Conversational AI and the Machine Learning Foundation

Whereas many business applications used to be about reducing costs by digitizing processes or increasing efficiency by simplifying processes, in the future the real added value will lie in automation through learning systems.

For this, offers for the use and integration of artificial intelligence and machine learning in own business applications by the large providers such as SAP, Microsoft or Amazon are numerous.

The portfolio of consumable AI services on offer varies from provider to provider, but overlaps in the case of easily deployable services. This refers to services that can be provided with either little or no training effort.

Nevertheless, SAP also offers a range of services and tools for the use of AI and machine learning to meet the requirements of automatable business processes and next-generation applications. From the customer's point of view, there is thus the possibility of combining transactional processes with intelligent tools.

In addition to the Predictive Analysis Library (PAL) as a complement in the Hana environment, SAP offers both Conversational AI and the Machine Learning Foundation with Leonardo Machine Learning. However, the challenges of modern business applications in interaction with AI are so individual and manifold that all these offerings can be applied to generate real added value.

With Conversational AI, SAP offers the possibility to automate a wide range of service processes with bots. Thanks to the ever-improving technology for understanding written or spoken natural language, queries in customer service or human resources, for example, can be largely automated and synchronized with SAP systems.

The SAP offering still shows weaknesses here in intelligent dialog guidance when a wide range of services is to be mapped as a bot. For those who want to apply AI methods to the processing of their business data, PAL offers a useful tool for optimizing their business applications directly in the Hana database.

With it, you can, among other things, recognize patterns in your sales, production or purchasing data and use them for forecasting. Use it to develop procedures for customer classification, for product recommendations or even for the early detection of serial defects and integrate them directly into your applications via Hana.

However, for performance reasons, it is recommended that you only work with data within your Hana environment and not with data that is integrated via cloud interfaces.

In the cloud environment, on the other hand, the services of the Machine Learning Foundation are the obvious choice. While these do not differ significantly from the offerings of other cloud providers, they offer the decisive advantage that Tensorflow models can be used directly within the SAP environment for business applications.

With Tensorflow, Data Scientists train highly specific models that specifically address the particularities of an individual process, a company or an industry. This creates the special added value in their own business applications.

This is the only way to achieve the best possible user experience in these applications. However, one drop of bitterness remains. The training of Tensorflow models is not yet supported.

AI-driven business applications are no longer dreams of the future. The missing components of a fully integrated AI process can already be implemented and operated using container concepts such as Docker. Continuous training of models and their deployment are also easily possible with this.