Andreas Welsch 0:30

Welcome back for another episode of What's the BUZZ?, where leaders share how they are turning AI hype into technology outcomes. Today we'll talk about the top three success factors for agentic AI, success factors beyond technology to be more precise. And I could be more excited to welcome our guest to the show, Sebastian Wernicke. Hey, Sebastian, thanks so much for joining.

Sebastian Wernicke 0:50

Hi, Andreas. Thank you for having me. Excited and looking forward to the conversation.

Andreas Welsch 0:55

Awesome. Hey, it's been a couple months since we've met in person.

Sebastian Wernicke 0:58

Yeah.

Andreas Welsch 0:58

Earlier in, in the year when I was launching my book. And I know you have a new one coming out. We'll talk about that in a in a couple minutes. But maybe for those in the audience who don't know you yet can you introduce yourself real quick and share a little bit about yourself and what you do?

Sebastian Wernicke 1:12

Yeah, of course. So I'm Sebastian. I always like to describe myself as an extremely curious person. I think that's who I am. And so it's no wonder that I work in data and AI day to day. I'm a partner at Oxera, a consultancy where I lead the data science and AI team. And now I'm also an author. We recently released the new book called Data Inspired. And yeah, this is as you know also as an author, this is now becoming a second job that I have,

Andreas Welsch 1:38

wonderful. I'm so glad you're spending your time with us, and I know you- you've been in, in the industry for a long time. You've also built quite a following on LinkedIn. Folks, if you're not following Sebastian yet, head over there real quick and give him a follow. Yeah. Lots of good information about data and AI that he posts there on a regular basis. Now, In good old fashion, we'll start by kicking it off with a little icebreaker. And this one is- Okay one that that I've been running for a long time. And- Do

Sebastian Wernicke 2:04

I need to hit a buzzer button or anything or is

Andreas Welsch 2:06

this- I will. I will do that for you automatic. Oh you

Sebastian Wernicke 2:07

buzzer. You buzzer.

Andreas Welsch 2:08

Okay. Something that would show up here, this is the one that I will press. So let's see. If I press it, the wheels will start spinning and you'll see the questionnaire. If AI were a plant, what would it be? 60 seconds on the clock. Go.

Sebastian Wernicke 2:22

If it were a plant.

Andreas Welsch 2:24

What's the first thing that comes to mind?

Sebastian Wernicke 2:27

I'm go- actually, okay, so I'm going to offend my, my fellow biologists. I'm not gonna use a plant, I'm gonna use a fungus here. So if AI were a fungus, I think it would be mycelium which is that fungus that some people might know is growing inside of the earth, and it's connecting the trees and various organisms. And AI is of course also this really big network of connections. So I think a fungus is more fitting here.

Andreas Welsch 2:56

Okay. I like that. Some people have said it- it's a cactus. Others have another- Tricky

Sebastian Wernicke 3:01

I like that. Good With

Andreas Welsch 3:01

the little pointy spears. But fungus, that, that's a good one, right? It's probably been there for a long time too, and it keeps on growing.

Sebastian Wernicke 3:09

Yeah. Technically incorrect, as I said but,

Andreas Welsch 3:10

Yeah

Sebastian Wernicke 3:11

we'll take it,

Andreas Welsch 3:12

ex- exactly. We're not biologists here, at least few of us are. Now, boy how do I segue in into our actual topic? For for a fungus to to grow and plants to grow we have some success factors. For agentic AI to work we need them too. And it's not just about the technology. So we'll talk about three, and I know we spoke about this a little bit before we went live, but maybe if you can share what do you think are the three key success factors that you're seeing? And then we'll talk a little bit more-

Sebastian Wernicke 3:39

Yeah

Andreas Welsch 3:39

about your book and what it means to be data inspired.

