Clarence Reynolds, TelecomTV (00:07):
Autonomous networks offers significant benefits to telcos. But where are we on the automation journey here to discuss our Manish Mangal global business, head of 5G and network services at Tech Mahindra, Tomek, Gerszberg, head of Axiata Group Future Networks, and Tammy Whyman, global head of telecom partnerships at AWS. Thank you all for joining me today. Thank you. In our industry, lots of people have definitions of what automation means. What does it mean to you? Start with you Manish.

Manish Mangal, Tech Mahindra (00:41):
Oh, fantastic. This is the new word that we have. Of course the industry has been talking about for the last year, year and a half, about what does it mean to be autonomous network. And this is not new from a word perspective, but our vision towards what does network need to look like it from an end-to-end life cycle from planning, design, deployment and operations and all the pieces that we can put in place to allow network to run by itself. I mean at the end of the day, we have a situation in our industry where the new technologies are coming in faster pace than we can get. The skillset skillsets developed and the innovation that we require to enable the services for our customers need to be enabled at the faster pace as well. So the only way to do it at the cost-effective nature is to drive this autonomous network journey that allows us to really create a full life cycle of a network, which includes actually all the different users of AI and gen AI into the embedded parts of the network lifecycle.

(01:39)
And to me, the autonomous network, the end state for that and TM Forum has defined the stages of the autonomous network. So that's a very great blueprint for us to be able to follow. But how do you bring to life is really dependent upon how we create the technology solutions that makes the end game come to life, which basically means that you can auto deploy, auto configure, auto provision, auto heal, auto operate, and you have a closed loop automation that basically requires the least amount of human intervention when it talks about the end-to-end life cycle of the network. So to me that's a vision, that's the future and I think there is a long way to go before we achieve it, but we are taking all different steps. And from tech Mahinda perspective, we have put a very detailed playbook in alignment with what TM forum is defined in the different stages of the autonomous network and in partnership with AWS to really create that sort of the roadmap that allows a operator to get to the autonomous network.

Tammy Whyman, AWS (02:40):
From my perspective, I think it's important that we're witnessing a step change in the industry now we've have automated networks for some time and now we're starting to talk about autonomous networks and I think there's still some variations around what that means, what the benefits are, but really for me, when you get into autonomous networks, it is about the ability to learn. The network is able to understand the context, make a correction and learn from that so that it does not repeat the same thing again. And that's where you start as network operators. You feel like you might be losing the control of this. And so I think there's a culture shift that needs to happen at the same time, but the technology is in place to have autonomous networks and we're starting to see some early implementations of this.

Tomek Gerszberg, Axiata Group Future Networks (03:33):
Yeah, actually I don't have much to add because I am just in this opinion that autonomous network needs to be, needs to be aware of the services of the users on the network. It needs to be able to change itself, optimize how to heal and so on. So this is sense. Yeah, I agree. This is also a big mindset change because we are now moving to this kind of declarative approach that is already well known in the cloud world, but due to the telcos, because very often telcos still follow the process, they optimize the process, they automated the process, kind of like robotic process automation. But now this new approach with the declarative way cognitive networks is changing a lot. And I'm very much dreaming about autonomous four zero network in my workplace.

Clarence Reynolds, TelecomTV (04:34):
But you are where a lot of companies are right now as well. For the two of you who are maybe further along on that journey for a network to self configure, self heal, what components are necessary for a company that may want to start down that journey?

Tammy Whyman, AWS (04:55):
Yeah, well I think they're kind of basic components. So one is the ability to think and that is using gen AI to be able to put that brain behind it and give it the ability to think. But then once you have that, you need to be able to detect what is an anomaly, what is something that's business as usual, what is the scale of this? And then once it detects how quickly can it detect, that's where you get into latency issues. And I think that's been one of the major impediments on autonomous networks to date has been solving the issue for latency because milliseconds can create major catastrophes and network operations and then once you have that ability to quickly detect, then the network needs to be able to heal itself and reroute. So all of those components together really are what enable an autonomous network. And so we have the technology that's available, especially with gen AI and a lot of the large language models which are being created by some of the announcements here like telco alliance, what's bringing together network data from different operators using the open brand standards. And I think we'll be seeing more and more data that's available to help make better decisions when we're looking at the decisions that the model is making. But there's still a lot to do around the processes and the enablement of the field force.

