SPEAKER_00: 0:02

Good evening, Kim. Nice to have you here.

SPEAKER_01: 0:04

How are you? Hey Ben, I'm doing great.

SPEAKER_00: 0:07

And uh thanks for having me. Very happy to have you. Um you've been playing a big role in Telnet's data and AI journey. So very much looking forward to hearing those stories. But maybe let's start at the beginning of your career. When you joined um machine learning startup in 2001, back then what was it like?

SPEAKER_01: 0:28

Yeah, so uh I started my career as a data scientist when that term didn't exist yet. So, like you said, in 2001, it was with a startup. Uh, we were working for large companies, creating predictive AI models for them, but you know, it was a little bit early. Um, to illustrate that, um, we got data from these large companies, typically on a CD, uh, which we would then you know take with us to our office and then upload that data to a giant supercomputer uh where we would then run the models and then we would put the output back on another CD that we would then hand over to the customer together with a presentation. So you can imagine these were different times. And um, so yeah, for me, probably the insight is that uh if you're working in innovation, timing is everything.

SPEAKER_00: 1:19

Yep. Never heard about uh CD stories before.

SPEAKER_01: 1:22

Because it was so early. Yeah, I can imagine a lot of things around it uh didn't exist yet. So the the underlying AI technology existed uh to some extent, but a lot of it around didn't, in terms of storage and terms of computing and yeah, indeed.

SPEAKER_00: 1:37

Yeah, that's what I saw in recent papers too. That purely in terms of algorithms, it was there, but we didn't have the infrastructure to support it. Exactly. Yeah, yeah. Um later you moved into marketing at Telinet, so in the telco industry. Yeah. What attracted you to that industry and to the marketing field?

SPEAKER_01: 1:56

Yeah, so uh when working for those large companies as a consultant, I really loved the conversations that we were having with the business people and typically marketing people. And um I was really interested in what they would do with the insights that we created and to make their plans better, to make their campaigns better. And I wanted to be closer to that. And so I decided after a while, after a few years, to make the move to such a large company, and that became Telenet in 2005, first as a data analyst, but then quickly to become uh a product manager in the marketing department. And um, you know, these days, the that era that was like the golden era of product marketing. Uh, that was a time when uh Steve Jobs unveiled the iPhone, and I thought that product manager was the coolest job in the world uh at the time.

SPEAKER_00: 2:47

And during that marketing chapter, you had some standout projects.

SPEAKER_01: 2:52

Yes, um, several, but I would say that Vigo was definitely the one that stood out the most. Vigo was the first um bundle that combined everything fixed and everything mobile together in one subscription on one bill, um, which now seems obvious, but at the time uh customers were typically having uh everything fixed, internet and calling and TV with Telenet, and then mobile with Proximus. And this combined the two, uh, which made it easier for families, uh, everything on one subscription, one bill, uh, and it became a huge success, a huge commercial success. Um, so yeah, for me, that was uh let's say the the takeaway was that real commercial innovation happens when you create uh simplicity that makes things better for customers.

SPEAKER_00: 3:41

Cool. Um, eventually, though, you transitioned back to data and AI at Telanet. What motivated that move and what what is your mission in this world today?

SPEAKER_01: 3:51

Yeah, that was uh back to my first love in 2016. So after we go, after almost 10 years in product marketing, I felt like I'm I'm up for something new. Uh, that was the big data era. Uh, companies were investing a lot of money in into data and big data, and I thought maybe now the timing is right uh to go back to that. Uh, I had the chance to lead everything data and analytics in the marketing department first, and then later as of 2020, uh, for the whole company, uh, which I'm I'm still doing, and yeah, my mission is really yeah to connect the two, connect business and data and AI to make sure we get the most out of it.

SPEAKER_00: 4:33

How would you describe the overall approach to data and AI within Telinet?

SPEAKER_01: 4:40

Um, I would say, yeah, so we we often say uh uh data and AI are the fuel and the rocket that launch our business to new heights, which is to say that it's not a goal in itself, it's a means to an end, and that is to improve the business. It's really to accelerate our strategic shift uh towards becoming the leader in customer experience. Uh, and that is not a faraway dream or a future vision, it's already today's reality with the the the portfolio of uh data now products that we have that are really mature, that are actually doing that already uh yeah across the company.

SPEAKER_00: 5:18

I'm interested in that portfolio. Do you have some clear examples to make it a bit tangible? Yeah, sure.

