Cross-domain collaboration: Removing barriers to data

Transcript 

[00:00:00] Kimberly Boyd: Welcome to Pragmatism in Practice, a podcast from Thoughtworks where we share stories of practical approaches to becoming a modern digital business. I'm Kimberly Boyd, your host for today's episode, and I'm joined by Marina Ribke, Vice President Enterprise Architecture Management and Digitalization at Exyte, and Sebastian Werner, Head of Data and AI Solutions for Manufacturing and Energy at Thoughtworks.

Today, we're going to discuss how organizations can remove barriers to data to enable collaboration. Welcome to both of you. Very excited that you could join us and have this discussion today. Sebastian and Marina, before we get started, it would be great if both of you could take a few moments just to introduce yourselves to the listeners. Tell us a little bit about you and your roles.

[00:00:48] Marina Ribke: Yes. Sure. Thank you for introducing me, and thank you for having me. I'm Marina Ribke. By the way, I like the Ribke, the American way of pronouncing my name. Currently, as you said, I'm the Vice President of Enterprise Architecture and Digitalization at Exyte, a company in construction and engineering. We're building this wave of fabs. Without, there would be no chips in this world, so it's a very, very interesting business. Before that, I was in the automotive industry, was Chief Enterprise Architect of Mercedes-Benz, and was there establishing data collaboration, data ecosystem called Container X for the automotive industry. Happy hear to introduce myself as an architectural digital innovation based on data.

[00:01:35] Kimberly: Wonderful. It sounds like you have lots of relevant experience to draw from when we're talking about data collaboration, so look forward to learning more. Sebastian, how about you? If we could hear a little bit about yourself.

[00:01:47] Sebastian Werner: Very pleased to join you. I'm joining you from Munich. My name is Sebastian. I'm leading the Data and AI Solutions team for manufacturing and energy at Thoughtworks. I'm originally a chemical engineer by training, so that's pretty close to what Marina is doing right now from the manufacturing aspect. I spent most of my professional career on the intersection of data IT, meaning the computational part of it, but also on developing solutions based on the data in that set and in that set industry.

[00:02:18] Kimberly: Data has been positioned as the new gold, the new oil, the hot thing for some time. Now AI's come in and try to steal its thunder, but data is very much still a part of any AI discussion. With organizations increasingly starting to realize and tap into the value of their data, are you seeing an increased need for data sharing across your organization, Marina, or Sebastian, across the organizations you consult and serve?

[00:02:52] Sebastian: I would say without any question, but first a remark. I believe that data in itself and the, let's say, comparison with oil fits quite well for the very reason data itself, when you store it, it's quite dangerous because it can easily get lost and so on. What you really want when we work with data is actually the insights that we can generate for it. It's a raw material for making, let's say, all of these businesses run right now. Despite that, data surely is the new oil. It makes the organizations run, and there's definitely a case to use it more effectively and efficiently.

[00:03:28] Marina: I absolutely agree. What I see in many industries currently is that many industries come at the end of the way. They organize in an industrial way. The way of organizing supply chains, the way of organizing innovation, the way of organizing collaboration has come to an end. I have seen the side in the automotive industry where supply chains broke, where innovation becomes difficult, and sustainability, as another topic, is the way of the industrial era, the way that we're successful, that what brought us there does not bring us into the next era.

What I've seen in the automotive industry and what I now also see in the construction industry, and many industries, by the way, as well, is that the way going forward with resiliency in supply chains, with sustainability in the whole business models and also the supply chains. With innovation, the limiting factor are information. Information is based on data. That's where this oil comes into place. The raw materials for information is data and everything is limited by data currently. If we collaborate better based on data, that would be a leapfrog into the new era and in resiliency, in sustainability, in innovation because limiting data is, today, the limiting factor of the industrial era.

[00:05:12] Sebastian: To add to this, the surprising part is that is true both within organizations themselves where you have isolated departments that try to share data with each other. That already is difficult, but what Marina is outlining is the part it becomes even more difficult if you try to share data from one organization to another one because that's really where, let's say, the trend is going that you no longer, let's say, print out the freight manifest and stick it to the palet that you're shipping alongside the supply chain, but instead directly handing the data over in a digital way without the data loss with all the quality problems and so on.

