Dive into what’s next for enterprise AI in the latest episode of Tech-Driven Business. Mustansir Saifuddin welcomes SAP expert Andrea Haupfear for an in-depth conversation on how SAP is helping organizations turn AI from hype into measurable business value. If you’re navigating transformation across finance, supply chain, or operations, this episode is a must-listen. Andrea breaks down what makes enterprise AI different—why trusted data, business context, and governance are non-negotiable—and how SAP is embedding AI directly into business processes to improve speed, accuracy, and ROI. Tune in for a real-world example and practical guidance on how organizations can start small, prioritize high-impact use cases, and prepare for what’s coming next with Agentic AI.

Andrea Haupfear is a Business Process Architect with over a decade of experience driving digital transformation through artificial intelligence and advanced analytics. She specializes in designing and implementing AI-powered solutions that enhance operational efficiency, decision-making, and adaptability across diverse business environments. Andrea is recognized for her strategic leadership in translating complex technologies into scalable, real-world applications, making her a trusted advisor in navigating change and unlocking value through innovation.

Connect with Us:

Innovative Solution Partners

X: 

YouTube

or learn more about our sponsor Innovative Solution Partners to schedule a free consultation. 

Episode Transcript

[00:00:00] Mustansir Saifuddin: Welcome to Tech Driven Business, brought to you by Innovative Solution Partners. I'm honored to have Andrea Haupfear of SAP. Join me today to break down how SAP is helping organizations leverage AI to drive efficiency, reduce risk and deliver measurable ROI. We'll also look ahead at what's next and how you and your team can prepare as AI moves from experimental to essential for enterprises to thrive.

[00:00:35] Hello Andrea. How are you?

[00:00:37] Andrea Haupfear: I'm good. How are you?

[00:00:40] Mustansir Saifuddin: Doing well, doing well. I'm so excited to have you on our show. So thank you for coming on. Today we would like to talk about the latest SAP's AI journey and the business transformation. And what it really means for SAP customers. How does it sound?

[00:00:57] Andrea Haupfear: Sure. No, it sounds great. This is one of my passions that I love to talk about, and so you know, happy and excited to actually share a little bit about what we're doing with AI at SAP and what we've seen in the field with our customers. So, super excited and it's a pleasure being here.

[00:01:14] So thank you again for inviting me here.

[00:01:17] Mustansir Saifuddin: Awesome. Awesome. Let's get into it, we know we are at an inflection point, right? AI is moving so fast and it's actually turning from experimental to essential, right? For in a lot of different cases. So let's focus in how is SAP's AI strategy fundamentally different from consumer AI? And why does it really matter for enterprises?

[00:01:39] Andrea Haupfear: Yeah, absolutely. So a couple of things that I wanna kind of touch on here. So oftentimes, and you mentioned this, right, we, we've used, we've used AI in our personal and daily lives for, you know, the last decade plus, right? I mean, when you think about AI, a lot of people think about Siri or Alexa or ChatGPT, right?

[00:02:01] And you know, when I personally think about AI you've got your broad and creative tasks. What we've done in our personal lives, you know, everything from creating a grocery list to, editing a, a photo, right? A family photo. But from a business perspective you know, an enterprise AI, it really has to change those business outcomes.

[00:02:26] And really when you think about this, think about, you know, everything from could be closing the books faster or. A faster on time delivery rate or reducing risk in my supply chain. And how do we ultimately do it with the highest level of governance, auditability, cost control. And so SAP's approach is really built around that flywheel of applications, data and that additional layer of artificial intelligence on top of it.

[00:02:58] So there's, there's that aspect to it, but then also thinking about it in a couple of other ways of how we're doing this. Is, you know, yes, we're embedding it where your business or where the work happens. So making it easier for our end users to be able to leverage artificial intelligent capabilities and even machine learning capabilities, not just

[00:03:23] from a digital assistant or a chat perspective, but how do we integrate it and infuse it within their specific business, day-to-day business processes to make their lives that much easier? But then also thinking about it from a strategic perspective, how can I obtain that high level of ROI by leveraging artificial intelligence.

