Why the future of AI depends on process management with Carina Siemen

In this episode, Liam O’Neill speaks with Carina Siemen, Global Quality Systems Leader at Nexperia, where she leads global quality systems, process improvement and digital transformation across a highly complex international manufacturing environment.

Carina’s route into process and quality management is not a conventional one. Before moving into aviation, operational excellence and global quality leadership, she started her career in fast food, managing restaurant operations at 19 and opening a brand new restaurant as a manager at 22. It was there, long before she formally studied quality management, that she first learned what process really means under pressure.

This is a practical and wide-ranging conversation about why process is fundamentally about enabling scale, why so many organisations misunderstand quality management, and why AI is about to make process governance more important rather than less.

They discuss:

•  Why fast food taught Carina more about process management than most corporate environments, and how operating under real-world pressure shaped the way she still thinks about standardisation, execution and scale today 
•  Why standardisation, when done properly, creates freedom rather than bureaucracy, and how clear processes allow organisations to move faster without losing consistency 
•  What the business world still gets wrong about quality management, and why many organisations become “audit-ready but operationally fragile” by focusing too heavily on compliance, documentation and approvals 
•  Why process improvement only works when the people doing the work are involved in designing it, and how communication, operational credibility and stakeholder alignment determine whether change actually succeeds 
•  How Carina approaches process and quality systems by starting with how work actually flows through the business first, then mapping standards onto that reality afterwards 
•  Why AI agents still require clear rules, governance and process frameworks to operate effectively, and how process repositories are becoming a critical organisational asset in the age of AI 
•  The difference between uncontrolled AI data environments and governed process knowledge, and why organisations with clean, trusted process frameworks will have a major advantage as AI adoption accelerates 
•  How Carina’s team built an AI-powered internal process chatbot, why prompting AI is fundamentally different from using Google, and what organisations need to do if they want employees to use AI effectively rather than become frustrated by it 

This episode is particularly relevant for anyone working in process management, quality, operational excellence, transformation or digital change. Especially those trying to understand how AI, governance and process thinking are starting to converge in large organisations.

Host: Liam O’Neill, Managing Director at bpmd

Guest: Carina Siemen, Global Quality Systems Leader at Nexperia

Transcript

SPEAKER_01 0:01

Welcome to Work in Process. We are Liam O'Neill and Sam Lewis. This show is for leaders who are responsible for improving how their organization actually works. If you lead process, transformation, IT, enterprise architecture, data or operations, and you are accountable for turning strategy into execution, this podcast is for you.

SPEAKER_02 0:22

Across the organizations we work with, we see a lot of investment in teams and tools, in programs, different softwares. You get dashboards built, processes modeled, programs launched. But that doesn't always translate into real business outcomes, into bottom line results, into something you can point your finger at and say, yeah, that's worked. If you are serious about building a structured, data-led and outcome-focused approach to change, we are glad you're here.

SPEAKER_01 0:51

This is Work in Process. If you get any value from this episode, please subscribe. You will get a brand new episode before anyone else. The views and opinions of our guests are their own and do not represent those of the company.

SPEAKER_02 1:02

Welcome to Work in Process. Today I'm joined by Karina Seaman, the global quality systems leader and Experia. Karina leads global quality systems, process improvement, and digital transformation across a really complex international organization. And she's doing some really interesting work on the role of AI in quality and process management. And what I really liked in our briefing call, our little call beforehand, was that Karina's view of process didn't start in corporate. It didn't start in quality. It started in the fast food industry. So Karina, love to start there. When you look back on that, what did it really teach you about process?