Sebastian Wernicke 3:42

Oh, I think the analogy is actually perfect because for a plant to grow, you need the right substrate, right? Not every plant will grow in every environment. And so I think for AI, and particularly agentic AI, there's three components that you need in your soil for it to grow and to flourish. The first one is data, because even though it's a very fancy technology, it's all grounded in data. So unless you have- the right information and you have it set up in the right way, there's nothing that AI can actually use to work on. The second one I think is culture, because where are we using AI and agentic AI? It's not in a vacuum. It's not a pure technology topic. We are putting these technologies into enterprises. And so enterprises even of the future, even if they are autonomous enterprises, they will still be comprised of people operating the whole thing. They will still have people making the decisions. And so we need the right culture within an organization for AI agents to actually unfold their full power and not, for example, just be used as a replacement for the process that should be replaced in its entirety anyways. And I think the third factor for me is strategy, and I know this is a topic that can sound quite abstract to many people, but what I essentially mean by this is... So in the book, I have this headline where I say, "Data second." Consider- strategy first, consider the purpose first. And I think with AI and agentic AI, that's become ever more true. If we just say, "Let's do AI" and don't have any idea what that means or why we're doing it or what the purpose of that is, it usually fails

Andreas Welsch 5:34

I like that. Data second. A couple weeks ago I was in an event and someone, a friend from municipal government in attendance and she said, "We're AI last." And everybody looked at her and "What do you mean?" She's we have so much catching up to do that while everybody talks about AI first, we're actually on the other end of the spectrum to become AI first." So I have a lot of empathy here for data second because we need to think about what is our strategy to begin with.

Sebastian Wernicke 5:59

Yeah.

Andreas Welsch 6:00

Now in February, I think it was we met near, near Philadelphia and I saw one of your talks a- about what it means to be data in- inspired, but I don't... Obviously not everybody here in, in the audience has had a chance to to see it or to-

Sebastian Wernicke 6:14

Yeah

Andreas Welsch 6:15

learn more about it. What is Data Inspired all about? We've seen so much about data. We've re- we've read so much about data. There's been so much written about data, but data inspired, what is that?

Sebastian Wernicke 6:25

Why are we still talking about data even, right? Yeah. It so- sometimes amazes people why we still do that. And I think the... there's two things to, to mention here. So the first one is, I think oftentimes when we talk about data, we use the term data-driven, and that term has been around for quite a while. So we want to build data-driven companies. We want to make data-driven decisions. And the thing that I appreciate about that term is I think it tells you very exactly where you're trying to place data within the organization, right? You're basically saying "We are having too many decisions where we don't feel that we have the facts right, where maybe gut feelings are prevailing. We're having too many discussions." And usually when somebody says, "We want to become a data-driven company," what they mean is, "We want to have more information and we want to have that information guide us. We want it to help us make the right decisions. We want to help, We want it to help us be faster, be more autonomous, be more automated." And so there's a lot of clarity with the term. And just to be clear, when I say data inspired, and I'll explain in a second what I mean by that, I don't mean to replace that. I think data has that place. Data is very important for measuring, for transparency, and ultimately for optimization. You just become a better and faster business if you add data to the business and you make good use of it. But I also felt that this wasn't enough in the end because if you think about it, what's happening? Every organization is collecting more and more data. Most organizations are luckily becoming a bit better, at least at organizing the data, even though that's still a huge topic, and they all have access to the same tools to make use of data. They all have access to essentially the same dashboarding tools, now the same AI tools. So the question becomes if you only do that, if you're only striving to be data-driven, how is that going to give you an edge? In my mind, data-driven is essentially becoming very quickly the baseline because if you don't use data to improve your business You're not going to be around for much longer because it's just become a very essential tool. And so I was trying to look at what's the missing component here? And if data-driven is a lot about optimization, then being data inspired is still about that, but it adds another component, which is how can we use data to transform our organization? How can we use data to drive innovation? How can we use data to find the next breakthrough product idea? How can we use data not just as this, steering wheel but more as a compass that shows us where the future of our company might lie. And so data inspired is a term I think that I found it was already around before that, but I found it, it was used in a way that was a bit wasteful of that word inspired. It was used as in a we have data around and we might be a little inspired by it." And I thought no. You need data actually to drive you towards the future." And so I was just thinking, let's give that word data inspired another chance, put it on the book title- Yeah and then hopefully we can actually have an i- inspired use for that term.