Manish Mangal, Tech Mahindra (06:26):
If I add to that, actually Tammy one area and you touched upon that, is really the data part of it. And I think for the autonomous network to be successful, one of the things that is the biggest impediment that I have seen is really the ability for us to collect data across all the different sources in a unified way and make sure that the data is a very meaningful data so that we are able to then actually apply the learnings and the different engines on top of it. So that's the first step from a building block perspective that I would look at and say that how do I get the data lake built that is a common data lake. The second part is actually creating an observability infrastructure so that the telemetry data that comes from the infrastructure can actually feed through the data lake into the engines that, like you said, learns itself.

(07:10)
And then from there it builds variety of different models, whether they're the models to actually do more deployment efficiently or the models to really operate in service assure at the most efficient cost effective points. So I think if I kind of layer build the three different layers, the one is observability layer that allows the info, then the data that collects the data in a very cohesive way. And then the AI models and gen AI models that sit on top of it that actually are purpose built to use the data to allow for a specific type of use cases. And these thing can actually mature into a variety of innovative situations because no single provider in the world can actually innovate across all the life cycles. So once you actually create the marketplace and ability for anybody to come and innovate and build different types of specific use cases that can actually happen when the rest of the plumbing, like the data lake and the telemetry with observability in place. So if I layer into three things, I think those three things are the fundamental building blocks to really achieve the autonomous network

Tomek Gerszberg, Axiata Group Future Networks (08:08):
Mission. I'm not so much concerned about this part because the telcos, they have a lot of data, they see a lot of data. Over the years we've been investing in observability and this kind of businesses already, I would say well established. We can monitor every level of the network and in fact we are throwing more data away than we are really keeping. One of the reasons is because at that moment we can't really get use of those data. And I hope you know that the gen AI will help us to filter out what is important and how to turn it into the knowledge and into actionable tasks. My concern is rather related to the ability to execute because we may have the knowledge about the network status, root cause analysis for any topics, but how to execute it if you have a lot of legacy that simply does not tuck in your language. So actually the right path for the evolution of the networks to get benefit of the technologies. As you mentioned, it's like the technologies are just in our hands, we could use it, but how to do it in the way that those networks and this equipment that is still now in use and has not built to be automated or it has been not built to be a part of the autonomous organism. This is the challenge.

Clarence Reynolds, TelecomTV (09:51):
And along those lines, on the other end of the spectrum, what strategies are you using to deal with partners or providers that are coming to you with different kind of strategies or different products that then you have to try to create this end-to-end ecosystem? What is the strategy for getting through that?

Tammy Whyman, AWS (10:12):
I think you present a really realistic case, and so it's really good to understand what's happening in the life of someone who's in the trenches every day. And when I think of well, what are the parallels, what I think it starts with getting that legacy into an environment where you can make use of the data and the automation. So for example, we're finding that and networks are difficult, that's why there's still the environment that has resisted moving into the cloud to a certain aspect or in certain conditions. So we're finding that operators are moving, for example, their disaster recovery onto the cloud because it's less invasive. You don't have to undo what you already have. And so that could be one piece. So then one process of disaster recovery would start being in that piece and then you might move to your IMS system. So we're finding that it's not all one greenfield network.

(11:07)
That's rare the cases that appear, but really looking at the brownfield and how slices of that network are being able to be moved into the cloud. So then you start building this environment where certain processes can be automated. I think for me as well, it's also the promise of having an interface which can be interacting with your operators in more of a natural language search ability that's more friendly. So it just helps create that information flow in a more human way and it helps them make decisions more quickly. So it's kind of a balance right between the old and the new and there's a path and it's a

Tomek Gerszberg, Axiata Group Future Networks (11:45):
Process. It'll be built step by step. Yeah, that's obvious because it'll start with the use cases probably that are now the most urgent ones, how to reduce the energy consumption. So with those things, because on the other hand we also need to remember that networks are very physical. There is a piece of fiber, optical waste, there is a piece, there is an antenna, there is a connector for this antenna, there is a corrosion somewhere. And my, so it's like there are a lot of things that always will be physical and there will be always kind of manual, but a lot of the services like the service creation, this adaptation, this can be I believe done in the first step.

Manish Mangal, Tech Mahindra (12:33):
Yeah, no, I mean autonomous car is a great example of being still physical and how its software comes and uses it. And I think you asked the question about partners as well, and I think one of the observations that we have is that there is a lot of partners have to come together to enable this journey. No individual, either A OEM or a technology provider or a system integrator or an operator can actually enable this vision of autonomous network that we have been speaking about. At the end of the day, everybody will have to play a role from their strengths perspective. The technology innovation has to come from the technology innovators like AWS who build the capabilities and the tools. Then a system integrator will have to play their role in terms of how do you take the toolkit of a technology and help bring that to life given in the environment where there is a lot of coexistence of the existing infrastructure and the new infrastructure is always put in place.