SPEAKER_00 1:37

Yeah, you're absolutely right. Fast food isn't really where people expect a global quality or process leader to start. But honestly, that environment really taught me more about process than almost anything else. So right after high school, when I was 19, I started a vocational training program in a system gastronomy. Vocational training, Ausbildung, that's a very typical German model. So you split your time between working at a company, getting very hands-on real-world experience, and spending one to three days a week at a vocational school where you learn then the theory part of that. Normally this kind of program takes three years. I was able to shorten it to two and a half. So system gastronomy, fast food is all about scale. Same products, same service, same branding, no matter where you are in the world. There might be some small local tweaks here and there, but in principle, you get the same burger menu with the same service everywhere. So the restaurants all run on identical processes, and I think technically you need at least three locations to even qualify as system gastronomy. Anyway, I was really incredibly lucky with my trainer. She was amazing. Within just a few months, she gave me and two other trainees who started at the same time a lot of responsibility, and she involved us in all the key tasks. A little bit earlier than typically, I would say. After about six months or so, she had to take a planned medical leave for several months. Before she left, she split responsibilities between the three of us because we were kind of newbies still. So the three of us basically ran the restaurant together as a leadership team during that time. And at that age, 19, that was really a huge amount of responsibility and also a serious challenge. But really, at its core, and also the vocational training in general is management education. So how do you run a business? And this includes hiring people, conducting interviews, scheduling shifts based on the expected revenue, planning the inventory, placing orders at the supplier's leading employees during the shifts, actually. Also guest service, direct customer interaction at the counter, but also preparing food in the kitchen. Also, after the shift ends, late at night at 2 a.m., you close the business, you run your numbers, and then also cleaning the entire kitchen. That's also part of that. Grills, fryers, ventilation, milkshake machines, everything. Still today, I can take apart any piece of kitchen equipment in my sleep, clean it, and put it back together. Yeah, but coming back to the process part of that, so as a shift leader, every single hour in the fast food industry, you're monitoring not just the food safety standards like meat temperatures, but also sales numbers, labor costs, material usage, and then you adjust your stock, your stuffing accordingly. So I learned every single aspect of the business from the ground up. And to make that level of precision possible, no matter which restaurant you're in, as the franchise owner, they had like seven or eight restaurants, and sometimes they place the different shift leaders in the other restaurants to help out. But it doesn't matter at the end. No matter which restaurant you're in, anywhere in the world, you need the standardized processes for everything that I just mentioned. If you look at the operation side, like making a burger, every single step is documented. I mean, there are even guidelines for how long it should take someone to place two tomato slices on a burger or exactly how much sauce goes on it. That allows you then as a shift leader to objectively assess an employee's skill level and set up the right shift setting. Maybe this is really an extreme level of standardization, but this process thinking shaped me massively at the very beginning of my career. Still does today. Of course, it doesn't fit in every environment, that's also clear. After completing my training, the franchise owners then, so basically the business owners that gave me the responsibility for an own brand new restaurant, so to say, where I acted as a restaurant manager when I was 22 years old. I was involved from day one, so from construction, recruiting the first employees, the opening, and then getting the operation fully up and running. I did that for about two years and then I decided to study industrial business engineering because I wanted to better understand the theory behind all the practice that I've learned and add a little bit more technical and engineering thinking to the management part and the entrepreneurship. So coming back to your question, fast food is really a process under pressure. You're dealing with high volume, real customers, face-to-face, really raiser thin margins, zero tolerance for inconsistencies. So there is no room for theory getting divorced from reality. So if a process doesn't work, you feel it immediately in customer complaints, longer wait times, then stressed employees, or then also lost revenue. So what it taught me is that process is not about documentation, it's really about enabling scale. So same product, same service, same experience no matter who's on the shift, or no matter the location, no matter how busy it gets. That only works if processes are simple, are clear, are um executable by people with very different backgrounds and skill levels. It also taught me the power of standardization done right. So not as a bureaucracy, more in the matter of freedom. So when the basics are standardized, the people can focus their energy on execution, on customers, on handling exceptions. So standardization is not about controlling people, it's really about making the success repeatable. And quality and speed are not really opposites. So when processes are well designed, quality actually enables speed, and then the work becomes predictable, decisions are clearer, the mistakes go down, and that insight stuck with me and still shapes how I think about quality and processes today. Of course, at a different scale at this moment. So while it might look unexpected on the servers at the beginning, this environment in the fast food industry gave me really very practical, very realistic foundation for everything that came after, basically.

SPEAKER_02 7:53

I must admit I also spent quite a few of my formative years in fast food as a sandwich artist. I always remember as I went for an university interview once at Cambridge, and the first question we asked was on your CV, it says you're a sandwich artist. What does that mean? It's a very floral title for a job, but when you look back on it, absolutely I couldn't agree more. Having that really structured, standardized lean process makes such sense. Speaking of you know, keeping those processes really lean, you know, you see fast food as absolute kind of paradigm of success on having lean process clean lean processes. Is keeping things lean something you still focus on today? Is that something you still carry with you?

SPEAKER_00 8:30

Yeah, like I just described, so everything, every step, every workflow had already been thought through by the headquarters, and it has been optimized for maximum standardization and profitability. So even if there were promotions like a special burger or limited time offers, those always came with very clear instructions, step-by-step guides, briefings, the training rolled out across all locations. Today I look at this very much through a business lens. So every extra step, every deviation from the standard process, something customers don't actually pay for directly eats then into your profit. And if from an entrepreneurial perspective, I leave too much freedom or ambiguity in that kind of environment, it hurts not only profitability, but also then the customer experience. So customers they know what the burger is supposed to taste like, they know how the fries should be, and if that expectation isn't met, they are disappointed and doesn't just affect this one restaurant, it impacts the brand as a whole. So worst case, the customer simply doesn't come back. Another important factor in system gastronomy is the workforce. So many people working in the kitchen or at the counter are, let's say, untrained stuff in matter of if we look at this vocational training or education or studies, some may have fewer opportunities in other work environments or they're doing the job as a temporary solution. So that means you're naturally need more guidance and a clearer structure. At the same time, getting employees up to speed quickly is really beneficial for any company, in my opinion. So the faster the people know what to do, how the movements work, what the rules and expectations are, the better it is for the business and for the customers. And yes, as customers, we might be a bit more patient with someone new at the beginning. Sometimes they have this badge over here. But if we are honest, we still want them to get to, let's say, normal, quick operations very quickly and then deliver a solid performance. And in my view, this applies across all industries, all functions, all roles. And if I zoom out even further and think entrepreneurially, especially about startups, you see the same pattern. So at some point, every startup reaches a moment where processes become necessary to take the next big step. Early on, the founder does everything themselves. So then the first employees join, the team grows, investors see the potential. And at some point, things start to break if all the knowledge lives only in people's heads while the organization is then scaling. And that's exactly where defined processes make a lot of sense to transfer the founder's vision, align the teams, and enable faster, then more sustainable growth.

SPEAKER_02 11:13

And on that theme, then going kind of from your initial background in fast food, going into the rest of your career and going into being able to set those processes, set that kind of landscape for the way an organization works. What did you find changed as you went from fast food into kind of the later stages of your career?