Andreas Welsch 9:41

I think this part is so im- important about the inspiration because there are so many technical books and works that, that cover how do you do the foundation. There are so many strategy books. But a lot of times we miss this part that you said when we talk about data driven, we don't talk about where do we actually want to be, what is the end state- Yeah and what could we do and what can we do with the data and the information we have available. And we know that y- your data is never as complete and as fresh and- as clean as you would like it to be or as you need it to be. So what's the inspiration? What are we going to do? A-

Sebastian Wernicke 10:14

and so deal with that also, right? There's also some parts where I think we just have to spill a bit of truth and say of course your data is never going to be perfect." And by the way, when you make an important decision, of course the data is not going to tell you precisely how you should decide. Yeah. That's just not how complex decisions work. And yet I think, and you've heard that sentence as well, where somebody will come and say, "Oh, if only we have the right people to the right Sorry, "The right data to the right people at the right time, then suddenly magic things will happen." And first of all, usually for many decisions, that's just impossible to do. And then the other part, so there, there are two chapters in the book that I snuck in there that are about psychology. Ooh, and but it turns out that- You know, this whole right data to the right people at the right time is also an illusion because it suggests that we're essentially just operating on a data deficit, and as soon as we get the right data, we will be doing better and we will be making better decisions. And human psychology for decades, and the s- the science behind it for decades has now shown this is not the case. You can give people perfect data, and unless they are psychologically, and in an organization that also means culturally prepared to use the data, nothing will happen. They will just dig in or dismiss the data that doesn't agree with them. And so this is why, besides having the right data, that cultural element becomes so important because if you don't have a receptive culture for data, and that essentially means being prepared to be wrong, to change your mind, you don't stand a chance even if you have the best data around that you could imagine

Andreas Welsch 12:14

So seeing why we're saying beyond technology, right? The success factor- Yeah beyond technology. And it seems that there's a lot more of that human element again in, like you said the culture the leadership culture too. In, in some countries we talk about a failure culture and being open and accepting that- on the path to success you will encounter some hardships, some challenges, some failures, and they're all part of the process of learning of getting better. But we also need to encourage this learning to happen. Otherwise, we we're- Yeah we're too afraid of risks to the point that nobody's willing to take them because there's such a big stigma around taking risks- Yeah and potentially not succeeding.

Sebastian Wernicke 12:52

And the most interesting data comes from failing, of course. Yeah. Yeah. And so if I want to generate interesting data, I need to be prepared to fail. I need to want to fail in a way. Of course, in a risk-controlled environment and all of that where I'm not betting the company, but still I need to be willing to do that. Otherwise, I'm just generating boring data that, may allow these stepwise changes, but isn't gonna drive transformation in the end.

Andreas Welsch 13:14

Now, a l- a lot of times that, that seems so easy or it sounds so easy. Yes, we need to em- Yeah embrace more failure, we need to embrace risk-taking, we need to em- embrace innovation. But then you get stuck in the day-to-day, you're in the- the quarterly

Sebastian Wernicke 13:28

goals. Yeah.

Andreas Welsch 13:28

Things that, that we do every time. You have some quarterly goals, you have some annual goals that you- Yeah really want to hit or exceed, you get dinged if you don't, but you've taken the risks. What are you seeing there when it comes to culture? How can leaders really, live and breathe that they kind of change?