SPEAKER_01: 5:25

Segment of one uh is a good example. Um we know for every uh customer, so for each of our two million customers, at any given time, what is the next best action for that customer? So that can be a commercial offer, that can be you know addressing a pain point before it it becomes a reason to churn. Uh it can become uh a service action like when you need a new modem. And so yeah, this is yeah, this is embedded across all of our channels from the call center to our to our app. And uh it's starting to deliver real value. Um, for example, we see significant churn uh improvements because of that. And uh yeah, so I guess uh there's real value in AI if you can connect it directly to your customer interactions.

SPEAKER_00: 6:17

Yeah. Earlier today I was browsing through the app and I noticed that Tilnet also has a chatbot. Could you elaborate on that?

SPEAKER_01: 6:25

Yeah, you spotted that well. Uh indeed, a few weeks ago we introduced our um our chatbot uh based on generative and agentic air. And it's uh it's very new. We um introduced it fairly under the radar uh because it's so new and we still want to learn, uh, but it can handle FIQs already really, really well. And um so gradually we will improve it and extend it, you know, add new capabilities. Um, it will be able to know your personal situation and related to Telenet, of course, uh, and and act on your behalf. So uh yeah, just getting started, but um yeah, super promising. Uh yeah.

SPEAKER_00: 7:08

Looking forward to further developments. How do you look back at the journey of Telenet in terms of AI?

SPEAKER_01: 7:17

Um so we've been working on AI for a while now, uh so many different years, and you can maybe split it up in predictive AI on one hand and then generative and agentic AI on the other hand. Um, and predictive AI and segment of one is a good example uh of that. Uh it's about automating millions of decisions um by predicting something. Um we know that very well. We're good at it, and we know how to do it, we know how to create value out of it. And the focus is really on exploitation. Um, so for us, it's like the it's it's like the banana plantation almost. Uh, whereas uh um generative and agentic AI is more like the jungle, it's new, it's exciting, um, it's also a little bit risky, but there's a lot of potential. And so the focus here is on exploration. So we have these first real-life deployments like uh the the chatbot that you mentioned that that we explained. Um but we're getting we're just getting started. There are many use cases uh in the pipeline as well. Um, so for us, I I guess it's key to balance really exploration and exploitation because you want to reap the benefits of your previous investments on one hand, but also make yourself ready for the future on the other hand.

SPEAKER_00: 8:39

You talk about agentic and generative AI, the field is changing very quickly. How do you handle the pace of change in that field?

SPEAKER_01: 8:50

Yes, indeed, it's uh changing uh incredibly, uh, so it's ridiculous. Uh and and indeed you need to be able to deal with that um yeah in multiple ways. Uh so on one hand, uh we want to make sure that that we spot all the opportunities. So we do a frequent um yeah AI roadmap check because of you know because there are new technical advances, which opens the door to new use cases. So we want to spot that. That's so that's one way to deal with that.

SPEAKER_00: 9:20

How frequent is that?

SPEAKER_01: 9:22

So in the past that was something every few years, and now it's something that we do every year, uh and and and also with updates in between, even. And so it's something that we do really frequently now because it's changing so fast. Um, and on the other hand, when you're building something, you have to watch out that you're not building something that is already outdated before you've introduced it. Uh, and so yeah, we need we yeah, we make sure that we work with product architectures that are flexible and modular so that we can swap in and out components as the technology evolves. Uh so um so that's uh these are a few ways in how we deal with that crazy speed.

SPEAKER_00: 10:07

Yeah, sounds like a sustainable approach. In your view, what foundations are most critical to drive both stability and also value?

SPEAKER_01: 10:18

Um, so we you need to look at it end-to-end. Uh so uh we talked about use cases, uh, but that's that's one side of it. Um so we talk about the 5P framework. Uh so the first P is potential, these are the use cases, what are current use cases and what are future use cases? So you need to have a portfolio view on that, so that's one. But next to that, there are other aspects, and and so the second P is the P of platform. So, what what type of platform choices and decisions do we need to make to realize all of those use cases now and in the future in a scalable and sustainable way. Then there's the the P of process. Um, if you really want to get full value out of AI, you need to also redesign some of your processes because just keep doing yeah, working in the same way as before with a new tool, that's not gonna cut it. And then you have the P of people, which is about the workforce. Uh so yeah, are people ready for it? Are they adopting it? Uh do they have the skills? Uh so there's a lot of uh people side uh to it. And then finally, there's the the P of potential, sorry, the P of policies and risks. Uh so how can we do all of this in a safe way, uh in a in a sustainable way, in a in a way that that follows rules uh and in a responsible way?