SPEAKER_00 11:31

So moving from fast food into, let's say, more quality process management. And it actually wasn't until I was studying, specifically in one of my first quality management lectures, that something clicked for me. And up until then, processes, the KPI monitoring, all of that felt completely, let's say, normal to me, just standard business practice. I never really questioned it. I never thought, oh, this could be done differently, or this is something special, or this even has a name. It just felt logical. It felt like the right way to run and manage a business. I mean, when I was really young, I was also very active in the fire department, and there you do have very clearly defined procedures, routines, and protocols. Even today, I have a private pilot license and in aviation as a pilot. You work with very clear procedures, safety protocols. Every single flight you check yourself against these procedures, no matter how often you have done that. And in that case, more of course for safety and risk management reasons. But for me, I might have just a natural favor of using processes and working that way just makes sense. But back then, sitting in those quality management lectures, it was like a light bulb moment, almost like something suddenly clicked into place. And I remember thinking, yes, this is what I've been doing for the last few years. Now I know why this was set up this way. And that's when I realized this is the theory behind the practice I had learned. And in that moment everything came together for me. Yeah, the picture is then wasn't finally complete.

SPEAKER_02 13:02

And actually that's really interesting. Having that academic piece of understanding quality management, having done specific learnings and trainings around that. What did you see are some of the biggest differences between how it was taught, the theory, and how it's actually done in practice?

SPEAKER_00 13:17

So, first in the system gastronomy, it was basically process management in practice. So hands-on executing what others have designed basically. During my studies, I then ended up in the aviation industry as a working student. The role was in operation excellence. So there was very strong focus on lean management. I was actually introduced to lean methods during my studies in the US. It was really a major topic during that time over there. And for my bachelor's thesis, and again, that's a very German thing, I guess, because you need a practical project as part of your degree. I worked with a company that manufactured escalators for shopping carts, and I completely redesigned their final assembly line for the escalator manufacturing based on these lean principles. So we analyzed half of the factory floor using spaghetti diagrams, looked at waste principles, redesigning the layout, created new 5S-based workstations. Many of them we actually designed and built ourselves for welding. We did get a bit of help from the professionals. But coming back to the aviation operation excellence part. So as a working student, I then analyzed the production lines using these lean methods together with the employees. So we developed new concepts and then actually implemented them. And that's one of the most important elements, I think, of process design. And what is also, I think, uh the challenge, or not the challenge, but you need to consider this. And that might be then the difference of the theory part and during the studies. You have to design or improve the processes together with the people who will later work in them. So they experience firsthand what doesn't work, what could be improved, whether a process really works in the day-to-day reality. And I think this applies to any type of process. Whether we are talking about production, engineering, procurement, sales, maybe we'll discuss also later the AI agent-based processes. So you always have to include the people executing and working in this process and work together with them. When it comes more to process governance, that's more my role today. It still helps enormously to actually understand the process yourself, maybe work in this process also time to time. Of course, not with the same intensity as in my earlier roles, but it creates really an understanding, also credibility, and definitely drives acceptance in my point of view. Sometimes on LinkedIn you see, or I see, I don't know, airline executives stepping in and working as cabin crew for a day. So showing presence, being visible, understanding what the people in the processes do, what matters for them, and also the communication part. Yeah, keeping the bigger picture in mind, being clear about why the organization is heading to, how it all fits into the overall company strategy, aligning the stakeholders, staying close to what's happening right now. And that's something I learned over time, not during my studies in the theory part. And that's getting more and more important, really the people, the communication part of processes, so behind the scenes basically, to make this all work.

SPEAKER_02 16:18

I think that piece you mentioned about getting the people on board of it as well, and understanding what they're going to be doing, whether it's getting your hands dirty as a senior exec, getting on your airlines, understanding what works and what doesn't, or even just being careful with your wording, that's really critical. Now, you're in a quality role, and in terms of getting people on board, the word quality can sometimes be a little bit tricky. It sometimes comes with some preconceptions. What do you think the business world gets wrong about the perception of quality?