Sebastian Wernicke 13:44

Yeah. I think leadership sets the tone for this, and this is a very important one. And what I always like to joke about is that I hear too many leaders trying to achieve efficient innovation in a way. So we want innovation, but we want to do it in a very secure way. That is not possible. And so ultimately I think it comes down to, and this is where we come to the topic of strategy what is the strategy driving all this? If your strategy is we want to be efficient quarter by quarter, we want to hit our quarterly goals precisely for the next quarter, for the next two quarters, three quarters, four quarter- That's fine. By the way, that is not a strategy at all. Even- by definition that is not a strategy. But, i- if you say that then go ahead. But we are, of course, realizing, or everybody's realizing we are living in a time of rapid change, of complexity, of, if you will, mega trends, right? Supply chains becoming extremely brittle aging population. We need to see where does our food come from. We do have climate change to deal with. We have aging populations. All of these big topics, and I think these big topics will compel you over time to be innovative. And not just in the sense of let's be innovative once and reinvent the company, but actually build companies in a way that they are innovative on a continuous basis, that they become resilient, that they change is the new normal, even though that's a bit of a cliche, I realize that. But that is the way that you need to build- a company. And so I think this innovation topic becomes very quickly strategic, and if you want to be innovative at speed and at scale, then all of the other topics that we start to, to talk about, AI, agentic AI, data, they become inevitable. But this is, turning it around and not saying, "Oh, how can I add AI to my company as it is today?" But rather saying, "It's a strategic imperative that we build around change and innovation," and the only way to achieve that at the speed and at the scale that we need is to embrace AI, agentic AI, data, all of these concepts

Andreas Welsch 15:57

L- let's switch gears a little bit. And I know that's a question that we didn't talk about before. We had very little prep and we said let's l- let's have a free-flowing conversation- And we already had a buzzer, so let's go for it.

Sebastian Wernicke 16:07

but you were

Andreas Welsch 16:07

really ending it. Yes. But something that that just crossed my mind because you talked a lot about culture, you talked about data, you talked about strategy. But I'm curious as an author, what is something that, that you've learned in the process of writing the book because I'm sure it's not a, four-week project, anything but that. What was this one thing that you that you learned, this big revelation maybe even compared to where you started now with where you ended with the published book?

Sebastian Wernicke 16:32

Yeah. So the book started out with many fragmented thoughts or so little frustrations. I always say that there's a bit of a not so noble way that I started writing the book, which was whenever I got into a argument with a client or on a project, I said, "Okay, I'm gonna write that in a book one day so I can say, sorry, but it's in the book," I didn't have to have that argument anymore. But then as you continue writing, of course, you're trying to find what's the red thread, what is keeping all of this together? And I struggled with this for a while until I had lunch with a friend of mine who works in organizational psychology, and she said, after we were discussing about organizations and culture for a while ultimately, what is an organization but a decision-making machine?" And this notion of it's all about decision-making, I think that was my biggest revelation that came to me because that ties all of it together. Why are we using data? To use... We're using it to make better decisions. That's what it's all about. If we don't make better decisions, then we don't need all that data because it's quite expensive, and then when we talk about AI what is AI? AI is something that we can hand decisions to mostly in the way of a recommendation engine or, before that we had decision rules or we had machine learning. So it's all stepwise, we have data to influence our own decisions. At some point we decide let's hand a bit of that over to the machine. With machine learning, it's an algorithm that has learned what we care about and we've been very precise about defining goals. With AI, suddenly that's no longer the case, right? If we define AI in the sense of large language models, they're basically trained on does the user like what I'm saying? A very broad goal, yeah, which we also then should be careful about. And with agentic AI, it's of course then the complete handover. We're essentially- Yeah just saying, "Oh I'll just tell you my goal, and I'll let the machine figure out everything." But decision-making is what ties it all together, and so I like to talk about the decision framework that an organization gives itself and operates itself in. Because if you look at it from the lens of decision-making, it becomes very clear what kind of choices you have to make ultimately. What am I going to hand over to an algorithm? And if I do that, what is the nature of that algorithm? Is it something that's very controlled, or is it something where I'm happy to say I'll specify a general goal and hope for the best," fingers crossed and everything. But it becomes very clear through that, and that, I think was my biggest revelation in, i- in the end. And it's a very useful framework to apply to this whole technology discussion in my mind.