SPEAKER_00: 11:48

Very interesting. Today TillNet is seen as very advanced in data and AI, but that wasn't always the case. What were the biggest early challenges?

SPEAKER_01: 12:00

So, yeah, when when I joined the the data team, um I used to say we got 99 problems, but data ain't one. And that was to say that we weren't in lack of data, we had plenty of data, it was the big data era. Uh, we were storing a lot of it, but we didn't know how to create value out of it. And um, that was because we had a few problems. Um, one of them was that um it was very complex to work with data and with data teams, you know, on the business end. And I I had been there, uh uh, I had experienced that. Um, as data and data teams, they were very scattered across the company uh and siloed, not always working well with each other. So that was one problem. Uh so it was complex. Uh, secondly, there was no big focus on additional business value creation. So you were doing a lot of reporting, a lot of proof of concepts, but there wasn't a view of really how to you know go beyond that. And then the third problem was that so in Dutch they say unbekanned is onbemint. Uh so data and data people weren't really visible or well known in the company, and then you have a disadvantage if you want to you know transform the company with it.

SPEAKER_00: 13:19

Sounds like a messy and fragmented situation. How did you turn that around?

SPEAKER_01: 13:24

So messy and fragmented, but you know, there were all already uh a lot of good things, uh of course. But we did turn our turn it around by making everything data and then later also AI easy, relevant, and famous. And so we made it easy by you know bringing all of the activities together into one team so we could make one plan and speak with one voice, organize ourselves in a logical way, also logical for business, uh, with ways of working extremely close to the business, working hand in hand. So that that was uh yeah, a way to make it easy. Um and then to make it relevant, so we started by creating uh a data and AI strategy uh that is in tune with uh company strategy, with business strategy, that focuses on those areas where you can make the most impact uh and then and then identify use cases from it. And then we made it we made it um famous uh by beefing up our story telling skills. Um yeah, being able to tell the story of how we create value with it. Uh we found numerous ways to do that. Uh one one example is data stories, um, yeah, where we it's like that talks, but then for data and by data people uh to inspire the business.

SPEAKER_00: 14:46

Maybe some context for the listeners. I also work at Telenet and have worked at Telenet. And I remember data stories where there were hundreds of people actively listening, so that's a milestone too, I assume.

SPEAKER_01: 14:59

Um that's still the case, and we do it now in a bi-monthly rhythm, uh, and we still have yeah like 150 to 200 people tuning in uh real time.

SPEAKER_00: 15:11

Amazing. You've recently evolved toward the data and AI product management model. Why was that change so important?

SPEAKER_01: 15:19

So that was really important. Um, and we decided to do that in 2023 because we understood that the name of the game was to get to continuous value, uh, maximum value at scale, maximum adoption on the business. And so, yeah, when if you think back prior to 2020, um we were treating data as a request. Uh so that means that um yeah, we got questions like make me a sales dashboard, what is the churn? Looking back, a lot of reporting. As of 2020, we shifted to you know looking ahead and using data and AI in more advanced ways. Uh, in sort of we called it boost topics, boost projects. Um, it's sort of business improvement projects using data and AI to do that. Um, and you know, we covered a lot of ground, we explored a lot of opportunities, get to we got to value proof points a lot. Um, but you know, at the end of those projects, it wasn't always clear how to then proceed with that. So that's why we shifted to treating it as a product. So yeah, we looked at all of our data and AI assets and then uh organized it in into products, um, focus on on a business goal, um, making sure that there is an owner, there's ownership, there's a a roadmap, there's an adoption plan, there's training, uh, you know, and then basically it's it's owned and so that we can get to sustainable value.

SPEAKER_00: 16:48

Where does governance and quality fit into that story? Because you mentioned it earlier. Yep. How does it fit into the story of data product management?

SPEAKER_01: 16:58

Uh yeah, it needs to fit within uh within data. And uh so yeah, data governance can be an abstract topic, uh, can they can be complicated. Uh, we try to simplify it and make it work for us and to support that data product way of working. Um, so we look at it like the rules of the game. Uh so um it's like like in football. Uh so if you make a serious fall, you get a red card. Um if uh yeah the match is 90 minutes, and and so these are the rules of the game that allow you to be able to have a good game. And so for us, it's the same. Uh and for us, data governance is like the rules of the game, and so that means in our data product way of working that every data product, every AI product needs to have seven product policies in place, and policies that are about that are about ownership, and so there needs to be clear defined ownership on the data side, on the business side. That's an example. Uh, that needs to be data quality monitoring. Um, uh data needs to be easily findable. Um, there needs to be documentation and so on. And so for us, it's like the rules of the game that make sure um yeah we can get to sustainable value.