[00:05:51] Kimberly: I can absolutely see how that can become more of a challenge. With everything operating in much more of a ecosystem manner, seems like that will become increasingly complex. I want to follow up, Marina, on what you were talking about, data information really being that limiting factor to being able to drive real collaboration within organizations, even with outside parties. Where should organizations or individuals begin if they want to remove those limitations and start to make strides in being able to collaborate and share data?

[00:06:35] Marina: That's a really good question. Where to start? It's really a multidimensional challenge. I very often see that people start with technology. They said, "I need a technology. I need data sharing. I need technology," but from my experience, that doesn't last really for long because if you don't have a compelling why and a compelling why do you do it, you won't have the energy to provide to the long way of the paradigm shift and transformation you are going.

To make an example, so when we start building a data ecosystem for the automotive industry, for example, we always said we bring solutions that solve problems for the whole industry. When that was the starting point, we said, "What are the real business problems that we must solve for the whole industry so that it can survive in the age of digitalization, sustainability, broken supply chains and really focus on our business cases and use cases and focus on resiliency? We must get this data along the supply chain fixed."

Sustainability. Sustainability is an information problem. Product carbon footprint, recycling, all that, human rights. Following down the information from the raw material to the final product along the whole supply chain is all information. That has been to that time mainly collected manual by people talking to each other along the supply chain, collecting manual contracts, and getting information right away.

To identify this data that we collect manually, that takes so long, we need so many people to do that. That is where real business value comes into play to say, "If we have this data available for sustainability, for resiliency, then it's a really, really good starting point because you have a compelling reason." Without a compelling reason that really makes a significant change, it will not last long. Then comes, of course, next thing would be the technology when you start with the business.

[00:09:00] Sebastian: Let me directly complement. I complete agreement, and to phrase it slightly differently. The key leading question when dealing with data-- and, in the end, for corporations, making use of data is nothing new. The question of how can we efficiently drive value generation by data first across the entire organization, but then subsequently along the entire supply chain, that should be the guiding north star of your entire operation still. Doing a tech push first is, I believe, not the way to do it. We've tried it many times. It really doesn't work.

If you have a compelling business reason, compelling business problem that you try to solve, and you, of course, want to do that efficiently, but the most important part is that you do it effectively. It really moves the needle, that it really has an impact on the cost, meaning the bottom line of the organization and, in the best case, even on a top line, meaning improving innovation or sales and so on. That, I think, should be the guiding principles that you follow.

The last part, and I think that's most important we deal with data is also keep in mind we need to manage the risks. Loss of data is also a challenge. There's personally identifiable information potentially in there, so keeping that in mind as well in these whole thoughts should be from a, let's say, compliance standpoint, also something to consider.

[00:10:22] Kimberly: Beginning, not with the tech or the tool, but beginning with a common shared understanding of what challenge, what business need are we trying to resolve or collaborate with data, with information, sounds like step one for folks. Perhaps I'm reading to both of your responses to that, but it sounds like that's probably also a common problem or challenge people make with not starting in that step one. Could you talk a little bit about what some of the other barriers are that organizations might need to overcome when they're thinking about how they want to set up cross-domain data sharing or even cross-department, cross-organization data sharing?

[00:11:09] Sebastian: I live in Germany, and in Germany, often the first motive when a new idea comes out is to immediately start thinking and listing every possible risk that you could somehow imagine in that way. Whereas I know from living in the US for quite a while that the opposite approach is taking there, first thinking just about the opportunities that are attached to a certain thing, and then later on finding out maybe that wasn't the great way. Of course, let's say, from my experience, truth is always in between, but there's also an opportunity cost of not moving forward with, let's say, the use of data.

What I mean with that is that we're not just here alone. There's global competition and so on, and there's a competitive pressure to basically do whatever we are doing faster, better, and cheaper. Integration of, let's call them, the data value chains, as Marina described earlier, is an opportunity to go forward there to cope with external factors such as uncertainty, meaning all the supply chain challenges we've heard about in the past couple of years, competitive pressure, but also experience loss in the job market, and so on and so forth.

