How Telenet Keeps AI Under Control as It Scales

Werner Meert:

Of course, at the end of the day, it's human. It's to help people. It is not that we only trust AI. At the end of the day, you can outsource everything but accountability.

Werner Meert:

There was one particular use case they showed. It was 43,000 tokens. That was the first result. And then they used the harness, and then it resulted in something around 7,000 tokens only.

Werner Meert:

And that's my personal belief, that we can accelerate benefit realization, because, at the end of the day, it should not just be tech for tech, but we need to focus on capital value creation within companies.

Werner Meert:

So we are very happy. We have onsite representation of Infosys within our team, and the individual who is acting there, is worth his value in gold. That's the message I would like to bring.

Werner Meert:

For me, it's really the whole shift from the experimental thing into the proven technology, into proven. I expect to be able to get plug-and-play elements, atomic capabilities and add them on top of the current landscape.

Vinod Ravindran:

Hello, everyone. We are at Dreamforce at San Francisco. I'm Vinod Ravindran, part of the Salesforce practice within Infosys Fluido. I'm glad to be joined here by Werner from Telenet. So Werner, if you can just quickly briefly introduce yourself and then talk about the program that you are involved in, Telenet, with the digital transformation, how it is going?

Werner Meert:

For a small introduction, so Telenet has two distinct parts. One is the B2C part, which we call residential, and the other one is B2B, more focused on business. And my personal role is I'm driving business transformation within the B2B space. And within that area, we focus on processes, people, and tools. And when it comes to tools, our focus is to unify, to cluster legacy CRM systems into one new target version. And for that, we're happy that we are one of the 15 customers worldwide that are adopting the new Commerce Cloud solution of Salesforce.

Vinod Ravindran:

What is your take in terms of some of these announcements, and how do you think these product innovations can actually be brought into your program as well?

Werner Meert:

Well, to start from the beginning, I'm really excited for everything I heard over here. But to be honest, there is natural evolution technology. And we see that, especially with Salesforce, they build up a lot of capabilities from the ground up. And today we can talk about an ecosystem containing a lot of metadata. What I'm hearing now with the whole AI attention is that we finally have a solution that can accelerate unlocking the value of all the work that was done historically. And I think that's a really very strong point, because we can move much faster. We don't have to do things ourselves as human. The boring lookup type of things and analyzing. I've seen nice use cases where salespeople can go in and say, "Show me my top three potentials to close my gap," and it's done in minutes. However, there's one danger, and that's also a very important message that I captured that we need to watch out for, is we need to handle AI with care. So that's one thing. Secondly, there was a nice presentation on the AI harness, which will give more context to using AI. And I'm also concerned that we are over-utilizing AI right now, and that reflects in the number of tokens being consumed. And with this AI harness that has been presented, which is available by Salesforce, I see that we had a significant reduction. But there's one specific action item, what you say, is that a lot of people doing experiments with AI, they use their own login and credentials. And that's a dangerous topic. You really need to encapsulate the AI agent which is working for you to be clear on the purpose, what do you need to control it. So it is not because a sales rep has access to everything, that his digital twin also needs access to everything. You might have to focus exactly to reduce the technology power that you need, talking about the tokens again, to have a really focused, context-driven, query type of logic. At the end, you need to see your AI agents as just an extension of your current workforce, and you need to apply proper user access management logic to that as well, and manage it accordingly.

Vinod Ravindran:

Obviously, over time, things have evolved, probably over the last three, four years, there is more of AI in the SDLC life cycle, the product innovations, everything is coming up. So in the program that you're currently in, how do you see AI playing its part? And over the years, how do you see it has kind of transitioned from the way the programs were run when you had similar roles in other organizations to what you are doing currently?