[00:03:47] So we're seeing it in a couple of different flavors from our customers. And then also what we're developing from a, from a product perspective as well. So we're thinking about it from a couple of different angles. [00:04:00] Additionally, thinking about it from a AI operating system for our developers and for our consultants, leveraging the AI foundation on the business technology platform.

[00:04:13] So think about what used to be our developers would have to generate thousands upon thousands of lines of code. Now that's no longer the case, right? It can take them you know, a, a minute or so now to develop the, these applications and these lines of code to where it's, it's easier for them to go about their day-to-day jobs and their, their tasks where now they don't have to spend days upon days trying to develop these different applications and agents.

[00:04:45] It also lets you and I just mentioned around agents, it also lets you create and govern custom agents to read and write back to SAP and non SAP systems. Thinking about this automation and accountability, not just getting those pointed answers, right? And then last but not least, kind of how I think about this is, yes, you've got your trust.

[00:05:09] Think about trust not only in the data from looking at it in your SAP systems, but also think about non SAP systems. Think about your third party applications that you're going in and looking at the data, whether it's geographic information customer sentiment information, or it could even be, asset related sensor information, right? So bringing in that data as well as looking at it from your SAP business data context but then also looking at it with responsibility. So SAP has that responsible AI program in place really aligned to the UN UNESCO principles and the ISO 4 2 0 0 1 certification to really prioritize

[00:05:57] those ethics and compliance and human oversight within our, our AI applications such as, you know, AI core and our digital assistant juul. So really taking it from a full spectrum approach and looking at it at the holistic level and how we can bring in artificial intelligence within these, these areas.

[00:06:20] Mustansir Saifuddin: I love the way you kind of package it all together, from an overall perspective, especially, you know, the two things that really stuck out to me was. Again, coming from a business side, what is in it for a customer, right? What is the real value?

[00:06:35] And you touched upon two things, embedding it. So a person who is currently doing a job and they're used to doing it manually. Now you can. Embed these ai component to their daily work streams, right? And how they can, you know, utilize that. And the second part I really loved it is you talked about ROI really what is the return I'm getting on this investment, right? And then lastly, you talked about data. So let's, let's talk about that. You know, here's the uncomfortable truth. AI is only as good as the data it learns from. We all know that. We all talk about it. And we have always heard this term garbage and garbage out, but what that sort really mean when we are talking SAP AI, making recommendations to customers and we are talking the effect in terms of millions in revenues or supply chain decisions. How would you like to address that? What is, what is SAP's approach on that?

[00:07:33] Andrea Haupfear: Yeah, so a couple of things and you know, I've heard that term many times and coming and being an ex consultant. It's, it's definitely right. You're only as good as, as I've heard, you know, as the data that you have. So my thought on this is really when you have an AI agent that recommends, say for example, expediting a shipment or reclassifying a receivable, the truth that it relies on, [00:08:00] is your master data.

[00:08:01] It is your transactional history and any sort of process constraints that lie in between. So ultimately, when I think about this, it's not just your, your master data. It's a, it's a multitude of things that ultimately will help the AI model in the end. In the end game to take those three pillars and turn garbage in into good decisions out.

[00:08:27] The other piece that, how I think about this are semantics and not necessarily just schemas. So think about when we have some of our solutions such as Business Data Cloud, which carries those semantics and lineage into your AI workloads. So say for example, the customer, the plants, or an open po, it means the same thing everywhere.

[00:08:51] And that's really critical for explainability and audit purposes. That's another way how I think about this. And then also just looking at this from a context perspective. So I also think about, you have to train a model when at first go, right? And be able to provide it some context, some instruction, some understanding that says, this is where this particular business process lies.

[00:09:17] This is how the process should look and feel. What does good look like? And that's ultimately what we explain and tell our customers is we need to train our model to understand what does good look like, and this is where you have that context, rich retrieval and not just kind of that blind prompting that just says, go do this.

[00:09:36] But the model will try to establish what it thinks good looks like, which may not necessarily mean what you think good looks like. This is where SAP HANA Cloud will bring in that vector engine, so those semantic retrievals and documents and notes and images.

