If you missed it, SAP made several big announcements at SAP TechEd this month. No surprise, but Generative AI took center stage in those announcements. If you want to learn more, the SAP TechEd News Guide is a great place to start.
In this month’s ASUG Executive Exchange Digest, I want to unpack one of those announcements and share why some of its potential implications are important to technology leaders. The announcement: SAP is incorporating a Vector Engine into the SAP HANA Cloud solution in Q1 2024.
Why is this important? The short answer: Generative AI. The longer answer: Vector stores enable data to be described based on concepts in the data rather than just keywords or values. This conceptual identification lets GenAI engine algorithms make all the connections and inferences that they produce. Of course, this is a very simple explanation of the underlying technology, but it does demonstrate an important idea: adding vector capabilities to your existing business data allows it to be part of the body of data that tools like ChatGPT can base their outputs on.
Again, why is this important? The short answer: Data Security. The longer answer: LLMs can only produce outputs based on the dataset they have at their disposal. By adding vector capabilities to your business data stored in the SAP HANA Cloud, LLMs can now process your business data in your “native” data store. So, there is no need to upload your data into a separate LLM data store and potentially make your data available to anyone else who uses that engine. You can keep your critical business data “in-house.”
I do not know the commercial implications of the SAP vector engine. That is a conversation for you to have with your SAP account team. There is a big tech issue that this announcement highlights that I want to point out.
Technical Debt, Data Style
I have talked to SAP customers about data management for years. In all that time, I have only met one customer who thought they had a solid data strategy that provided meaningful governance and data consistency (they sold data management services, so no big shock that they felt they were good at it). For the rest of us, all the Generative AI/vector engine noise creates another area of technical debt that must be addressed. There is probably a formal term for this, but for now, I’m going to call it “contextual debt.”
For GenAI engines to draw meaningful inferences from your data, they must understand how your data relates to your business. What is the context of your business data? Inconsistent data management, ETL processes to populate analytics tools, and a lack of clarity of “what data is the truth” can all combine to create a world where the tools can’t determine the context of the data.
So, here’s a key takeaway. Just figuring out all the technical, security, and use challenges for AI solutions is not enough. You also must (finally) address the contextual debt challenges you have around your existing data. You’ve heard the term “clean core” thrown around in conversations about adopting SAP S/4HANA. Now, AI adoption is driving the need for a “clean data core” as well.
The Role SAP Datasphere Might Play
The (potential) good news is that there may be a solution out there. This past spring, SAP announced Datasphere. One of the big promises SAP Datasphere is supposed to deliver is maintaining the context around your business data across disparate systems. I’ll admit that at the time, I did not fully appreciate the implications of that promise. As I continue to learn more about Generative AI, I am at least starting to connect some dots on the real value behind this idea.
Wow—when I first started writing this, I wouldn’t have guessed it would end with a pitch for SAP Datasphere. The message I wanted to deliver is this: As you evaluate Generative AI use cases for your business, be sure you are also asking whether or not you have the underlying data to base your answers on. Think about how you are managing your existing data so that it can effectively support your decision-making, regardless of whether or not there is a GenAI tool in the mix.
Now, whether SAP Datasphere plays a role in your company’s data strategy is something you must determine yourselves, but effective data management and data governance are becoming a “must have” instead of the “nice to have” where many of us find ourselves today. So, here’s this month’s call to action: What works for you in your current data strategy? What doesn’t? Let me know where you are in your data journey. Let’s get your stories into the ASUG Executive Exchange community and learn and improve together.