Werner Meert:

Okay. Well, talking about the short term, so our team is, let's say, using a lot of traditional way of software development lifecycle management. And we have seen some hiccups, things that were not going as smooth, as fast as we hoped for. And one of the key problems was that we had overlaps of requirements. So typically, when you have requirement in traditional way of working, the business analyst will come up with a requirement. They will define it, and once it's defined, they will hand it over to the function analyst. And the function analyst will take it from a functional perspective, and they will take the requirement and translate it into a functional design. But with the AI, what we're seeing is both the business analyst and the function analyst, they're now starting with the same knowledge base. It's like they have direct collaboration with your team, because they're doing exactly the same. We have the business analysts, you have the function analyst, and they work seamlessly together, and we have the same data source to start from. So that's where AI is helping both teams and the interaction is going really, really good. So that's one part. The second part is that when epics are being created, that we ask AI to check on incompleteness, unclarities, redundancies, overlaps, just to avoid that when we hand it over to your implementation team, that those people need to say, "Oh, what is happening here? That's not correct," and send it back and over, and that's where you lose time. So AI is helping us to sort of harness, to do a quality assurance of the epics before we hand it over for the true implementation. And last but not least, what you get for free is the creation of test scripts that you can use later in the game. We see a lot of saving of time. So that's for me an important part. That's the whole SDLC part. Our business value is, of course, documented within the overarching business case, and this is not new in our telco industry. You would like to increase your revenue, you would like to reduce churn, and you would like to act more efficient, that your time to market is faster. That's not Telenet specific. That's a common industry standard within telco and other companies as well. And AI can just help us to accelerate that because just the example I gave you earlier, that the sales rep is able to go in a couple of seconds to his core opportunities, thanks to AI, that will increase his selling power and his results as a sales rep. Secondly, talking about efficiency, time to markets. We will be able to launch new products much faster, thanks to this whole AI technology as well. Because we can learn from what we have, and we can accelerate thanks to the speed of the AI engines.

Vinod Ravindran:

Talking about the partnership itself, Werner, I think obviously we have been engaged on this program for more than a year. But if you look at the partnership, if you can probably think about a scenario where things are not happening properly, can you just talk through how we as Infosys work along with your team and then kind of resolve the situation? Any instances that you can really think of?

Werner Meert:

Well, starting with the team. So we are very happy. We have onsite representation of Infosys within our team, and the individual who is acting there, is worth his value in gold. That's the message I would like to bring. So he's really great guy, brilliant guy, and very collaborative as well. And then secondly, we have also your account manager who is seeing me personally every week. And our CTO every month. So there's a very close collaboration, and we're very open and transparent to name and to identify where we have areas for improvement. It is not a blame and shame exercise. It's a true partnership where we say, "Listen, we see things, we observe things, and we see room for improvement. We get suggestions from our end, we get suggestions from your end, we get even suggestions from outside, the two of us." But there is absolutely an openness to look into that, to evaluate, and to focus on resolving. And maybe you can't resolve it right now, but at least there's always a plan. A plan that says, "There's observation that we're going to do about it," and then we have a tracking mechanism. And I'm very happy about that. And also proud of this partnership because we see that this is working really well.

Vinod Ravindran:

Okay, great. Thanks, Werner. I think if I may ask a final question, I think the theme for Dreamforce from an Infosys perspective, we are talking about less AI work and more AI that actually works. So when you hear that, what does it translate for you?

Werner Meert:

For me, it's really the whole shift from the experimental thing into the proven technology, into proven. I expect to be able to get plug-and-play elements. Atomic capabilities, and add them on top of the current landscape. And I count on your team and your implementation power to be able to create an implementation which allows us to have this open plug-in logic as well. That's really important. So next to that, AI for me is really what we call a co-pilot and digital twin for the workforce. So to do the boring work, but we need to make sure that we keep it controlled. And we hope and expect that Infosys, you and your team, will evolve in the same way, and we can continue to inspire each other and create some enthusiasm to focus on the real work that we can do together.

Vinod Ravindran:

Sure. Thanks, Werner. Thanks for making time. I know it's a busy Dreamforce, but then you found time for Infosys, and you are here with us. Thanks for sharing your perspective, and we look forward to working more closely with you in the next couple of years.

Werner Meert:

Okay. Same from my side. Thanks a lot.