[00:09:53] Think about a knowledge graph. So it brings in those specific facts and relationships from your ERP, but then also thinking about a rag model as well, reducing those hallucinations and making those citations explainable. So why did the model or an agent be able to go through the process that it did?

[00:10:16] Well, that's because there's multiple steps and instructions that the model has to take in order to provide an accurate response. Those are some of the things in, in which we leverage today with our customers and really making it so that way it's, yes, they may not always have the best data, but let's provide additional context to really help, again, make those good decisions coming out.

[00:10:41] Mustansir Saifuddin: I liked the way you tied it together, right? We talk about business semantics being so important, BDC, the business data cloud. How is that coming into play in this conversation? And then coming from business semantics into a context. A context really is required for the answers to make sense and be business relevant.

[00:11:03] So I really love the way you kind of connected together. Let's zoom in. Let's pick an industry. And there are so many examples that manufacturing, retail, financial services. Can you walk us through one compelling use case where SAP AI is really creating these breakthrough value?

[00:11:21] What was the business problem? And how did AI solve it differently?

[00:11:26] Andrea Haupfear: Yeah, absolutely. Mentioned at the beginning, SAP is investing heavily within artificial intelligence and machine learning capabilities, not just from an embedded AI perspective, but also think about it from a tailored AI perspective. So I mentioned

[00:11:41] Business Data Cloud, being able to pull information and data not only from your internal SAP systems, but also external and third party information. And I wanna give you an example in a use case real world use case of a dairy co-op out of Wisconsin that is actually doing this today [00:12:00] from a very innovative approach.

[00:12:02] Their challenge was around their performance at the subcontracting level. Ultimately these guys have a dairy co-op with their local farmers or farm base. They bring in the milk to not only from an internal manufacturing perspective to process out milk cheese, butter whey, et cetera cream, but they also subcontract it out as well.

[00:12:28] And so this is really where they wanted to be able to get a better understanding, not just insights perspective on their data at their subcontractors from a yield output perspective. How much dairy, how much cheese was being was an output or yield, but what was going in and then going in versus going out.

[00:12:50] And so ultimately what they wanted to be able to do was, yes, be able to look at the yield perspective from an insights, but they wanted to be able to leverage and infuse artificial intelligence from this process to ultimately help with their reduce of shrink. And contributing to a 1% KPI, which ultimately makes up to you and I roughly 10 to $15 million.

[00:13:16] Okay. So just to kind of put it in perspective here of how much we're talking about. And so what they did was, yes, we have the insights from an analytics perspective, they wanted to be able to make it easier for their dairy supply chain planners to be able to, in real time through natural language processing, be able to chat with a digital assistant to gain insights around, Hey, what is my yield output for specific plant?

[00:13:42] Tell me my highest and lowest plants that had the yield output. Tell me un understanding from a scrapping perspective how much waste is going out. So they wanted to be able to look at these specific metrics and be able to get a better understanding, hey, which particular plant is performing the best versus the worst.

[00:14:03] So that way they can help to be able to retain and possibly improve some of these plant relationships going forward. Additionally as kind of that part two, what they wanted to be able to do is they receive manual yield output reports on a weekly basis from these subcontractors. It's typically in a PDF format, Excel, PDF format, and oftentimes these can be miskeyed into their S4 system.

[00:14:35] And what they wanted to be able to do is be able to have an a little bit more of an automated process. Of, yes, not only uploading these reports that they, they have to manually key in today, but they wanted to be able to provide some intelligence behind it. And so this is where we've put in outlier detection on these attachments to where now I can see, okay, was there a miskey or an oversight that says, okay, you know, this should have been 2,622 versus 6,222. It can detect those mis keys in real time to say, Hey. For my supply chain planner, this doesn't necessarily look right. It's way outta whack compared to what was previously entered in, in the previous weeks.

[00:15:24] You should triple check this, right? So it's, it's being able to provide a little bit more of, think of like big brother watching over you before it actually goes and hits into their ERP system. This is ultimately contributing to their supply chain process and has a direct impact on their KPI metrics that they're leveraging.

[00:15:44] In this case, it's it's within shrink, so really getting a, a better handle on that.

