AI Can't Fix a Broken Process

In this episode, I sit down with Nabeel Siddiqui, AI leader, product strategist, and Head of Tech and Automation at SAP Digital, to talk about one of the biggest misconceptions surrounding AI today.

Most people think AI is the solution.

Nabeel believes that's the wrong place to start.

What matters first is understanding the process, the people, and the problem you're actually trying to solve.

Because technology can accelerate a good system.

But it can't rescue a broken one.

Transcript

0:00

AI cannot solve everything that is wrong uh just because it is AI right. AI only works if the best practices in the underlying process are in line and they're working, AI is just to help you to last money. It cannot fix your property. There's no replacement of a process in place whether it is in healthcare or in technology or you have to rely on it. Welcome to the I Am Life Podcast, where the biggest questions in healthcare and wellness meet the collective wisdom of industry leaders and innovators. I'm Dr. Laura Hurry, and here we explore the intersections of AI, care delivery, and health through the lens of those who are reshaping healthcare. This is not just about theory, it's about lived experience and real industry insights. Let's dive in and get wiser together. Welcome to today's episode of the I Am Wiser Podcast with Dr. Laura Purdy, where I explore the real stories behind groundbreaking healthcare innovations and the human impact they have on patients and providers. On that note, I'm delighted to welcome Nabeel Sidiki, Global Senior Director and Head of Tech and Automation at SAP Digital, a seasoned innovator in AI, product management and digital transformation. Nabeel has spent over a decade leading complex technology projects, driving AI strategy, and mentoring teams on delivering real-world impact. Today we'll dive into his insights on AI workflows, agentic mesh, and what it truly takes to be successful in AI product management, all while keeping patient-centric innovation at the forefront. Absolutely. Thank you so much, Dr. Purdy, for having me here. And to the entire team for organizing such a great podcast. I'm looking forward to sharing some information with you. The honor is certainly ours. So let's dive in. Why don't we start by you telling us a little bit about who you are and what you're up to these days, what your projects you're working on and what you're up to. So I'm Nabil and I'm in this industry for over a decade now in different uh roles. Currently, I'm leading um the development of an AI platform uh at SAP. Um I'm here to share a little bit more about my experience and actually what I'm doing in role because I cannot share much uh due to NDAs, uh, but here to share much about like more about the knowledge and experiences that I have had in general in my experience in technology. I love it. Well, let's talk about the journey. So, where did you start and how did you get from there to here? Tell us about your journey in tech. So uh it's been a fun ride and lots of learning uh on this curve. Uh started uh during my bachelor in computer engineering. Uh, and uh from there on it was all about problem solving, different sorts of problems. Technology kept on changing, and we got like new technology and tools in our hands, but uh the knack of problem solving that remained uh constant throughout my my journey. Um uh so I have been with SAP for over a decade now. So this has been the company that I've uh I've worked throughout my career, grew here very, very happy uh uh to be here as well. Uh but along the along the lines, I've done done some projects during my uh time at Duke, uh, which was related to healthcare, uh, and also um you know to during projects which I have undertaken uh just as a hobby uh as well, outside my work. So um uh so yeah, so constant, the constant and underlying and foundation is problem solving, and I've got have gotten multiple opportunities where I was able to do so. The new problem solving mechanism is AI, and that's where we all are working uh towards. But you know, have been fortunate to work in machine learning in and AI for many years now now, uh, before even it was uh famous uh for us to even talk about. Before it was cool. Yeah, before it was cool. Before it was cool. Well, I can imagine that could be challenging, right? Because you went to college several years ago, and the technologies that you live and work in today did not exist. You weren't taught about them when you were in school. They might have even been theoretical or in just their first few years of development, and now you might not even be using any technologies, or some have been completely obsolete that you might have learned about when you were in school. So, how do you make sure that you're always keeping up to date with the technology as it emerges and develops? Great question. So I believe that the framework still remains the same for processing information and solving, right? Technology evolves, but software engineering practices or how you solve it uh is constant and underlying, right? For example, uh, let's take the example of AI, for example. AI is new, but how we process information