AI Models as a Commodity and Why Data Foundations Decide Who Wins - with Guillermo B. Vazquez of SAP

**Daniel Faggella** (0:12)
Welcome everyone to the Emerge AI in Business Podcast. Today's guest is Guillermo Vazquez, Chief Architect in the Business Transformation Services for SAP America. Guillermo joins Emerge's Nick Gersch on today's episode to break down how enterprises can strengthen the core foundations required for AI-enabled ERP. He explains why harmonized data, standardized processes, and a focus on long-term maintainability are essential to ensuring future AI-driven adaptation happens smoothly rather than disruptively. Just a quick note for our audience that the views expressed by Guillermo Vazquez on today's program do not reflect that of SAP or its leadership. According to Edison's research, 79 percent of Americans listens to Online Audio Month for AI brands trying to reach senior decision-makers. Podcasts are one of the few channels that in 20 plus minutes of focused attention from VP plus leaders. Emerge reaches one million listeners annually. See how other AI brands are driving pipeline on emerge.com/ad1. That's emerge.com/ad1.
Now for a conversation with Guillermo.

**Nick Gersch** (1:32)
Guillermo, welcome to the show. It's really lovely to have you here.

**Guillermo Vazquez** (1:37)
Thank you, Nick. Thank you very much.

**Nick Gersch** (1:40)
Guillermo, virtually every week, we're seeing the release of a new model. In fact, I think I saw the other day that SAP itself recently released its own foundation model, RPT-1, succinctly described as an LLM for spreadsheets.
And it really got me thinking. As the modalities of AI start to broaden, of course, the applications do to the possibilities broaden, and of course, additional dangers come into the picture. But many who have or are currently walking this path seem to hit the same stumbling block. And it's an issue we hear many of our guests speaking to. The point comes up over and over again is regardless of the industry, regardless of the vertical, data quality is a big issue. So how do you think about addressing fragmented data, quality, lineage, interoperability, all of these challenges across global enterprises?

**Guillermo Vazquez** (2:44)
That's a good point.
I would say, first of all, on the AI technology, our approach is that we see the AI engines, the AI models as a commodity.
I think all corporations should see them as a commodity, because once that you have your foundational, well-defined, and you have the ways that you are automating with AI, and making a GNTK AI part of your core, doesn't matter if someone comes in three days with a fantastic new model, you just plug and play in that model. And it's now in an advantage for you. It's not decremental. It's not that you are obsolete because your AI model that you're using is less effective than the new one. So you use it as a commodity. I think that's overall the principle. And coming back into your question of data, I think that's one of the fundamentals, right? If we think about you be effective in the new artificial intelligence world, I think that the fundamentals, the foundation is everything. And one of the core foundation is data. And data now, depending on global corporations, they are submitted for different sort of rules and laws, like the European data privacy laws are very strict. Asia has also other types of data requirements, North America as well. And also who sees the data and who cannot see the data, right? For example, in the US, restrictions of citizenship to be able to see some sort of data, et cetera, et cetera, right? So the point is that first, you need to invest in the harmonization of all your data, identify what I call in a global template mode, meaning what data is going to serve your global corporation, and which data will need to reside where. That's where you define your architecture and the cleanup of your data, right? But bottom end is that harmonization and globalization of data that needs to be upfront, and taking into consideration the restrictions and statutory restrictions that you have to be able to define who sees what and where does the data needs to reside. That will give you a very clean map or blueprint of how to address your data.
The concerns of, I will tell three, five years ago, we're not talking about more than that. It was about the cleansing of the data. Now we have fantastic AI tools that will support the teams to go through it. And even the owners of data that are the ones that are going to sanction and validate the quality of the data, they could do also these verifications with these new tools and technologies available today.

**Nick Gersch** (5:47)
I love the idea of kind of treating AI as a commodity. I think it goes a long way to normalizing the use of AI, goes a long way to, I guess, embedding it throughout the enterprise. You mentioned, of course, that data there becomes the foundation of everything and ensuring you have this robust framework and architecture from global and then cascading out to the remainder of the business.