SPEAKER_00 16:48

I think you're touching in a very real and I also think a very damaging misunderstanding, and that's exactly why I'm quite careful with the word quality, or sometimes I'm trying even to avoid it altogether. So in much of the business worlds, quality management is still equated with, let's say, compliance mechanics, quality assurance instead of value creation. What I see is all behind quality processes. And I think that's the core problem. What people get wrong is I think rooted in the language we use, like quality compliance, quality management system, QMS, that haven't kept up with the, let's say, modern business language. So they still trigger associations like bureaucracy, overhead, cost centers, or we have to do this for the auditors. And what's missing is more a reframing. For example, instead of saying quality management, say process excellence, process management, or QMS is more how we reliably run the company. Or instead of compliance, I use trust, consistency, organizational resilience. So what I'm trying to do is connect quality to how value is created, not how the rules are enforced. So quality is still talked about mainly as control, not really as an enablement. And the language is then dominated by these words controls, reviews, approvals, sign-offs. Yes, control is necessary sometimes, yeah. Don't get me wrong, but when control becomes really the headline, quality feels a little bit more defensive and backward looking. So in my perspective, quality should primarily reduce the rework, increase decision confidence, make outcomes more predictable, enable teams to move faster in a safer way. In let's say mature organizations, quality is what enables speed. It's not what slows you down. Another issue, what I think, is that quality is often, let's say, delegated to the quality department. The quality department is taking care of the quality. So it's often positioned as a policing function or as a support function at the very end of the process, quality control when something has already gone wrong. So that creates also a very dangerous dynamic. The business owns the speed and the results, while the quality owns constraints and risks. And that almost automatically makes quality feel like kind of a friction. Real quality management, process management is about embedded ownership. If quality isn't owned by process leaders, by product leaders and managers, then I think it will never scale. Quality is also often still seen as documentation instead of outcomes. So many organizations treat quality like procedures, checklists, forms, audits, audit as a big event instead of asking the question, okay, does this actually help us deliver better, more predictable results for customers and for the business? And when quality turns into, let's say, paperwork, people just optimize for passing the audits, not for building really a robust process. And then the result is an organization that is audit ready, but from the operational perspective, very fragile. And from my perspective, quality is about how work actually gets done, not how well it's described, then in the documentation part. And I also see quite often that quality is treated as quite static. So many systems are designed as if once a process is documented, it's done. Change becomes then the exception and the improvement part, the speed, the changes, they live in the projects, but not in the daily work. And that mindset completely clashes with today's reality. Our environment is changing so fast at the moment. Customer expectations are changing fast. And the organizations they need to adapt really constantly. And typically they do. And what worked a couple of years ago might not longer fit to today's processes, the structures or the operating model. And the whole organization is changing. So also quality, the process management part, needs to be adaptive. It needs to be learning oriented, feedback driven, continuously evolving with the business in sync. So in practice, quality should function more like a living operating system than a very strict rule book in this sense. So, bottom line, in my view, the business world often gets quality wrong because it treats it as an administrative obligation. We have to do this, instead of a strategic capability of the organization. And organizations that, let's say, get it right, at least in my perspective, is that quality is about clarity, about ownership, learning, predictable execution, especially under change. And if it's really done well, from my perspective, people don't experience it as quality at all. So this is why I'm also trying to avoid this word. It's just the experience that work is running smoothly.

SPEAKER_02 21:36

That kind of evolution from quality as a team that said you should do this to a team that says, here's why we should do this together on a continuous, ongoing basis, making sure we are not a tick box, we're getting some things done, but really helping things tick along in a more effective way. So next period currently, you're responsible for global quality systems across a ton of different areas. What's a Is part of making the quality system or the system you oversee at an experia really help the business take along, actually deliver on that value you mentioned before?

SPEAKER_00 22:08

Yeah, so as I said, quality needs to get the work easier, not really harder. So there are a couple of things that I'm doing to make or to implement this and to get this approach integrated in the organization. That's just for example, if you design a process, I do it the way that we start, let's say, with the work first, then the standards. So lots of organizations do it in the other way. They ask first what does the standard requirement and then the design process to satisfy that. But then you end up in two different worlds, to be honest. So I encourage to start first, okay, let's see what the work actually needs to flow to deliver value, make that visible, make that explicit, and then map the standards onto that flow, not the other way around. And honestly, lots of the requirements, if you say ISO 9001, IITF, aviation standards, it doesn't matter. They aren't really rocket science. So they're usually not from good business practice because they reflect what customers actually care about. So when a process generally works, the standards tend to fit naturally. And then the documentation becomes the explanation, not the justification. Then I also encourage to make the standards invisible to most of the people. That's also a little bit maybe a different approach. So I think one mistake is also pushing standard language down to everyone: clauses, paragraphs, numbers, audit terminology that just creates cognitive load or distance, if we also talk about language. So instead, we translate the standards into the clear operational expectations. Keep the ISO, the standard language in the system, not in people's daily work. So the team should work with purpose, with outcomes, with decision rules. And typically an operator or an engineer or a manager, they should be able to say, I know what good looks like here without ever quoting a clause. And also shifting the how we think about audits, that's also some aspect from inspection to navigation. So if an audit asks where does work it unnecessarily hard, instead of where do you deviate, then the audit becomes more valuable to the business. Documentation is another big bond. It should be the side effect, not the goal. As a rule of thumb, I always say if a document is only useful during an audit, it's already failing. The same goes for approval layers. Many systems pile on sign-offs, gates, approvals, not because the risk is well managed, but because authority is unclear. So that also makes it quite cumbersome then.

SPEAKER_02 24:39

The approval one.

SPEAKER_00 24:49

So in general, I would say quality system that actually makes work easier, defines who decides what, clarifies what good decisions are based on, and minimizes the escalation by design. So fewer approvals don't mean less control. They mean control is built into the process and not applied after the fact. So that's as a general feedback for these approval layers. And typically, approval layers are a cause of findings and then the quick fix is okay, let's put another approval layer on top of that. But this is just a quick fix. It's not really addressing where the process didn't work right well. So it's just fixing it, putting a plaster on top of that instead of really making then the process more simpler. And it's not because the trigger is that the people don't understand the process, it's that the process is badly designed. So that's something I would use as an approach for making less approvals or the right approvals at the right process step. My current role and maybe what's the hardest part of making a global system feel useful to the business. So all what I've just said, and especially introducing that in an environment that already exists. So you're always dealing with a setup that was built by someone else, for example, based on a different understanding of quality management. And that's okay. Everyone brings their own perspective. People make decisions based on what they believe is the best approach at that time. And of course, things like the company's history, the overall setup, other departments, the past experiences all play a role. And at the same time, of course, I bring my own background into the organization. So I come from the lean process management, scaling businesses in the fast food industry, completed a power MBA a few years ago with a strong focus on startups. So all of that has shaped how I think and how I approach my work. And that also shapes how I work with my team, how I turn ideas into the actual execution, and how I convince my stakeholders and make changes stick in a sustainable way. So I think, like so often, it really comes down to people, communication, and change management. Like the argument we have to do this for the audit. That's something I don't like at all. I don't want to hear that argument because at least I think it's not correct. We don't support auditors or audits, we support as process managers, as quality managers, so we support the company, we support the organization. So processes need to fit to the business and not to an auditor.