Andreas Welsch 19:02

That's awesome. I love that. Yeah at the end of the day it's about decision-making. It's to your point about having the data to, to make a confident decision whether or not the reality turns out to be, right or wrong based on decisions- Yeah That you've taken. But just having data, having facts to make a decision and then making it quickly too. So to your point now with agents, when the- AI say we, we delegate more agency, we delegate more autonomy to these systems and components how can they make decisions that are a- aligned with our values, that are aligned with the facts that then represent the reality-

Sebastian Wernicke 19:37

Yeah that- And psychology becomes so interesting here, I find. So the so I distinguish, as I said, very clearly between machine learning and AI. And the interesting thing is machine learning is something where we control the decision environment quite tightly, right? So it's an example of, let's say, ah, the classic example, the tens of thousands of cancer images, and the machine learning algorithm learns is this dangerous, is this non-dangerous or benign. And the thing is that people intuitively, they don't trust machine learning. So there are studies where you can show people that, for example, a prediction algorithm is superior to the predictions that you are going to make. And then you ask people, "I'm gonna let you bet some money. Are you gonna make the choice? Is the algorithm gonna make that choice?" They see the algorithm making a single mistake and they will say, "Ah, I'd rather do this myself. Yeah, I don't trust this thing."

Andreas Welsch 20:28

Yeah.

Sebastian Wernicke 20:28

And now with AI, it's so ironic because we are, AI is trained in a much, much less controlled environment. As, we're throwing all of the data at this huge thing which is impossible to understand in its entirety. And we're training it on this goal of do I like what you say or not? Yeah. And still because of the interaction mode and because it can sound so natural, suddenly we trust this thing. So there's machine learning which statistically think we should trust because we can prove it's better than a human. Then there's AI where we don't, basically we can't prove anything, but we say, "Ah, yeah, but, let me tell you about my relationships," ask you for relationship advice ask you for strategy advice. Let's hand the business over. It's so ironic, I find.

Andreas Welsch 21:13

For sure. And to, to me that, that comes back to how we've learned to trust automation and- how automation has traditionally been deterministic. Every time I press the button it switches the channel, If then else,

Sebastian Wernicke 21:24

yes

Andreas Welsch 21:24

then we've ad- adapted more to I need to train my thermostat, for example, especially here in, in the US when Nest was a big thing. And it learns when I like it hot and cool, and when I come and when I leave the house, things like that to, to your point now delegating more and more to, to these systems. I I'm curious though, talking about data in- inspired, yes, there's data a- as a foundation, but we also want to see where we want to go, how data culture- and strategy fit into this. What's the one thing, what's the first step that anyone and any leader should take to become a more data inspired business or business unit?

Sebastian Wernicke 21:59

Yeah. I would recommend two steps, and they can go in parallel. I think the first step is really that strategy discussion and finding the purpose. Because without it, you are entering a large transformation without having a reason to do and so I think it's very important to understand how is doing this critical for the business? How is becoming a company that permanently innovates with data critical for the business? Unless you've discussed and answered that question, you can of course start the projects and the innovation, but we know from transformation projects, and there's tons of research on that, if you don't have a clear purpose, if you don't know where you're going, these things fizzle out. So sometimes I say transformation is like climbing Mount Everest, but not with oxygen, but with PowerPoint, and that is just very hard. So you need to know why you're getting yourself into that.

Andreas Welsch 22:54

I love

Sebastian Wernicke 22:54

that. Now, the other part of that is to, I think, realize that the culture discussion and the technology discussion Must not happen and cannot happen in separate rooms. And so we cannot just sign the check and say, "Let's buy the tools, let's buy the AI," which by the way, you need to do, yeah? This is not anti-tool. This is not anti-technology. Of course you need the tech. But if you don't supplement the tech with that culture discussion and saying, "How do we want to make decisions in the future? How are we prepared to hand over decisions to these algorithms that might or might not be trustworthy, and where our intuition about trust might be different?" To have all of these other discussions. And by the way, also the classic decision-making with data isn't going away, so you also need to discuss that because people usually are quite bad to make good decisions even when presented with very good data. But you need to initiate that as well. So the purpose part and the how do we tie culture and technology together, I think are two essential stepping stones that will lead you towards that path of inspiration. Ah,

Andreas Welsch 24:01

ha.

