Generative AI Roundtable at the SAP BTP Executive Summit
Meet the Experts
⇨ Leveraging generative AI from within business applications is the goal for many organizations.
⇨ AI and machine learning capabilities are already available inside solutions like SAP S/4HANA 2020
⇨ Security must be prioritized when using AI with business applications and data.
Generative AI has been at the forefront of conversation for most of 2023. This is as true at SAP as it is elsewhere. SAP recently announced the generative AI assistant Joule, which will be embedded in several SAP solutions. SAP also partnered with Google Cloud to bring AI to SAP Datasphere. Given the importance of generative AI, it is likely that SAP will make more announcements in this space over the next few months, particularly in the TechEd timeframe.
With AI being top of mind for many organizations, SAPinsider attended the generative AI and machine learning roundtable at the SAP BTP Executive Summit to learn more about what SAP is planning in the space. The session discussed AI in general, rather than any specific features or capabilities. SAP is already looking to leverage AI in multiple ways across SAP BTP services and offerings that are connected to the platform.
Like any organization, SAP’s internal teams looked at use cases for generative AI. One example given was to create the capacity to answer questions based on public SAP documentation around SAP BTP. Something like this could be leveraged by any organization with internal teams that wants to search documents. While there is no functionality available supporting this capacity today, these capabilities are likely to be achieved using the SAP Cloud Application Programming Model (CAP).
SAP is pursuing two different AI strategies. The first is with embedded AI, such as with Joule. These embedded capabilities will be available inside different SAP applications like SAP SuccessFactors or SAP S/4HANA. But another approach is what is coming through SAP BTP. In the SAP Datasphere announcement, SAP said it is working with Google Cloud Vertex AI to analyze automobile industry data from Catena-X and combine it with data in SAP Datasphere. These capabilities are designed to connect and leverage SAP data with AI that is not embedded within the application.
While new AI capabilities are still under development, SAP is also focused on ensuring any AI development is secure. For example, SAP is only leveraging publically available data because many large language models (LLM) will store the data that they review and then it can be accessed by anyone. However, SAP does plan to provide target architectures for whatever it releases. This is important because SAP wants to ensure that any data on which an AI is trained is used appropriately and no private data is consumed.
While the focus may be on generative AI, SAP already has AI capabilities available in SAP S/4HANA 2020/08 and later. An example of this is Intelligent Scenario Lifecycle Management (ISLM), which supports more typical machine learning scenarios using prediction, automation, augmentation, and pattern recognition. SAP also has intelligent data applications in SAP Datasphere. This includes leveraging SAP HANA Cloud data science capabilities using an existing Python or R environment to trigger calculations in SAP Datasphere or training a machine learning model on persisted and managed data in an HDI-container or open SQL schema.
SAP is also looking to use SAP Datasphere with the AI capabilities that are already available with hyperscalers. An example of this was using Google Cloud Vertex AI, but capabilities that leverage Amazon SageMaker or Azure Machine Learning are likely to be announced in the future.
What Does This Mean for SAPinsiders?
Only a small number of SAP customers are in a position where they are likely to be able to utilize capabilities like the generative AI assistant Joule simply because they do not have current RISE with SAP contracts. For the majority of SAP customers that have either not yet moved to SAP S/4HANA or are not using RISE with SAP but do want to leverage AI and machine learning capabilities, what are the alternatives?
- New AI features and capabilities will be coming via SAP BTP. Although only the partnership with Google Cloud using Catena-X data has been announced, this is just the start of AI capabilities that can be leveraged via SAP BTP. In the future, organizations are likely to be able to use SAP BTP services to connect to LLMs that they have trained, or that are available from other cloud providers, and connect that to their SAP data.
- Machine learning capabilities are already built into SAP S/4HANA 2020. SAP has been developing AI and machine learning capabilities in solutions for years. Many of these are already available to any customer running SAP S/4HANA 2020 or later. SAP continues to add to these capabilities. Do not think that AI capabilities are only available to RISE with SAP customers. Leverage what SAP is already offering and is continuing to build into existing solutions.
- Security will be a huge part of connecting SAP data to AI. When planning for any AI use, it is crucial that organizations understand the security implications. Any data that is made available to the LLM that powers OpenAI’s Chat GPT solution becomes visible to anyone else that uses the LLM. This could be extremely problematic for SAP customers. One possibility is using frozen models that will not learn from data that they interact with, such as what Google Cloud is doing with the Vertex AI integration with SAP Datasphere. Another may be to ensure that only public data is used. However this is approached, organizations must ensure that they understand the security implications of connecting any AI to data in their SAP systems.