SPEAKER_02 27:18

Yeah, I mean, too often you see quality teams who are going through and it is a tick box exercise. It's to prep for an audit, and the rest of the business is getting so little back from that. I love your point about having documentation just to pass an audit is not good enough. It's a waste. You have this fantastic asset, this investment of time that people have made, and you're giving them nothing back for that. So it's a really good mindset how can you create value for the rest of the organization. There's a lot of different priorities though, especially if you're running a quality team across such a big organization as Nexperia. You've got multiple different strategic programs, projects, or you can invest your time and effort in. You've got lots of different initiatives, a lot of different options, a lot of different things to look at. How do you decide which ones take your time, which ones to prioritize, which ones to focus on?

SPEAKER_00 28:04

So from the very beginning, we deliberately pulled ideas from the community. So just like with process improvement in general, the best ideas come from the people who work in the processes every single day. So we asked that they submit suggestions from their perspective, from the community, where they think it really made sense to launch an initiative or set up an improvement project. And then as a leadership team, we reviewed that input, of course, added our own experience and perspective and discussed what could create then the most value. So based on that, then we decide on uh initiatives and improvement ideas projects. So if you focus on these coming from the the ideas coming from the, let's say from the ground, from the people who actually work in the processes, matching it with the let's say bird perspective from the leadership team, I think it's then a quite well-rounded picture on then on the projects you can decide, okay, this is the focus for this year, this is where we want to improve on, for example.

SPEAKER_02 29:05

Yeah, you can't really just make that decision yourself, right? That hits all the production team, it hits the sales teams in terms of what we're distributing to the market, customer service teams in terms of managing this new product flow. You know, that has to be done in conjunction with all those diff at least for sponsors of those areas and some of those impact for the stakeholders. So uh earlier on in your career, in one of the previous calls, you mentioned that you built a web-based quality management system and kind of built that from a ground up. Given you know we've talked about your background in fast food, in lean, in uh aviation, none of those are in IT. So building this system from a growing up, what does that give you in terms of insights into the way IT and the way digital transformation works?

SPEAKER_00 29:59

So there was during my time as a working student in the aviation industry. Back then, I actually build the global web-based quality management system uh using SharePoint, so about 10 years ago. And so often happens the IT owner for the SharePoint system didn't really have the time. So he basically said to me, okay, I can either build this exactly the way you describe it, or I invest a bit of time teaching you SharePoint, and then you build the rest yourself. If you get stuck, or YouTube tutorials don't get you any further, then just reach out. So that's what I did. I sat there, taught myself but very basic programming knowledge. So it was really pure hands-on learning by doing. But to be honest, that's really what tools like SharePoint or Power BI are designed for. So it's very plug-and-play. Even with limited IT skills, you can already build surprisingly powerful applications yourself. And I see the same thing today with my own team. So behind many of our processes, there are apps running, and the process flow are very tightly connected to the apps itself. So you need to have a very good understanding of the application. Some IT knowledge is also very helpful to map the processes to the IT applications. So you need to have a certain level of basic IT understanding, how these tools work, how they can be extended, further developed. So, as a general setup, especially for the processes that I have under my responsibility, we have kind of a competence network concept behind. And there is then one competence network champion. And one of the tasks is to hold all the, let's say, the threads together, coordinating everything with some key users that have the local functions or local execution of the process. But one of the functions of these champions is to coordinate everything, but be also serve as a key contact for the IT, for the tool providers, serving as a system administration. So having this technical ownership. So I think it's very important and it gets more and more important being a process manager, process leader, that you have some basic IT understanding. It doesn't need to be very in-depth, but in understanding. So IT literacy is becoming more and more important, especially as we move further toward AI. Also the agent-based setups, that you understand the interfaces between the different applications, how the applications work, but also then later on how AI works. I think we will come to this also in a couple of minutes.

SPEAKER_02 32:25

No, definitely in kind of a process space, you're seeing that you don't need to know how the systems you're designing around and incorporating work, but you do need to know what they can do. If you're a designer processor where you're working, you've got to recognize the new capabilities can do this, that they can automate invoice matching. You can have intelligent document process and it takes documents in and passes that info very quickly. It's interesting. You mentioned the competency framework though, you have kind of sitting behind your team supporting that. What are some of the specific new competences that you're adding on to that framework that you're seeing becoming increasingly important?

SPEAKER_00 32:56

I think absolutely AI knowledge. This is really important. And this is also what I'm really encouraging that they understand how AI works and what is possible with AI. I think just a few people understand the potential of AI and also then the potential AI plays with process management and the tools we have, and at least now from quality perspective. There's lots of potential. So you need to understand how AI works and how you can apply this then to the processes, what's adding value or how the developments will be with AI agents. So this is something I think is really important building your knowledge around AI. Also if you're working process management.

SPEAKER_02 33:36

It's going to be very disruptive. If you're trying to control what people do, it's going to massively change what people can do, what systems can do. So it's got to be really fundamental to that process capability moving forwards. When did you first see a couple of use cases, a specific area where generative AI was going to be genuinely useful in experience, in quality in the space you're in now?