SPEAKER_00: 18:20

Cool. Let's maybe talk a bit about leadership. How would you describe your leadership style?

SPEAKER_01: 18:28

My leadership style. Um, I would say I try to lead as myself, um, trust my intuition, trust my strength, but also at the same time um try to empathize with my especially with my direct reports, so that I can understand them, so that I can you know adjust my style and my approach a bit uh to their individual needs. So that's one component I would say. Um, next to that, I also give a lot of empowerment, uh, autonomy. Um, so I'm leading mostly directors, they're super capable leaders and managers themselves. Um, so yeah, they do most of the work, but they can come to me for guidance. And then there are a few big challenges for which I need to step in. I want to step in decisively. And and for those big challenges, I take the lead, but then I quickly go into co-creation mode with them, with my direct reports. Um, so I put the big challenge in the middle. Um so yeah, and assemble the right people, and then we go over the objectives, the challenge, and also yeah, what's the desired output where we need to get to. Um, and then I follow a bit the the leverage principle, which means that for every work, for every hour that I work on it, uh, they work on it five hours or ten hours after that meeting, after that first meeting, and then they come back with uh they've worked it out, uh, and then we go over that. Uh, and then you know I can steer it in the right direction, and we do that iteratively until we get it right. So I I guess that that's a little bit that balance between giving empowerment and then stepping in when it's when it's needed. Uh yeah.

SPEAKER_00: 20:21

How important is talent management?

SPEAKER_01: 20:23

Yeah, extremely important. So we work with uh with all these technologies, but to make it work, yeah, you need the people. Uh so the people behind it are actually super important to make it work. Um, we're also in Talon at Inside, we're we're known as a talent hub, uh, where people make a lot of promotions inside of the data and AI team, but also yeah, towards other teams uh in the company. So this is something that that's happening a lot and has recently happened uh as well.

SPEAKER_00: 20:54

How do you see the short-term talent drain of what you're talking about versus the long-term opportunity of that?

SPEAKER_01: 21:02

Well, talent drain, um luckily when people leave the data team, they typically go to business teams and and take on business leadership roles. Uh so I would say it's like a a short-term local hassle for the data team to then yeah uh find the new leaders, uh the new people. Um, but it's uh let's say uh a long-term and global uh opportunity for talent as a company because if you think of it, we're putting data-driven people, people with knowledge about AI, we're putting them in leadership positions in the business. So this is also another way to fulfill our mission.

SPEAKER_00: 21:45

Yeah, and you can more easily collaborate.

SPEAKER_01: 21:48

Exactly.

SPEAKER_00: 21:49

So, what is the secret sauce for attracting and developing talent?

SPEAKER_01: 21:54

Secret sauce. I don't think we have a secret sauce. I just think we have consistent and good professional uh talent management practices and processes. Um, we have, for example, so it's a large team, uh, so more than 150. Um, but we have, for example, talent managers, talent uh leads we call them, uh dedicated to certain clusters. Um, we have these bi-weekly talent management meetings uh where we discuss uh yeah talent management. Um but I also think that I I jokingly often say it's like the law of attraction, because uh we have such good talent, and I believe that talent recognizes talent and wants to work with other great talent. So this is a dynamic that we see often that yeah, that allows us to keep attracting uh new great talent also from outside the company.

SPEAKER_00: 22:53

I can confirm it's pleasant to work with those talented people. Thank you. Looking back, what are the most important leadership lessons you've learned along the way?

SPEAKER_01: 23:04

Um I would say I would pick the period of 15 years ago. Um the manager I had at that time was really, really interested in making me better. Uh so he really believed in me more than I did myself at that time. And so that lifted me up, that made me better. And for me, that was really a game changer. I started believing in myself more, and you know, for me, that was the start of let's say the rest of my career. And so yeah, what I take away from that is that you know, as a leader, you have you have really a great power, um, you can have a lot of impact on on your people's uh careers and lives. And so, yeah, I would say that great people leaders are leaders that make others better next to delivering results to others.

SPEAKER_00: 24:02

Cool. Something to take away. Um, Kim, this has been a fascinating conversation. Thanks again for joining me today. And to everyone listening, I hope you enjoyed this episode. See you next time.

SPEAKER_01: 24:16

You have taste in a way that's meaningful to suffer people. Hello, I'm Bill Gates. I would I would recommend uh TypeScript. Yeah, it writes a lot of code for me, and usually it's slightly wrong.