(13:32)
So there is a lot of work that has to be done and the telecom operators have to play their role in ensuring the discipline and the direction that they give towards the vision and not let everybody disperse from that vision and keep them on course and ensuring that they're aligned to their business goals and delivering for the customers today as well as tomorrow. So as we all look at the different components of the industry and all of us have to come together now, good thing though is that the end game of achieving the vision of autonomous network is better networks a cheaper cost, which is a great thing for the customers and for our industry. So from that perspective, I think there is a strong ecosystem in play and just from tech manager's perspective, we have a very strong partnership ecosystem and with AWS that we are building towards enabling that mission.

Tammy Whyman, AWS (14:17):
Yeah, I think that it's a really good point about the ecosystem and I see in the industry we're moving more and more towards ecosystems where we're coming together, but particularly in this space or even to the point where we're seeing the telcos themselves come together to try to solve, so as I mentioned before, the global telco alliance where we have different operators coming to bring their data together to make a stronger, more accurate LLM initially for customer experience, but also using network data for that customer experience. And then you also see a lot around network APIs at this show and it's telcos that are opening up their networks to developers so that they can probe and create applications to help make networks run more efficiently, to have better inspection of them. And truly, there's a lot of innovation that's coming from the outside world of developers into the networks.

Tomek Gerszberg, Axiata Group Future Networks (15:12):
Yes, this is the pressure actually that we need also to feel to evolve ourselves because with the pressure, hopefully there will be also money coming and the evolution towards programmable networks that we can give this experience to the developers that they also, that we can build on their creativity. This is something what we really need and for this we need to have the autonomous networks. Without that, this change will not happen. Yeah,

Manish Mangal, Tech Mahindra (15:42):
Absolutely.

Clarence Reynolds, TelecomTV (15:44):
How do you see the intersection of machine learning and AI evolving in autonomous networks or how do you wish it would evolve?

Tammy Whyman, AWS (15:56):
That's a great question. Would you like to take that?

Manish Mangal, Tech Mahindra (15:59):
It is indeed a great question and I don't know if anybody can predict yet what scale at which the AI and the Gen AI will actually evolve and play. One thing is sure for though it will have a very significant and important role to play today we are starting to see some very specific use cases that are enabling a specific problem solving. One of the things that we have brought to life is an ability for gene AI to take the data in the operations and turn that into what we call the ticket resolution advisor so that when the tickets are getting formed, people with gen AI capability can start asking questions. The incident managers can ask the question through gene AI in terms of what likely is happening in the network and what should I do. And that becomes a lot more adaptive standard operating procedures rather than rigid ones and gives a little bit more intelligence in solving the problems as you've seen in the network. But that's just a very small incremental first step. I do see that it'll take its shape three, four years from now when we'll be talking today or just like we're talking today, we will not be talking about what role AI will play. We'll be really talking about the amount of disruption the AI and gen AI has brought in in terms of the management of the end to life cycle.

Tomek Gerszberg, Axiata Group Future Networks (17:13):
I'll just add one thing that the progress, what we have seen in with Jet GPT or with other tools, it's coming because this information to be processed was available. Telcos networks are still pretty closed. We don't share the information building, let's say strong models require access to the information. Right now, let's say we can charge GPT can program because GitHub was open for years and this is the knowledge that we share. I believe also building open ecosystems in the telco and open access to some of the data can accelerate this progress very much because like, okay, consumer data, this is one thing, we don't need to share it, but the information, how the network elements behave, but it's nothing competitive behind it. This is kind of common I would say if we can share it for the common benefit

Clarence Reynolds, TelecomTV (18:18):
Very well.

Tammy Whyman, AWS (18:18):
And then in the telco industry, look, we're a year, a year and a half into the gen AI revolution here. And I think you'll see the telco industry continue to be very productive with the relationship with the customer. So now we're seeing most of the gen AI use cases to be more internal, whether it's employee facing. And then with networks, certainly the challenge, as I mentioned before is around the latency. So it's a difficult use case, although it is internal facing a lot of data, a lot of systems to bring together and it needs to be real time. I think we'll see a lot of evolution towards autonomous networks. And I would say eventually we'll be seeing a lot more consumer application, so digital human interaction and retail environments. And that will be a stateful interaction that you bring with you as you have an interaction with your provider on the phone, whether it be in a digital environment. So I think you'll find that it becomes more pervasive with the consumer.

Clarence Reynolds, TelecomTV (19:25):
Well, thank you very much for your insights. This is an exciting time in our industry and you all illustrate that. So thank you very much for being here. Thank

Tomek Gerszberg, Axiata Group Future Networks (19:33):
You. Thank you. Thank you.