SPEAKER_00 33:59

I think that actually goes back a quite a few years, to be honest. So my very first real touch points with AI were during my time at uh Lufthansa Technik around 2019 or so. So they had this random lunch generator where you would get paired with colleagues for lunch completely by chance from different departments. And through that, I once met someone who had been researching and developing AI very early on. And he explained some of the projects they were working on at that time, and he really emphasized how big the potential of IA could be at that time. So it's seven years ago already. I mean, in aviation, the data collection started very early. So even back then, it was clear that this data was a huge asset, something incredibly valuable you could build on over time. One major application area is uh predictive maintenance in that area. I worked in the engine sector during that time. So aircraft, engines, they collect massive amounts of data on every single flight using hundreds of sensors. And that data is then used to predict failures or issues in individual components long before they actually occur. So that started at Lufthansa Technik as an internal startup. I think it's called it was called Zero G. And now they're celebrating the 10th anniversary. So they were quite early adopters. And after that, to be honest, I more or less parked the topic for a while. And then when ChatGBT became publicly available, I believe that was at the end of 2022, first started using it more privately, experimenting with it. And then not long after that, I invited that former colleague to the Chili Kongku webinar. That's something I'm uh hosting from the German Association of Quality. And he shared then additional perspectives on how AI could be used specifically in the context of process management and improcess improvement. And that really sparked new ideas for me. So from there I started developing my own thinking, educating myself further with AI, exploring what this could mean in practice with process management. And now everybody's talking about AI agents. So that's everywhere in the conversation. And from my perspective, that's the next big area where process management has really enormous potential.

SPEAKER_02 36:14

He said there's enormous potential playing devil's advocate. Why do you need process if you've got agents who can think and uh can do your process for you?

SPEAKER_00 36:23

So for me, okay, so agent AI or agentic AI represents a completely new form of, let's say, process intelligence. So in AI, agent is essentially a software program that can collect data, analyze it, and then independently take actions to achieve defined goals. Can range from creating audit plans, compare supplier prices, configuring data interfaces, or making decisions in the production line with data from optical inspections. But for agents to do that, they need guidance. They need to know what they are allowed to do, how they are supposed to work, and what the goals are. So what no-go's are, which decisions they are allowed to make. And all of that information already lives in process management. So whether you call it process management, QMS, document management system, DMS. So the rules, responsibilities, decision limits, workflows are already documented there. So that makes process management a massive data and knowledge treasure. I mean, it always has been there, but now it becomes even more important for the organization. Until now, the main focus was, let's say, knowledge management for employees. Going forward, it also becomes the knowledge management for the agents. And the interesting part is yeah, everything is already there. And AI can work with BPM models just as well as with what I would call, let's say, historically grown DMS setups that are still based on Word documents. From an NI perspective, the format really doesn't matter. It can work with both. Now you could say, okay, agent developers, often sitting in IT, they can just train the agents individually and feed them the necessary information manually. But then you hit the challenge keeping everything up to date and still as an organization, at least this is what I think, you want to control the results or let's say the goal, what the agent should do. If we take a simple example, an agent handles standard office supply ordering, maximum limit of $500. If that approval threshold changes now to $300, the agent needs to know that immediately so that it can act accordingly. Otherwise, it will keep making wrong decisions. In classic process management, there are typically well-established ways to implement a process change, communicate them to the relevant stakeholders and key users. But in the future, with many agents, not just one, where you can keep up with the maintenance, but if we have really lots of agents, keeping them all manually aligned would become a huge effort. So why not use the processes that already exist in a well-established process management system? And from my perspective, this is where the process management, the owners of that, they need to partner very closely with IT. So we need a shared understanding that the knowledge required for the AI agent, they already exist. It just hasn't been viewed that way before. So they can make use of that, connect them. And in the age now of agentic AI, process management becomes a new kind of, I would say, organizational asset, almost like a hidden treasure that's suddenly very incredibly valuable.

SPEAKER_02 39:27

Definitely. That centralized, clean, up-to-date process repositories, your rule book that sets the context for all of your agents, where all your rules are, all your ways of working, all the gold that is uh that drives your company. It's actually really interesting because you've seen a trend with Anthropic and OpenAI where they're both trained on public available data. Ton of that available, ton of soft content, books, publications, blogs, fantastic. What we don't have access to is that proprietary enterprise data, the process repositories, the hard performance data. And you've seen them start up these consulting branches with Anthropic, with OpenAI. And there's a revenue stream there, but there's also, more interestingly, that access to that proprietary enterprise data, the real asset, and these consultant arms being built up so that they can get that and use that to further build on what they have. If you've already got that asset, it's clean, it's accessible, it's fantastic. You can feed that, and that gives you a competitive advantage over anyone else who's building AI agents without that context. So I couldn't agree more.

SPEAKER_00 40:25

Just to add one more thought to this, lots of organizations have, let's say, co-pilot rolled out. Co-pilot from the default setting, out of the box. It's also trained AI. And of course, you can also raise your questions, but the difference is the data like behind that is somewhat uncontrolled. So it uses your own personal access to files and folders from SharePoint, from OneDrive. And honestly speaking, I don't clean up my OneDrive and I have old drafts in there, and Copilot doesn't know what's now the latest actual process, what are the latest actual rules, how we work as an organization. This is not controlled in co-pilot. It's an uncontrolled data lake and it's individual for everyone. Everyone has a different data lake in their systems. And if you have now your controlled environment, controlled AI, where you say, okay, the controlled data lake is the process management. You can be sure that the knowledge of the AI, how it should work, how it should operate, what are the no-gos, the guidelines, this is controlled and not individually for everyone. So that's the difference between, let's say, the standard, the default LMMs, and then the ones you build for your organization and where you use then the process management, the approved process management processes as the data lake for the AI.