has been uh very constant throughout the years. You ingest information, there are ingestion mechanisms on how you ingest uh information from different sources, could be via an APIs, could be something that you tap into as a database, or you know, there are modes of that. Then you process information because uh just any unstructured information is not useful for any technology, right? Be it AI or any uh even if you're creating a website, you cannot just put out information without processing it. Uh and then there is a delivery layer uh where uh actually you make some sense of the information, right? Okay, if X leads to Y, Y leads to Z, so what X leads to? So something like that. Uh in uh so that has been the case always throughout uh my experience as well. Uh if it is AI, yes, we got more tools to process information. If it was machine learning, yes, we had to build tools to process information. There were no uh generative models uh at our disposal. We had to create models from scratch. Uh but the these three frameworks, these three pillars, ingestion, processing, and delivery, uh remained constant throughout the experience, be it any technology. Ingestion, processing, and delivery. Yes. Those steps stay the same, but it's the way in which and probably also the speed with which we do them. That I mean, that's what I've noticed is that it gets a lot faster, right? With time. Absolutely. So earlier, for example, we didn't have to build we have to build models from the scratch for whatever machine learning automation we used to do. Now you have all these crazy generative AI models, which does it for you. As long as you present them in a nice format, uh, you pre-process it the right way, we call it RAG. Uh now, uh, you know, it it gives you the right results. So you don't have to build models from the scratch per se now. Wow. That's honestly, as someone who knows very little about the technical intricacies of this technology, it's very impressive to me. And I wonder, so how do you, how do you what could I say, how do you pursue that lifelong learning? Do you have to keep taking classes? Do you have to sign up for courses? Or do you learn by doing? Do you say, I'm gonna, I'm gonna practice some AI this weekend, and you jump in there and just mess around with it and see what happens? I mean, what does it look like in your workflow and in your career evolutions as someone who's in this field and adapting with it? Uh, you have to read a lot, you have to learn a lot. There are many courses, and you have to also understand and uh build it back to the foundation that you had, right? Like I was able to break it down into three pillars. Uh so you have to always bring it back to what you have learnt, otherwise, everything seems very new and uh uh and and you cannot like learn it or uh put it in your practical use. So um there's one book uh which I uh which was written in I think 1990s, uh, and the definitions that are there in that book on AI agents, on AI, it's still true uh right now. It's written by Peter Nowik. Wow uh it's Fundamentals of Machine Learning, and that book is still relevant right now. Uh just the marketing and the wrapping around it has changed. Uh, all of these blocks that we have, the podcasts like this one or any other podcast also helps us to keep abreast about the changing shift in technology. And then we have these um, you know, these research papers and these industry papers from Mackenzie Gartner, which tell us a little bit about technology curve, which way we are heading and how uh people are presenting technology rather right now. Wow. Is that so is there a certification? Do you have to get a certification or a diploma or a credential? Like how do you, as a professional in the tech industry, become qualified or certified to be a technician or a practitioner of the AI software? Uh there are multiple uh certifications by these great universities. Um those those are good refreshes because they teach you a lot about how things are. Even product management is evolving uh since that over the time. There are more tools now. Uh there are more complexities now because data has got larger. So you need to uh so there are multiple certifications that you can do. Uh there are books, there are blocks, uh, and uh then you can also listen to some cool podcasts. So multiple ways to capture information. Uh, but you know, I I employ uh all of these. I read books, I uh see tech blogs, I read Mackenzie Papers and Gartner paper to uh know about what was going on in the industry as well. And then I uh uh you know I listen to podcasts from many influencers to know about what they're doing and they're how they're implementing the AI uh life cycle. And I and I learn a lot from that as well. And that's a great piece of wisdom you're saying without saying is that as professionals, we should be dedicated to lifelong learning. Yes. We do that as physicians, journals, conferences, guest speakers, things like that. Research, some people do research. We must be committed to lifelong learning, or we will become stagnant and obsolete and no longer relevant in our field. 