natural language processing rely on a side-by-side approach using SAP Business Technology Platform (BTP). These advanced algorithms can be accessed from the “AI Business Services” product portfolio and integrated into any SAP or non-SAP system.
Predicting Scheduled Delays with a Preconfigured AI Scenario
As mentioned before, complex ML scenarios like prediction may need multiple algorithms to get their work done. SAP ships with over twenty ML scenarios, one of which allows for predicting schedule delays for outbound deliveries with SAP S/4HANA. Sales reps can oversee the delivery processing as part of the service offering to the client. But how does the sales rep get the information that may inform them of a delay in deliveries?
SAP Fiori has an app named Predicted Delivery Detail that presents the information processed by the ML agent in a simple, easy-to-digest way. Using the source information for how much of a delay (in days) there was for the creation and processing of the order, it can give an updated view (in real-time) of what the estimated delay on the delivery will be. The graphic display makes it easy to spot inefficiencies and seek out solutions for them. This approach also allows for just-in-time intervention for critical supply chain decisions, ensuring that deliveries get made within a reasonable time frame.
The Power of Machine Learning
If a human were to try to work out these delays, it could take years of data collection, by which the data would be outdated. An average human being cannot process elastic systems like supply chains because we learn so slowly. AI can iterate on its learning process rapidly, making it able to learn hundreds (even thousands) of times faster than the quickest human learner. Whether you’re shipping lettuce, or a phone headset, you should be able to give an estimated delay on the product’s delivery. SAP’s embedded AI and specialized programs offer you an excellent way to do so.