Ray Le Maistre, TelecomTV (00:05):
So we are in London at FutureNet World 2025. I'm here with Aaron Partouche. He is the Innovation Director at Colt Technology Services. Aaron, great to see you again. Thanks very much for joining us here.

Aaron Partouche, Colt Technology Services (00:17):
Hi Ray. Thank you for invitation.

Ray Le Maistre, TelecomTV (00:19):
Thank you for joining us. So this event, this conference, it's largely about AI and automation. How would you describe PO's AI strategy in relation to its network operations?

Aaron Partouche, Colt Technology Services (00:33):
Wide topic, wide topic. There are several part of, I mean AI that we could mention related to network operations. Firstly, trying to break data silos in the network. So we have a long-term journey on modernizing our networks. And it's important because hold legacy network architecture, were not built with data in mind. So we build, in 2017 our segment routing MPLS network created a source of truth with data collected from the network network topology segment. We are also including now power consumption to provide greeners path to our customer. So it's one way. So it help us create this foundation. But after we know that it's just the beginning of the journey, we are already looking at how we could expand this data to build a smarter network, a more sustainable network. And at the end data will create the value and AI will accelerate it. So one piece is the data. The other piece is also creating an ecosystem, a collaboration ecosystem. NAS is an important topic as you probably know. We are more and more looking at building a federation of operator of based on API.

Ray Le Maistre, TelecomTV (02:09):
Okay.

Aaron Partouche, Colt Technology Services (02:10):
We did also this exploration of connecting the RPI we are defining in math with camera API from GSMA and we did this pilot during mobile watch progress with a range. It's also a way to show that API will not be just horizonal, but vertical could propose options to developers, to ai, adjuncts and developers. So many piece, this is very few of them.

Ray Le Maistre, TelecomTV (02:45):
Okay. Could you talk a bit more about your data management strategy? You talked there about the importance of being able to collect the data from the network, et cetera, but that's just the starting point. How are you working with that data? Do you need to take a subset of it? Do you need to politician it? How do you work with that data?

Aaron Partouche, Colt Technology Services (03:08):
So yeah, very good point and difficult to anticipate. To be very honest, we are looking at different things. One is clearly for US enterprise dataset will be across different type of cloud platforms, private public cloud platforms and even maybe edge. So we are more and more looking at how we could define the good proposition on multi-cloud. And this is something that we are more and more focusing on this data in the middle. We are also looking and we are doing a pilot currently on AI inference at the edge. Again, difficult to anticipate what will be the demand from enterprise and what will be the traffic at the end, but we have to anticipate it and it's quite difficult to predict the traffic change before it was driven by OTTs. Now in the future it'll be more and more driven by AI application that will be much more predictable. But we are looking at this tool aspect.

Ray Le Maistre, TelecomTV (04:15):
I guess you can only figure this out by doing it. Are you doing that in your live network? Are you creating the copies of your network and doing that in AI and doing lots of modeling in basically, I'm trying to remember the term now for the copy, the mirror image of

Aaron Partouche, Colt Technology Services (04:38):
The network twin.

Ray Le Maistre, TelecomTV (04:39):
Yeah.

Aaron Partouche, Colt Technology Services (04:41):
So indeed. So we have built now an innovation lab in London that help us to incubate experiments of all of these new concepts, but help us also to work on something that is in parallel and will not in fact our network, our production network. So it help us, one aspect that I didn't mention is also of course network security. We all don't the security aspect of ai, but it's potentially one of the today's main request from our customer today, how we can secure the flows, how the threat will evolve in the future. So not only using AI to detect threat, but also how the threat will evolve. How can we do something? So we recently build a pilot on quantum safe that works because, and we see a real traction from customer real demand. So it's quite interesting. We are looking at how the traffic evolve, what we should do. But at the same time there are over aspect of it, but we can't guess that it's more and more requested by customers.

Ray Le Maistre, TelecomTV (06:00):
So there is real track and real thought going on with the customers around quantum safe networking.

Aaron Partouche, Colt Technology Services (06:07):
Yeah.

Ray Le Maistre, TelecomTV (06:07):
Okay.

Aaron Partouche, Colt Technology Services (06:08):
Yeah.

Ray Le Maistre, TelecomTV (06:08):
And is that just recent? Is that something that's,

Aaron Partouche, Colt Technology Services (06:11):
I think it's recent because we, it's difficult again to predict. It's all about prediction, but it's difficult to predict when quantum computer will be ready.

Ray Le Maistre, TelecomTV (06:21):
Yes.

Aaron Partouche, Colt Technology Services (06:21):
But what we know today is we can store the data, we can, this is a harvest now they could later.

Ray Le Maistre, TelecomTV (06:29):
Yes.

Aaron Partouche, Colt Technology Services (06:29):
And within maybe five years, 10 years, we don't know exactly when we could this data but potentially will be very relevant. So all the sensible industries are looking at protecting the data now, even if the quantum computer are not but here ready to get to decry this data.

Ray Le Maistre, TelecomTV (06:50):
Okay. Interesting to hear that that's really happening and that people aren't just talking about it. That's really good to hear. Just to go back to AI for the moment, is there a way, there's a lot of talk about how the use of AI can improve operational efficiency and so on so forth in network operations. Is there a way to measure the impact once you introduced AI tools? And if so, is that something that is a standardized that everybody's using or something that you've developed in-house yourself?

Aaron Partouche, Colt Technology Services (07:30):
So I have impression that I'm talking with our financial partner because indeed one of the key challenge we have today is when we are on the innovation engineering side is we know what we can do, what are the capabilities? But our financial partners ask us, okay, how can we measure,

(07:51):
How can we prove the value? And this is one of the big challenge that I think all the industry have, and particularly here in certain network world, a lot of things can happen, but we need to measure, we need to show the impact. So of course we are now looking more and more how we can measure and show impact in term of productivity reduction of dex. So reason why I was talking about data, but should go maybe beyond the network data, but we are looking at modeling financial insight plus network insight to be cost efficient and more and more cost efficient. So yes, now the way we are thinking our next generation of network is including this measurement to show the impact of ai.

Ray Le Maistre, TelecomTV (08:36):
Okay. And is that something that you feel that you have to the tools to measure this or the ways to measure that? Is this something that you are having to develop internally yourself first or is this something that your technology suppliers are able to help with?

Aaron Partouche, Colt Technology Services (08:53):
So very, very good question. More and more of our partners are proposing and pushing us. Okay. We can help you. We are still discussing it internally. It's important to be in control. This is the main point for us to be in control, to be able to control this data and not in five years, oh, we need to change, we need and we have again to have a kind of vendor type of solution. So it'll be key. So the way we control data, what we will do with the data will be quite key. So there are more and more options in the market because I think we are not the only one to see that. It's quite key. A lot of vendors are proposing options, but we try to find our way and be in control of that.

Ray Le Maistre, TelecomTV (09:42):
Okay. So that looks like a watch this space kind of scenario. Something for the whole industry to be thinking about clearly. Thanks so much for joining us here at telecom TV today and good luck with the rest of the event. Thank you.

Aaron Partouche, Colt Technology Services (09:56):
Thank you very much, Ray.