100%. Wow. That's fascinating. So all of the different places that people can get information from, I think you have an advantage because you are in the field and you're probably a better connoisseur of the sheer volume. I mean, there is massive amounts of information out there. And for someone like me who doesn't come from the tech industry, how do you recommend that we be good consumers? How do we vet the magnitude of information out there about AI? What do you how what do you look for in something that is a credible or legitimate resource? Uh you're absolutely right. Like information is everywhere right now as well, right? So you can just go and ask perplexity or any of these AI models, and that's also one way of uh getting information right now. So when I uh what I can trust is something that that I know. For example, if there's a paper or an article published by BCG or Mackenzie or Gartner, that they have done their research. And if they are putting out their research on AI and they're talking about whether AI is working right now for certain use cases, uh, I would take their word for granted because they have done their due diligence and there's a huge trust in equity uh behind their name. So to start with, I would say that uh uh you can go ahead with these uh uh marketing consultancies and consultancies which uh they publish these papers, these great articles, and they give a general idea of where the trend is, what are the use cases that people are working on, and then we can uh deep dive or derived into those use cases from other uh sources. So you're talking about real research with sound, solid data that is data-driven, data backed, and really scientifically or even empirically based, as opposed to other people's opinions. 100%. Yes. We have a lot of like research papers from IEEE as well on uh research going on AI. Uh everything that is published and can be trusted, that's my first go-to uh way of acquiring knowledge uh before I look uh just everywhere else. Yeah. I think that's great. And we see that a lot in healthcare too, where there may be um editorials or you could say articles that come out that are really just opinion-based and not necessarily fact or evidence-based. And it can be hard once that stuff kind of gets out. People read it and they take it for truth. But if you go back and look at the data, the research, what does the science say? The answer might be different. So we have to understand how to be connoisseurs of the information that's in front of us. Absolutely. Yes. Do you provide? I mean, I know you're engaged in several projects right now, but do you do any work with small businesses or tech companies? Do you advise startups? Or are you so busy in your day job that there's no time for things like that for you? Uh I do advise uh startups and um basically university students as well, if they are trying to do something uh cool where I can uh give my two cents and two cents of experience in. And you know that that has been uh my way of operating since I started. Uh uh I've also written books uh as well myself. So wherever I can share, or in these podcasts, for example, wherever I can share my experience and uh whatever I have learned, and I can also learn in the process, that's what I always do, and I interact with students a lot because they're the creators of the next gen company. And uh they're also a great way to learn about what's going on right now. Yeah. Uh and you know that that those conversations I do very, very frequently with uh students from the current universities in the US. It's important and it can also be fulfilling to be able to pass that on as you mentor the upcoming generation. So let's explore this idea a little bit. What are the top couple of, let's say, um, points of friction or maybe landmines or misperceptions that you find yourself talking about or encountering over and over again as you're working with students or small businesses? What obstacles are you facing in the course of your profession as you steward it right now? Yeah. So the major misconception that I see everywhere is uh people trust more uh put their trust more on technology than on process. Uh so for example, uh AI cannot solve everything that is wrong uh just because it is AI right now. Right? Uh see the example of healthcare. You cannot just deploy AI uh on every healthcare problem that you see, right? Uh AI only works if the best practices in the underlying process, the operational part of things, are in line and they are working. AI is just to help you the last mile uh and get get structured, help you with productivity and so on. It cannot fix your broken process, right? So that's where I see a lot of like challenges because when someone is learning about AI, uh it is uh romanticized in a way right now that it can just solve everything uh introduce more. So that's also not true, and that's something where um uh uh you know that's where the experience comes in as well, and that is with every technology curve. We saw blockchain at certain point a few years ago, where that was seen as the next next big thing, it will solve everything and all the banking systems and everything. Yeah, and it did not uh work that way, right? So uh that's where I think uh you know, first of all, fix the