Sebastian Wernicke 24:01

I know it sounds a bit lofty but I think ultimately, the whole as... when, when I was writing the book, sometimes I was just thinking, "Oh yeah, that, that actually, that, that company I'm describing, that's a company I would like to work for." Yeah. So I think ultimately it is about building more inspiring companies that spark joy in a way, and energy.

Andreas Welsch 24:19

We certainly need more of those, Yeah especially when we look at the headlines and things are pointing in the other direction of being less inspired, maybe inspired by, by savings, cost savings, but not inspired but by what could we do if we gave people access to this technology or to- Yeah.

Sebastian Wernicke 24:34

And yet still, i- isn't that the that is so ironic, right? These are all things that are so well-described because that in the end is the classic innovator's dilemma.

Andreas Welsch 24:43

Yes.

Sebastian Wernicke 24:43

And it's been described technology cycle after technology cycle. Now AI comes along, what do we do? Seeing eyes, we're going into the innovator's dilemma. Ah.

Andreas Welsch 24:53

Absolutely. Now, Sebastian we're coming up to the end of the show. I'm just wondering if you can summarize the key three takeaways for our audience today. We've had such a great conversation. We talked on or touched on so many important points, but what are the key three things that the audience should take away?

Sebastian Wernicke 25:07

Yeah, so if you're excited about AI agents, I completely understand why you would be, but please do these three things. First, make sure that the data is right because your AI agent will be using data, and it needs that soil to grow in and grow in the right way. The second one is make sure that you understand that AI agents are not just buying a technology. They're all about changing your business. They're ch- about changing your organization, how it operates, how you make decisions, and ultimately your decision-making culture. So it's a culture topic. And I think the third one is don't start the way of saying, "I have a business. How do I now bolt on this technology of AI?" You have to go one step deeper into the strategy discussion and really say think about, what's really threatening my business? What's hard? What are the challenges, and how do I overcome these?" And no worries, I think the solution in 99% of cases will be we need to embrace AI. We need to embrace agentic AI because otherwise we don't stand a chance to survive what's going to come.

Andreas Welsch 26:16

Perfect. Thank you so much. And what's the book again that people should read and go

Sebastian Wernicke 26:19

buy? Yeah, the they should also buy and read your book, so definitely. But of course I'm also hoping that, yeah, they will go for Data Inspired as well. It's- Yes it's a I love your book, I think it's a very complimentary read. I think this is, this is the culture book in hiding. I think yours- very much points out all the amazing possibilities that are out there and the way to go. So read them both. Read Data Inspired- Yeah and read The Human Agentic Edge, and-

Andreas Welsch 26:43

Yes

Sebastian Wernicke 26:45

you'll be doing well.

Andreas Welsch 26:46

So some good summer reading for sure by by the lake, by the pool, wherever you are. Don't forget that. They're also available digitally and mine as an audiobook.

Sebastian Wernicke 26:54

Yeah. Yeah. But send us pictures, i, I- Yeah I lo- and you d- you love that as well, right? When you get these pictures of- Oh, yeah of people in fantastic places, they're like, "Oh, I have the book with me." Do it, yeah. We love them.

Andreas Welsch 27:04

That's a great joy for me, seeing where it pops up all, all across the world and what people are saying. So yes, Yeah do share it with us, tag us on LinkedIn, follow us and we'll certainly respond. And as you can see- Yeah makes our author's heart beat super fast.

Sebastian Wernicke 27:19

It's one of the greatest joy... if you're wondering how to bring joy to an author, it's leave a rating somewhere where other people can see it and send those pictures. Yeah. Yeah. That'll be perfect.

Andreas Welsch 27:27

E- exactly. Wonderful. Sebastian, thank you so much for sharing your expertise with us. Was great learning from you how to become more data inspired, and most of all, also what are the three key success factors for agentic AI beyond technology.

Sebastian Wernicke 27:40

Yeah. What a joy. Thank you so much for this conversation.