[00:12:22] Marina: Yes, I fully agree. In sharing data to external likes like data spaces with your partners downstream and upstream in the supply chain, the biggest challenge is, oh my God, what intellectual property do I give out? What risk do I take in, for example, quality issues? It comes to there is a failure in the car. What liabilities come out of that? What risk occurs to the companies?

We have in Germany tend really to see the risks and tend to avoid collaboration as much as possible due to the risk factors. We're very, very high in seeing the risks and not the opportunities. That's one thing, but also see from the internal discussion in the company is if you need to provide data to others, you need to put effort into your data. It must be good quality. You must have an infrastructure for that. The key question is always why should I put effort in it? Why should I put effort in? Why should I share data? What's in it for me?

I think that's something that it cannot be answered from an individual point of view but must be led from the top, from the business. Collaboration in itself, what's the value out of it? That's why we were, again, with the why. To make very clear why is it useful for a company as a whole to come in the mindset of collaboration that is worth putting data into it and sharing it?

I was also leading the open-source practice at Mercedes and the governance for that, and it was the same. Why should I do open source and always say, "Why should I give in?" We always said, "Give in and share to get back so much more." I think this is the mindset that must be led from the top management, that information economics or the data economics and data sharing is really put in the effort to share and to get so much more back within the company but is also true for the whole industry in the end. That's what collaboration mindset makes in the end.

[00:14:49] Kimberly: You read my mind, Marina, where I was going to go with my next question. You've talked so much about the importance of having a shared understanding of why we're doing this, what's the business need for it. I was curious: what role do you think leadership plays in fostering an environment where people are really thinking in that way and thinking about the value and business benefit of sharing data?

[00:15:15] Marina: You cannot overestimate the role of leadership on all levels in fostering the data culture. Because it's much more than data. It's just that digital mindset of how you do business in a digital way or in the age of digital and AI. That is so different than the way you did it in the industrial era. Always I talk about digital transformation. I take the picture of an iceberg and say 20% is technology you see on top. Know that you have processes and data that is 40%, and under that, you have people and culture. You will never be able successful in a digital way with only looking at the technology, with only looking at the data and the processes.

You also need to care on your culture and also empower the people with data literacy, with how you really can handle the data, how you take decisions based on data. You must go in a leadership role because it's, in the end, a paradigm shift on how we lead businesses. Leadership can stop that by simply saying, "I really don't see where the impact is," and stop that or can go ahead and drive that. I've seen both.

One simple anecdote I always bring when I discuss with top management in the automotive industry was, "Marina, you're talking so much about digitalization. Why are we doing that? We did that three years ago, so it was not successful. We don't do that anymore." If you go there with that mindset, nothing will happen.

[00:17:13] Sebastian: Adding to this, the one thing that I've seen clearly as a recurring, let's say, success factor is if there's clear buy-in from the very top. People say, we really, really want this. Because otherwise, there's very little chance for success. Then on top of that, I would say this mindset change that Marina hinted on is the first part is to say, "Hey, be clear that if you're trying to gain additional benefits in an exchange of industrial revolution, you need to embrace the team sport." Meaning saying, "Hey, we don't play separately. We think in entire supply chains instead of just in one product."

On top of that, thinking of everything in the very end as a product, no throwing over the fence, thinking about the whole picture, and then building things to scale in that way. Starting small and being able to grow in the next phase. All of these principles, in the very end, if they are embraced from the top, can enable the next step of innovation, I believe, especially by data use alongside supply chains.

[00:18:23] Kimberly: Following up on that, no one can see our faces, but we're nodding in vigorous agreement to all these points. Is it the continual reinforcement of those principles, Sebastian and Marina, that you were just talking about within an organization to really drive that home? I guess I'm curious with where have you seen it work really well or where haven't you seen it work well to help folks understand these are the types of actions that'll drive desired behaviors.

[00:18:57] Sebastian: The first part is change is hard, and already changing within an organization is really, really hard. Change, by my experience, starts from the top, meaning leaders embracing, in that case, a certain change. Let's take data: if you take leadership that suddenly challenges each other based on data, then suddenly, the value of data within that organization will go up. Now, think alongside supply chains, what Marina said earlier. If suppliers request from each other that there is more extensive data than just the, what I previously said, that loading manifest that is coming with the pieces alongside the supply chain, suddenly the importance of that goes up.