process. There's no replacement uh of a process in place, whether it is in healthcare or in technology, or uh if I'm I'm I'm uh writing a software for for a particular uh you know for a company, uh the process needs to be 100% thought through. Uh and you cannot you have to rely on functional experts. For example, you are there in your profession for many years. Uh, if I just assume AI could uh could could fix uh all of the pain points that you had without even talking to you or doing focus groups or pilots with you, uh it would be just like shooting for the moon without uh having the right target in place. Yeah, you're right. But you can't fix broken processes with technology. That's a very valuable point. Yeah, yeah, you you cannot, yeah. So you have to fix broken processes with your experience. Uh you have to set the right standard of uh standard operating procedures, SOPs, best practice guidelines in place, and then you deploy these agents or AI uh on top of it just to help you in running that particular process. To run the process, to run the process. You have to set the blueprint and then deploy AI on top of it. That is that is one of the major misconception that we have right now that just AI AI could be deployed anywhere. No, uh, we are also seeing a lot of research as well where it is not able to uh you know be that productive for you. And the reason behind that is that we are not thinking about if this is the right process where AI could be deployed or not. So people come to you or come to a consultant or advisor and they think, help me use AI. Help me use AI to fix all of my problems. But really, if you look at the root cause of the problems, it might not be that the technology needs upgrading. It might be that the processes need fixing so that the technology can then be used to enhance the processes. 100% exactly. Yes. Yeah, I'm living that problem right now, honestly, because we just went through a major electronic uh medical record system migration, which is all fine and dandy, but it also needs to have automations and workflows, and it has to help usher the journey through the process. And it's the process that's broken right now. We have amazing technology, but the process is broken and it's very manual. And you can have all the technology in the world, but a broken process is still a broken process. Absolutely. Yes. Well, I'm sure that one of the most annoying or maybe obnoxious or maybe most frequently heard things that you hear right now is AI is gonna replace humans. It's gonna replace humans. We hear that on every corner, and it almost always comes up in every show on AI that we do. So I'm wondering for you as a professional, how do you respond? Everybody that I talk to has a different response when somebody says that to them. I usually just say, I doubt it. But as a professional, what do you say when people say, I'm afraid to use AI, I'm afraid to go there because I don't want to put myself or my staff out of a job. So I would use a very cliche um uh comment here, right? So AI will replace humans not using AI. That's a fact. Wait a minute. Say that again? AI will replace humans not using AI. That is very I don't that might be a cliche, but not in my industry. So that's actually really funny. That's the first time I've heard that. Tell me more about that. What does that mean to you? Yeah, so if right now you're repulsive to the idea of not using AI, uh, that is also not good because then you're not as productive as your colleague or anyone else in your in your organization. Because everyone else is using AI. And if you're saying that you're not gonna use AI, that's not working in your favor because the work that you could do, some of the work that you could do in, let's say, 10 minutes, is uh elongated to 60 minutes because you're not using AI. For example, writing your emails or uh having a summary of all the research papers uh at one place. So yeah, so that that is that would be counter counterproductive. So you will be replaced then by someone who's using that AI and doing your work fast. Uh so that's that's my way of looking at it. Yeah, but definitely I don't believe that AI can replace humans fully. Uh they can they they can fully replace or they can supplement some of the manual tasks rather that we are doing right now, and that's about it, but it cannot be a total replacement of humans. So, example, you are in the medical field. Uh I've not seen any AI right now that that that can replace a doctor, for example. It comes with years of experience, and that's a very uh unique experience as well for a doctor, and there's no mechanism to train a model. You can train the model on Web MDs and all of these places, but you cannot train the model on anyone's personal experience. So uh I cannot create uh an AI model for Dr. Purdy because you know your your many years of experience cannot be uh translated to a particular model. So uh yeah, so that's what I I think in, and and that is true for every other industry as well, that yes, it is it is able to help you uh with some uh your day-to-day manual tasks, but definitely it's not gonna replace a human being who's efficient and working on a particular, you know, working