[00:15:50] Mustansir Saifuddin: No, I think the, the way you explained it, it is a great example 'cause now I can see not only does it apply to this particular industry, but it can cut across multiple industries. [00:16:00] Right. Because the example talked about production at a plant level. At the same time, the supply chain mishaps that can happen.

[00:16:08] And usually a human eye can take so much

[00:16:11] or can detect so much, but you can't try and put it together in a way that you guide your, your workforce to look at anomalies that can really help you steer the ship in the right direction quickly and efficiently. So that's great.

[00:16:26] That really leads into my, my next question is, all of this is great, right? This is happening right now, we can see like the example you use,

[00:16:35] right? It is in action. It is in motion, and customers are seeing value. Let's fast forward, where is SAP's AI development heading? You know, let's take a time horizon, 18 to 24 months. What capabilities should organizations be preparing for? Because it is all about future proofing ourselves, right? And how should they architect the solutions today to be ready for that feature, you know, coming up so quickly?

[00:17:02] Andrea Haupfear: Yeah, absolutely. So a couple things that that come to mind. So number one and we've all been hearing kind of the next and elitist buzzword is around agents and agentic AI. So really how. I think about agents is how do we provide some of those tasks that, you know, may not necessarily whether they're they're still important, but, you know, maybe take up a lot of our time

[00:17:29] but being able to provide and have a AI model behind that. To really free up some of the workload and provide some of our end users more on the strategic front. Freeing up some of that time. So I think in my humbled opinion, Agentic AI but Agentic AI at scale. So a lot of our customers are looking at what we have and this is where we're embedding.

[00:17:54] Within our SAP applications agents within each one of our lines of business, but also custom agents. So this is something that is going to be released here in the next next several months, is looking at I have my embedded agents, but if I have very specific and unique, maybe differentiated business processes, how can I be able to integrate and infuse custom agents or an agent within this particular process?

[00:18:23] And so this is really where I think is gonna we're really gonna see a lot of value coming in from our customers that says, yes, I can use agents in a multitude of different ways. Second is thinking about as an organization, we're becoming more dynamic and, and open source for data and how we can process it in a business context.

[00:18:47] So thinking about, yes, I mentioned the Business Data Cloud, but also you know, strong partnerships. That was just announced with Snowflake as well. Right? So bringing in, yes, not only our internal data, but also our external data as well. How can we take that data and be able to normalize it? As far as from an architect's perspective, here are a couple points that I was kind of thinking about in my mind as we were going through this.

[00:19:13] Was around keeping the core clean, right? So making sure that, yes, we're using our, our business technology platform various extensions and agents and skills from from JUUL studio and avoiding really those those drastic upgrades. And also kind of how I'm thinking about this is adopting that data product mindset.

[00:19:40] So looking at, and I mentioned this as well, like Business Data Cloud, from looking at semantics and lineage, but also looking at retrieval methods. So vector engines and knowledge graphs. But then also thinking about it from a process perspective and [00:20:00] designing agent guardrails. So making sure that you have a much more standardization of an understanding of those roles and permissions.

[00:20:10] Understanding human in the loop checkpoints at what point should be automated versus, okay, we need to have a set of eyes on this to actually be able to say, yep, this looks correct. I think that that's extremely important.

[00:20:25] Mustansir Saifuddin: Yeah, for sure. And I think a couple of things that really stuck out for me. One is very near and dear to me, is the data part. And you talked about the partnership with Snowflake coming out recently, and I think it's important, especially when we talk about data for an organization. It's not just SAP data, it's like the overall, right?

[00:20:42] You know, this, what does it really make up my organization? So

[00:20:45] great approach from SAP, how it's trying to bring in like a business context around it, right? You have information within your ERP, outside of your ERP and then using the BTP platform. and the BDC platform to kind of bring it all together. So I think great segue, especially when we talk about agents and you know, we've all been talking about agents you know, for quite some time now, but now we can see the real value, how we can customize it and bring it together from a data perspective. So great conversation. On a personal note, how are you staying up, you know, on top of all these changes taking place in technology and business? What is your secret sauce?