With that, it becomes easier to drive than these individual business cases and these first couple of changes to integrate individual steps in that data supply or data value chain. I think it all starts from not just talking about it but doing it.

[00:20:00] Marina: Yes, I really only can support this. From the leadership, where can you really start making first step? I think this is, if you take decisions, get away from highest-paid opinions or just from opinions and really focus or really demand that every decision we take must be taken on data. We don't discuss without data anymore. Bring it to the table, and leadership ask their teams, "If we take a decision, bring the data, and we discuss on the data, and we take a decision based on data and not on opinions."

I think if you continue to do that and then say, "Okay," and then you have the next discussion, "Where do I get the data from?" If I need data to get decisions, I need data from the whole organization. Who owns the data, by the way? This is in a classical organization, a topic that no one can answer. A question that no one can answer is, who is owning the data? That's the key challenge. Define ownership. That's what we call data governance. In the end, is take ownership of your data, build teams that say, "You are responsible for that data. You are responsible for that data. You are responsible for that data."

Then create a sharing environment, an ecosystem. For example, start with a data catalog, then start with sharing based on usage policies, whatever. Step by step, you come into such a data-driven organization, but it starts again with the business. The leadership has to start with and can start simply asking for any decision must be taken based on data.

[00:21:43] Sebastian: If there's any leadership that themselves behave analytically and show passion for data and analytics and the company, let's say, has at least from the top a drive to make decisions from data, it's significantly easier. There's the cultural aspect. Basically, there's also a high standard of how you basically want your data to come to you, not just in a printed report that is six months old, but using live data. That's basically a next dimension to that. There's also the incentivization of using data by setting clear targets or setting clear numbers that you can actually control based on that live data so that leadership really needs to comply also with that.

[00:22:25] Kimberly: If I'm playing back everything we've discussed so far, it's starting with that clear business need and use case. Why are we doing this? Making sure that there's leadership support and buy-in, and they're reinforcing that message in the organization, but not just talking the talk, walking the walk, too, being that analytical data-driven leader and modeling those behaviors in the organization. The ownership piece, the data governance that Marina just mentioned as well.

I know we talked about It's not all about tools, and you shouldn't start with tools or tech first when you're really trying to accelerate the sharing and collaboration with data in the organization. I do imagine that concepts, tools, and technologies do play a role in successful data sharing and collaboration. I'd love to get both your perspectives on what concepts and tools have you found to be useful in driving data collaboration and sharing.

[00:23:33] Sebastian: I would say my first learning is that open standards is a big step. Using proprietary technology just makes it more difficult. First thing I would say is, embrace open and clear standards is for me a concept that is essential for success on the way forward. Even as a technologist, saying that will hurt a little bit, but hey, a tool or let's say a certain product is only an enabler to make a certain case run. We are not doing AI. We are not doing data for fun. We do it for a business purpose. I think that's a key point to realize. However, when we do that, to have certain guiding principles to make it a success. For me, open standards is the first one, and I think the most critical success factor.

[00:24:27] Marina: Yes, I fully agree. This is really open. This is one of the key things. When we were starting data ecosystem, Catena-X, for the automotive industry, we really relied on the principles that we said, "What do we really have in case or implemented that it becomes successful?" We relied very much on such principles. We exact, okay, we rely on interoperability. That's where we have data at the center. No matter what technology we use, you must be able to access data independent of your technology. You must be able to collaborate on one use case independent of your technology that complies and is implemented then by open standards.

That was the core of a data ecosystem that we defined is what are the standards for interoperability now, collaborating along one use case. Next thing is what we relied on besides interoperability is data sovereignty. Sovereignty means I never lose control over my data. I always have a contract at my data. When I share data, I can rely on that it's only used in that way that I wanted that it's used. Data is much about sharing, and in every sharing economy, someone owns it.

If you share your car with your neighbor, you want that the neighbor brings the car back again and that it's in a good state. The same is with data. I don't want if I share data to someone, that it turns over to another, to another, to another, and I got too much loss of control. Data sovereignty implemented, for example, by usage policies and implemented and executed by technology. That computable data contract is something we thought that is really crucial to get these things on intellectual property and risks and trusts to get that done. For that, next thing, of course, it must be secure. Security is one big topic. We relied on principles like interoperability, data sovereignty, security.