on something on his own. I think it's fun. I think it's wonderful to live in an era of transformation. I remember when we went from paper to computer. I remember when we went from landlines to mobile phones. I mean, I remember all of those transitions, even like, you know, cable to streaming. There's so many technological transitions that we've lived through. And I think it's really I think it's fun to be a part of the progress, but I myself do not fear being replaced. I look forward to be able to use them. And it is definitely important for scalability when you talk about small businesses or startups. We have got to be using these workflow automations and AIs, otherwise, we will get outpaced by our competition quickly. Yeah, exactly. Wow. How fun. So tell me a little bit about your job. And I know you can't say much because you're under NDAs. But when you are working on these projects, do you find them to be particularly challenging for a technical, from a technical perspective? Like, is it getting easier? Is it getting harder? Do you find that you're always having to solve more and more complex problems? What's it like being someone who works in this evolving technology in real time as it's evolving? Uh I would say uh there's always been a lot of problems to solve, right? No matter what time we are in. Uh, I was at a time working on IoT, uh, where you're solving the uh problems of um you know managing the data of windmills so that uh uh any problem on a wind farm or a windmill that could be automatically translated on a dashboard. So that was also a huge problem to solve at one point when we were going through industry 4.0 and IoT and everything else. Um we have solved problems on uh uh I I I during my time at Duke, I worked on this very uh amazing project uh by Dr. Nimi Ramanujam on Kaleoscope. So a Duke um Diversity and uh Women Health program, they built a device known as Kaloscope, which could be used to determine early signs of cervical cancer. And they were using AI basically to do uh anticipher imaging and what goes behind that. So that was also a very good problem um as well. I was helping them with uh, as part of my business school, helping them with value propositioning and putting it in the right terms in front of uh uh in front of to raise uh uh funds and uh to bring in investors to back that particular project. Uh the problem that you know a corporate uh uh where I'm employed or or others are employed as well, we solve is we have a lot of challenges from the customers that or our internal teams where we help them with some productive solutions or use cases uh so that their job could be easier. And for that, also we have to deep dive and delve into their role and function. We have to step into their shoes to understand that how our customers uh target a particular uh what the what the process is again, and um what the problems are there in the process, and how can we help by deploying a solution or a technology? So, in a sense, uh bottom line that there has been always some of the other sort of problems uh that I have encountered and I've learned from. And uh I think the secret sauce there is that uh understanding from the shoes or the point of view of the other person, uh, and not just employing or pushing technology uh at the problem where it is not needed. So that's the uh that's what I have learned in my career so far. Uh, and uh, you know, it it helped a great deal because then it gives a sense of to the other person that this is he is uh our understanding my process, my life cycle, my day in life as I like to call it. Uh, and uh he's able to point out the problems which could be solved and also eliminate the problems that cannot be solved. Well, there's a piece of wisdom. So to reiterate what you just said, not just pushing technology solutions on people and saying, this is what you need, this is what you need, this is what I do, this is all I know, this is what we do, but to take a moment to stop and to ethically consider the position and the situation and the circumstances of the person of the other side of that conversation. And then you can properly advise them on what might be the right technology solution for them, not just the one that you do for everybody. Absolutely. It has to be personalized, it has to be hyper-personalized to their problems, to their situation, than just me giving them a generic solution because that never works. Everyone's problem, everyone's process, it's very unique to them. Is that happening a lot in the consultancy or tech industry? Like is that a is that a fairly common practice that you see is people who just promote one solution and don't take a broader view of what might be helpful? Uh yes, yes. Uh I see it a lot and I see it also not working because then the adoption is not there, right? If I if I create something which is for which I assume is for like 20, 30 people or 10 to 30 customers, uh, but it's really not because it is only fixing the problems of two or three. The other 17 or 20, 18 will not adopt my solution. So that's why we all complain that, hey, I've created this amazing software, but people are not adopting it because you have not run pilots with