[00:21:26] Andrea Haupfear: Yeah, no, it's tough 'cause it's changing daily, weekly, right. And so being able to stay, have it stay top of mind. This is something that is part of, yes, not only my passion and what I do day in and day out, right, but also looking and getting, keeping educated not only from a process perspective by virtue of, you know, our internal processes, what we have in our products and our product offering, but also external with our clients.

[00:21:54] To say, okay, what are they doing today with their processes? And then how can we leverage AI within that? So yes, not only from an internal knowledge sharing perspective, from a functional and technical perspective, but also external as well. And then thinking about external blogs, news sources, those are just kind of some of the things that I try to stay up to date.

[00:22:16] You know, as best as I can.

[00:22:18] Mustansir Saifuddin: No, I hear you on that. It is, it is a constant learning and I think that's the key, right? Educating

[00:22:23] and educating and educating and be able to find your sources. I think that's the key. Great conversation. I know we are at time, what would you take out of this conversation that we just had and want to leave a particular takeaway for our listeners and folks who are interested in this topic?

[00:22:42] Andrea Haupfear: Sure. Absolutely. So AI is not just like, and, and I meant we mentioned this earlier on, right? AI is not just in our personal lives, but it's also in our workplace and it's very, very real. We are seeing our companies and our customers take advantage of infusing AI into their business processes and receiving the high ROI in their processes.

[00:23:05] The key is to start small. Strategically and identify which areas will have AI and will have that high ROI, but then also have the highest value when it comes from an impact perspective. We see this to where we run this as, as ultimately an ideation sessions with our customers and from the takeaway out of those sessions is they can start to craft an internal roadmap that will lead them to AI success.

[00:23:38] And ultimately our organization can help them get there along the way, even whether they're just dipping their toe in the AI pool or some, some customers that we deal with already have strong partnerships with large language model providers or they're partnering with universities, things like that.

[00:23:57] The end goal in net net is that [00:24:00] our organization can help to not only help identify those high value AI use cases, but also investing in you to create those. We offer a free proof of concept in as much as eight weeks. You mentioned this, you know how frequently this is changing.

[00:24:18] AI is changing. Right? And this is where we can develop these proof of concepts to where you, our customers can be able to realize those value in a very quick and short amount of time. And so that's ultimately where I wanna leave the audience with is that we're, we're seeing AI not just in our personal lives, but also in the workplace.

[00:24:39] And we're actually showing them and, having them realize it in real time. So you guys can, can feel free to reach out to me. I think we'll have my contact information at the end of the podcast here. But you know, happy to have further conversations with you and your organization.

[00:24:55] I wanna thank you again personally for inviting me to the podcast and to discuss a, a very, very close and passionate topic for me.

[00:25:04] Mustansir Saifuddin: It's a pleasure to have you, Andrea, and really, I think it was a great conversation. You touched upon so many different things and I think that was the purpose of this, was to kind of bring light to exactly what's going on in, you know, we talk about AI in general, but what is really happening at the inter-enterprise level

[00:25:21] and what is the real value when folks are looking at, from a business perspective. How to increase ROI in this new technology and what does really mean in terms of increasing business revenue and across the board improving efficiencies. Right? So it's all together. But thank you so much for coming on our show.

[00:25:42] Andrea Haupfear: Absolutely. Thank you for having me.

[00:25:44] Mustansir Saifuddin: Thanks for listening to Drug Driven Business, brought to you by Innovative Solution Partners. SAP is helping customers move from AI experimentation to enterprise value by embedding AI where work happens, grounding it in trusted data and business context. And ensuring governance, auditability, and control.

[00:26:10] Andrea's Key takeaway? Start small. Focus on the highest ROI use cases and build a clear roadmap because when AI is tied to real processes and real outcomes, SAP customers can unlock faster decisions, lower risk, and measurable impact. We would love to hear from you. Continue the conversation by connecting with me on LinkedIn or X.

[00:26:36] Learn more about innovative solution partners and schedule a free consultation by visiting isolutionpartners.com. Never miss a podcast by subscribing to our YouTube channel. Information is in the show notes.