SPEAKER_02 41:42

Yeah, and final note on Copilot. I think a lot of people I've spoken to on record, say they use it. And as soon as we're off record, they're sneaking onto COD and OpenAI, sometimes sneaker on compassing policy, which is another fun risk for quality teams.

SPEAKER_00 41:56

Exactly. Not speaking about okay, what kind of information should not be in there, especially if you use public available elements. Yeah, of course, that's also something quite risky for organizations losing knowledge if we talk about IPs, etc. That's also some a huge aspect to consider.

SPEAKER_02 42:14

So in terms of what agents are actually doing, though, what you can use them for, what you're planning for right next period, and what you've got live, we've kind of touched on people asking questions, using it to query documents. Are there any more interesting use cases that you have been playing with or have to turn around some value for Next Beria?

SPEAKER_00 42:32

Yeah, so regarding process management and let's say the knowledge behind it. So one of my first goals was actually, let's say, simple in terms of technology. Um, I wanted to make the process knowledge accessible to everyone in the company as easy as possible. So finding information has always been, let's say, hard, especially for new employees. So you often need to already know what you're looking for in order to find it, and that's a big barrier. So the idea was if I feed our approved current process knowledge into an AI, meaning the II scope is clearly limited to the validated company knowledge, then I can use it like a chatbot. So as an employee, I can just type my question into the chat using normal language, and the chatbot explains what I need to know. If I want to go a little bit deeper, it gives me a direct link to the original document. Also, we build this chatbot so that it appears directly inside the normal Microsoft Teams environment. So just as another chat window right next to the chats with my, let's say, real colleagues. From a technical perspective, that's actually not very hard to build. The other applications or the other use cases are maybe a little bit more tricky. But of course, you can't just release an I application to 13,000 employees without thinking it through. So there are rules and regulations we need to follow in this case, like EU AI Act, for example. So from the very beginning, the project team needs to be clear about these requirements and actively take them. into account. Another important point anyone who has worked with AI knows that good prompts lead to better answers. And not everyone lives deep in the AI world like I do or fully understands its possibilities and its limitations. That's okay. I mean for more than 20 years all of us have learned how to Google information. But prompting an AI is fundamentally different from doing a Google search. So I approach it in a completely different way. I go more an iterative approach. So the AI is a personal thinking partner. We work towards the answer together. So that means I need to provide context, scope, the goal, target audience constraints. And if I now just roll out a chatbot without any guidance to 13,000 people, the risk is high that people get disappointed very quickly. And then they use the Google style approach, what we have done for 20 years. They don't get the best answer, maybe they try it a few times and then they lose trust and think, ah okay, it's not really valuable. It's just a better, a little bit better Google search. So once the disappointment is there, I think people often become even more skeptical about other future AI applications. So the road for the other AI applications is even harder. So if you want to successfully roll out a first AI application in a large organization, you have to bring the people along. You need to explain the do's and don'ts what belongs in the tool as we just said something should be in there what doesn't so where are the limits and how do you prompt effectively in the simple let's say chatbot example we explain that there are basically two main ways to use the AI assistant. We call them precision and exploration mode. So in simple terms do you want to use AI for fact finding or as a brainstorming partner? And when you use the chatbot for fact finding you're looking for the specific verifiable information that already exists in the process library. This is what we call the precision mode. You have a clear question and expect a concrete answer. And in this mode the eye behaves like a very knowledgeable colleague who knows exactly where the processes are stored, what they contain. It gives you a short explanation and also links the sources so you can verify everything yourself. So this is really ideal when you need clarity compliance or confirmation. Then in the second mode the explanation or brainstorming part so here you're not looking for one single correct answer. You're using AI as this thinking partner so instead of just retrieving the processes the chatbot helps you generate ideas, explore options, structure concepts still grounded all in the approved process knowledge but you treat it like a collaborator not just a search engine. And in this mode AI combines the process knowledge with the creative synthesis and it turns this let's say static documentation into the actionable ideas, proposals, drafts or starting points. So for example you can say you're my continuous improvement coach I want to reduce the cycle time and our PPAP approval process. Please suggest three approaches one conservative, moderate and bold referencing the relevant procedures include pros and cons, the expected impact and dependencies. Or please act as a facilitator, propose a 60 minute workshop to improve our use of the 8D method provide an agenda roles, an interactive exercise and handouts based on existing process materials. Ask clarifying questions to improve the proposal. So all of this context in my view has to be provided before rolling out even a seemingly simple AI tool in a large corporation. So that's really important to get the people on board and understand how AI works, what are the possible outcomes, the limitations, the constraints behind that. Other than that, we have lots of pilots testing out but in a more smaller scope I mentioned this with supplier interaction that the standard requests we send out to suppliers checklist on their management system on their terms and conditions etc that the AI is using this information and evaluating the information and feeding this into our system or like in the production lines we have optical inspection and we have some known errors in the products pictures of these feeding the eye with these known failure types so that the AI can easily identify or predict possible failures and errors in the production line. So these are all pilots with a more limited scope but if you really roll it out globally to everyone you still have to think a bit bigger and teaching the people to use AI.