[00:26:49] Sebastian: Directly building on that, in the end, it's about creating accessibility in a systematic way by these standards and building it with resilience and governance in mind, and I would add one aspect to that that comes out as a benefit. If you assume that the technology will change from the very beginning onwards as a mindset, it will make it a lot easier to achieve this. Because if you just lock into one technology and assume they will always stay the same, I can already guarantee you this is exactly when you get the next big enterprise modernization waiting for you.

[00:27:26] Kimberly: Since this is, after all, a Thoughtworks podcast, and we're talking about data, I can't not ask about data mesh in the role of fostering data collaboration and data sharing. Sebastian, it would be great to get your thoughts on the role data mesh has played in accelerating or scaling organization's ability to collaborate with data.

[00:27:55] Sebastian: Excellent question. To just give our listeners a short intro, data mesh in the It has four core principles. Two of them are, let's say, in general, about keeping delivery and knowledge local and making sure that you truly enable the business to be effective in use of the data. These are the principles domain ownership, which built on, in the end, domain-driven design, as well as data as a product.

It is two further principles that are about providing infrastructure, strategy and policies in an aligned way to be cost-efficient and fast. These two principles are self-service data infrastructure and federated computational governance. Overall, we have the four principles: domain ownership, data as a product, self-serve data infrastructure, and federated computational governance.

That being said, these four principles, I believe, are fundamentally compatible with everything we've discussed already. If you, within your organization, already moved with data mesh, this decentralized, socio-technical approach to, in the end, bring data to a better use, I believe it is an easy step to add the additional capabilities that are needed to either go towards Catena-X or to any external data sharing ecosystem because you've already done the point. You've already capsuled your individual data as data products, and adding a second external port that will be externally exposed towards consumption in a Catena-X ecosystem should be a lot easier than if you try to go the whole way from the very start.

[00:29:34] Marina: Yes, I fully agree to that. Data mesh describes really great principles. They are also, in my opinion, really great to help me to do the transformation into a data-driven company. As I first read about data mesh concepts on Thoughtworks, it was great. I took that also directly at Mercedes into the organization, and we're really trying to experiment and implement, trying to experiment on data meshes. This was also very inspiring for me for the design of Catena-X.

By the way, these principles are our foundation also for Catena-X, and Catena-X, I can say, has guided my thoughts on the data mesh, has guided my thoughts on Catena-X, and Catena-X would look different without that data mesh article that was published by then. These ideas laid the grounding principles also for Catena-X, and if you have a data mesh in place, it would be easy to connect to any data space in the world because any data space architecture relies on the same principles in the end with different technology, but the principles are the same. If you have data mesh in place, you're good to go with collaboration for your own industry.

[00:31:09] Kimberly: Just for our listeners who might not be familiar with Catena-X, Marina, could you speak a little bit about what that is?

[00:31:15] Marina: Catena-X is a data ecosystem that was established by a consortium of 28 companies, that established a data ecosystem that has the ambition to share data along the whole supply chain of the automotive industry to connect 280,000 entities globally in this network, or in this global supply chains, and to share data along the whole supply chain from the OEM to after services down to the raw material in an interoperable, standardized, secure, and sovereign way.

This network is up and running. I was a founding member of that and was, in that time, responsible for the enterprise architecture in Catena-X, so was leading that from the enterprise architecture point of view. I was responsible for bringing in the principles and implementing them, like interoperability, data sovereignty, and security into that system. I think this is really a paradigm shift in innovation on how the automotive industry can share data. That improves resiliency in many use cases in the supply chain and also solves a lot of sustainability problems, like bottom carbon footprint, human rights, and all that stuff. That's what we did, building an ecosystem built on the data space architecture.

[00:32:57] Kimberly: No small feat, it sounds like, given the scope and number of players in that ecosystem. Probably one of the bigger data collaboration challenges out there. Sebastian, I heard you say the A word a little while earlier, AI. We really can't have conversations these days without mentioning it. I'll do it here and ask, what is the role of AI and machine learning in overcoming challenges to data sharing and collaboration? What role can they play in helping improve or better facilitate?