them. Yeah. Uh you have not done small-scale pilots, you have not revised it to their liking, and you're just uh one one sh one uh size doesn't fit all in this case. Wow. So it's important for people to be a good connoisseur of a vendor and vet them and get to know them before they choose them to help them implement some technology because otherwise we could make some big mistakes that could maybe even be expensive or time consuming. Yeah, yeah. Fail early, fail fast. That's the also one of the agile methodologies. So if you fail late in the process, you have already created the software that is not being adopted by 17 of your customers. So that's a big polyum. Uh then trying to fail early and also just doing it for five or six customers, for example. That's a great piece of wisdom. And for the founders, you know, the startup founders, the small business founders who might be listening to this and as clueless about AI as I am, I would encourage you to all be good connoisseurs of the vendors that are in front of you. Just because someone says, yes, this is what you need, this is what you need, without taking the time to listen to the problems that you're having or check to make sure. I love what you said about pilot and testing. How do you know if you don't test? Because technology is not a one-size-fits-all solution. So it sounds like you approach your profession and your field from a place of ethics, from an ethical standpoint, and strive to do good, not just to close deals. Absolutely. Yes, that's because it also builds your equity or your company's equity if you if you operate like that. If you lose trust to answer you lose it in buckets, it's very uh difficult to get it back. Uh so yes, uh, you have to be always true and ethical of what's possible and what's not possible, and uh just don't um uh try to relay a solution uh that's not actually for them, right? That's that will not solve their problem. It is also true for internal uh solutions as well that we have internally within the company, they we treat them similar um as our customers as well, as internal customers, and I cannot just like give them a solution which is not solving, which is just wasting their time. So pilots are very important, and constructing pilot is also very important because you have to talk to them about the KPIs that they have uh which are important to them. Maybe productivity is not the KPI that they are looking for. Maybe they're looking for some other KPI, maybe they're looking for uh increase in their revenue uh uh because of something else, uh not just productivity. So you have to understand their KPIs as well, then construct your pilot uh so that it can measure those eight to ten or five or six KPIs and then publish the results with them. That this is what we have found out with with our test or test run. And yes, uh, this is what we're gonna do projection-wise, if you go on a full-scale uh adoption or implementation of this, that's that says a lot about you know trust and building it with example, then just like you know, uh showing them a solution on a demo and asking or assuming them to just use it full-fledged tomorrow. Pushing it, pushing a solution. Yeah. Yeah. Well, on behalf of all clients of your services everywhere, thank you for approaching this from an ethical perspective because I understand that not everyone engages into such a partnership mentality when they're working as a consultant or advisor. And when you are an expert in a field that someone else is not, you're almost at a position of advantage, right? And the other person is at a relative position of disadvantage because they can be exploited in their ignorance. And so I appreciate the fact that you approach it from such an ethical perspective. Thank you. Well, you know, I think that's about all of our time for today. I really appreciate you taking the time to educate me and to share some of the wisdom that you've found in your journey. Now, can you please share with me if anybody wants to connect with you after listening to this episode, where can they find you? Uh so they can find me on my LinkedIn uh as well. Um I'm very uh I I talk to a lot of people on my LinkedIn through chats. That's one of my uh networks. Uh they can also reach out to me on my email um as well. Uh, they can write to me on nabil1591 at gmail.com, and I'll be very happy to connect and uh share or learn uh some something uh from from our viewers. Thank you so much. It's been a pleasure to have you today. I definitely feel like I'm a little bit wiser. I wish you all the best, and I really thank you for coming today. Thank you so much, Dr. Puddy, and thank you to your whole team for this amazing podcast. Take care. Thank you for tuning in to the I Am Wiser Podcast. Each episode brings us closer to a wiser, more human approach. If today's conversation inspired you or sparked new ideas, share it with someone who's ready to read the first. And if you have a story or innovation that can light the way for others to reach out, we'd love to hear from you. This face is yours too. Don't forget to follow, rate, and review us on your favorite platform. Until next time, stay curious, stay courageous, and stay lighted.