SPEAKER_02 48:45

I think it draws on two of the things you mentioned earlier, right? One is giving people the right structure to do things, using AI in the right way there's a process around how you do that the lensing the framing the role you're asking it to play. And then the other piece is VI you mentioned that sounds very similar to the kind of predictive maintenance that was being done in the aviation industry. Just uh catching up predictive maintenance is such an interesting trend. You did also just mention previously a webinar series but your own ChiliCon Q. Do you ever explain that and some of the interesting things you've heard there?

SPEAKER_00 49:14

Yeah let's say co-anchoring the ChiliCon Q webinar series from the German Association for quality we invite different people from that bring different fresh perspectives into the field. So with every single guest speaker we invite I really take away fresh impulses for myself and for my work. And as I said we really intentionally invite speakers who sometimes have a more provocative take on topics like process management, quality management, maybe also people who challenge established ways of working and question longstanding approaches and that doesn't mean that everything old is bad. Sometimes you just need to take a step back and ask whether it still fits in today's reality. We also bring in perspectives from other fields like software development. One example is the concept of shift left I just learned recently to be honest when I first heard the term it didn't mean much to me but the principle behind is really pure quality thinking. It's all about identifying errors or bugs as early as possible in the process. So the earlier you find an issue the cheaper and easier it is to fix instead of controlling or identifying an error or a bug at the end of it. Making lots of quality assurance checks at the end of a process. That idea is really deeply rooted in quality management. It's just expressed in a different language and we also invite speakers who really look ahead who talk about the future so how quality management is evolving, where it's heading next and this really fascinated me. It's something I've been thinking about for a long time and where also then my thinking about use of II in quality management and process management really shape how this could work in the organizations.

SPEAKER_02 50:57

I like the Schiff analogy it's kind of like a snowball right catch a snowball at the top of the mountain it's just a snowball at the bottom it might be an avalanche.

SPEAKER_00 51:05

Are there any other provocative more outvertakes you've had on a webinar recently that have been uh interesting I really had an interesting conversation with the one of the process management and that process management is really the foundation of I so that was really an interesting conversation I had with him discussing the potentials the constraints but also then the need of change in the mindset of the people in the organizations still seeing process management as a burden just as a documentation part of that. So that this is really the trickiest part getting the understanding the acceptance of that in the organization so it's a lot of communication stakeholder management influencing the people to move this or to use this hidden treasure in the organization for potential future agents. So that was really a very interesting conversation we had and different perspective from environmental inputs so people dealing with environmental sustainability topics. So there's always a little intake a little piece of trigger that I can make use of in my daily work or I could use this in my organization. So it's really a bouquet of flowers I would say I like that a bouquet of flowers.

SPEAKER_02 52:20

And where can people find the webinar series?

SPEAKER_00 52:22

It's in German unfortunately we have a couple of English webinars but you can find them if you look for the German association of quality and then look for the webinar Chili Kung Ku then you will find them in the internet.

SPEAKER_02 52:36

Fantastic it's been really interesting kind of going through your career and all the things that you've developed along the way different insights different experiences. Just as a closing question if you were to travel back in time and talk to Karina start of a career or someone else who's just beginning on that pathway donor process quality transformation career journey what would you say to them?

SPEAKER_00 52:57

Yeah that's a good question. I think quality first sounds as I said in the beginning sometimes it's still really seen as bureaucratic, old fashioned not really up to date so that's something that could maybe be scaring people to go into the process management world or the quality management world. But actually it's quite interesting how much you see of an organization. So process and quality work really gives you a bit of a startup like perspective inside an organization and that's exactly where I think or what makes the role so powerful and interesting. I've never expected this at the beginning at least what I've been taught during my study time. Of course I had the practical experience at the beginning and the insights how this fit all together but making this really as an advertisement for getting into this process management world you really see across boundaries so the processes they don't belong to one function or one side or one hierarchy level. They really cut across all of them so that means you're not locked into a single silo you see where handovers break where priorities conflicts maybe local optimization hurt the overflow so the rules give you that kind of the end-to-end visibility of a whole organization you also operate very close to real problems not just the reported ones so when something goes wrong it usually shows up in the process it delays you need to have to do rework confusion workarounds so you're not dealing with abstract strategy slides so you're really dealing with how work actually happens in an organization overall that's very similar to a startup mindset where feedback is very immediate and the reality hits fast. The role also sits at the let's say intersection of intent and execution so the leadership defines direction and strategy the teams executed and process and quality connect the two basically so they translate the ambition into the mechanism that actually scales without losing control. So it gives the role quite real influence maybe behind the scenes but still you have quite a lot of influence it's also powerful that you work with constraints not just with authority so usually you don't own the people or the budget in the classic sense. So you earn the progress by creating clarity alignment and trust it forces you to be very strong in communication in stakeholder management systems thinking again very much like in startups. And now with the AI and the automation coming into play process and quality work is no longer about documenting the past it's really about shaping decision frameworks for the future. So you are helping the organization to decide what should be standardized, what can be automated where human judgment really matters so if you do it well the role becomes really a leverage point. It might be not so loud not so flashy still very incredibly impactful. So when the processes work the whole organization moves faster with less friction and with more confidence. So it's a really great opportunity to work in process management.

SPEAKER_02 56:18

Super Karina, thank you so much for coming on really appreciate it. It's been an absolute pleasure speaking with you and suppose you listening thank you very much for listening to Work in Process.