[00:33:39] Sebastian: That's a brilliant question. With AI, you can refer to different things. I believe right now, a lot of the conversations we're having are around large language models in that way that brought this huge boost with products, chatbots, and so on that are coming in. One of the applications I can think of in the topic we are talking about right now is helping us to standardize, let's say, the metadata descriptions that we have during data sharing.

Writing or helping us to summarize them in a standardized way. Describing data sets before they are published in order to make it easier for reviewers to understand what they're actually trying to share. Also for legal evaluations that are that might be there. That would be the first things that come to my mind. Meaning don't use the AI directly in the operational part but in all these tedious administrative tasks that you have to do within an enterprise before you are able to share data externally.

[00:34:38] Marina: Yes, I fully agree. I think this is tedious tasks. That's where we can really see a lot of value, also I see a lot of value in that. Also not only in metadata but also data cleansing. Get data clean, get the quality of the data assessed and right. It's metadata and data quality where I see, where you can start with.

[00:35:03] Sebastian: Also, to add to this, yet another case that we've recently seen with customers to be very interesting is taking, let's say, historically valuable systems that produce certain types of data but are written in programming languages that are very hard to understand. Meaning FORTRAN, COBOL, and so on, but using large language models in getting a first description, what does this actual module that we are using for many years actually do?

With that summary, use that as a briefing for experts to then help to, I would say, almost lift and shift. Take that module that you have and take it from the existing space into a modern infrastructure, let's say into a data mesh architecture where then it is refactored in terms of a data product to then be valuable and, in the end, reduce the time and effort that is required in order to do this modernization.

[00:35:58] Kimberly: It sounds like there's definitely a significant role it can play in the upfront. Doing a lot of the grunt work so people can then get into the interesting "What is the power of this? How is it actually going to help us address whatever the business use case is?" Going back to beginning of today's conversation of where we need to begin. I know you've both been very generous with your time to have this data chat today. Maybe one more question before we wrap up our conversation. I'd love to hear from you both what are you most excited about in the space of data, in what the potential holds for really unlocking data collaboration and data access?

[00:36:43] Marina: I think data is where we're coming to the beginning, and then the data is the raw material for information. We're in a paradigm shift, I think, of our whole world. We're transforming from the industrial era into the information era already right in between. I'm for decades now helping to transform the world, industries, companies, and people to transform from the industrial thinking and organization and the way of doing business into an information and data-driven-based way of making business or doing and organizing in industry.

Then that's what's fascinating with me, with data at the core and organizing technology organization all around that to get most of that information to help people better to collaborate, to innovate, and to create a better technology for a better life for all of us. That's what fascinates me for decades and every day more and more.

[00:37:57] Kimberly: That's a great thing to be excited and hopeful about.

[00:38:02] Sebastian: For me, it's really that data really starts to move the needle. Meaning it becomes a core part of how businesses operate, and that we are more and more tackling some of these complexities and break them down into simple individual parts. Companies also realized that there's no way back. There is not only no way back but there's also no way of avoiding it. What makes me excited about that is, it's that even in Germany, companies think about opportunity or start to think about the opportunities, what they can do with data, and how to overcome these long-existing paradigms.

"Hey, we cannot share that because of intellectual property, this and that and that," but that they think, "Hey, wait a second, we need to compete in a different way with our competitors, and we need to put the customer in focus. What the customer wants is that he has the data accessible at his hands, at an instant, and that it's no longer an added feature but it's a necessity." That makes me excited because it's only the beginning.

[00:39:10] Kimberly: I love both those answers a lot to be excited and hopeful about. It sounds like, Sebastian, perhaps there's some American approach to data that's rubbing off and eroding a little bit of the German risk aversion there. Marina and Sebastian, thank you so much. I really enjoyed our conversation today, and it's been wonderful to have you on Pragmatism in Practice.

[00:39:37] Sebastian: Thank you very much.

[00:39:39] Marina: Thank you. It was a pleasure to be here.

[00:39:41] Kimberly: Thanks so much for joining us for this episode of Pragmatism in Practice. If you'd like to listen to similar podcasts, please visit us at thoughtworks.com/podcast. If you enjoyed the show, help spread the word by rating us on your preferred podcast platform.

[00:40:04] [END OF AUDIO]