Guy Daniels, TelecomTV (00:12):
Hello, you are watching the Cloud Native Telco Summit. I'm Guy Daniels. Telcos are endlessly looking for new ways to reduce their costs and improve automation. Using cloud infrastructure has enabled some economies of scale, but there are still high hopes that AI can deliver even greater savings and innovations. AI relies on having huge amounts of data, which telcos often have, but what can telcos do if that data is difficult for AI to access? Well, joining me to discuss how legacy system data can be unlocked and support wider AI and cloud Native telecom business operations are Gary McDonald, VP Product Management at Wavelo, and Robert Curran, Consulting Analyst with Appledore Research. Well, good to see you both, and thanks so much for joining this year's Summit. My first question to you both is what is the relationship between cloud native AI and data? Gary, can you tell us more about this relationship?

Gary McDonald, Wavelo (01:22):
Yeah, for sure. First off, thanks for having us, guy. I think it really is a symbiotic relationship, right? Much of the telecom systems and data for the most part already live in the cloud to a degree and to power the development of AI applications and the development of AI models, you really need the large volume of high quality data and the ability to handle the spikes in the computing and processing of that data, whether you're training AI model or just simply drawing inferences from the data. And I think this is really where cloud native infrastructure is more suitable, right? It supports the elastic scaling up and down of compute and storage resources on demand for these models. It avoids the costly provisioning of those resources to meet the peak demands for capacity and processing that data for ai, and also it allows for containerization and distribution of the data and the models that are built out of AI so that they can be accessed in pretty much any environment at any time anywhere.

Guy Daniels, TelecomTV (02:31):
Thanks Gary and Roberts, as Gary was saying, there is this relationship, isn't it, between cloud native AI and data?

Robert Curran, Appledore Research (02:37):
Yeah, very much guy. Ultimately, it's really all about speed and flexibility and ability to provide more intelligent responses to customer and network situations in as real time as possible. And that's really the trio that we have here. As Gary's indicated, cloud native software on cloud infrastructure, it provides that flexibility and that adaptability, but it's not just about where you're running workloads, it's also about the whole life cycle. And so what we're seeing now is the awareness of a combination of cloud native software, the need for AI that can be distributed to where it's needed, both from a training and an inference point of view, and then feeding that with data. That's why you get this kind of trio of technologies and capabilities absolutely essential to achieve that level of flexibility and real time. This if you like, for the kind of customer situations that we have today.

Guy Daniels, TelecomTV (03:32):
Great, thanks Robert. And Robert, let me build on this. I'd like to ask you, how are operators looking to deliver improved experiences for their customers by using ai? What data do they need to do this?

Robert Curran, Appledore Research (03:45):
Yeah, I think where we are today, we are definitely looking at a future that is AI enabled and even into agentic common reference point here. The issue with that is the way in which data is scattered. Customers want and expect a level of personalization, a degree of immediacy, and that barrier, that bar is constantly being reset by all the other providers of digital services. And so that means bringing together data not only past history for a given customer or a given service, but also network information, weather data, segment information, all kinds of additional context if you like, for a customer to make sure that at the point of interaction you get a very well-informed, very responsive, very intelligent interaction with a customer. That's really what we're aiming to do, what telcos are aiming to do. That means bringing all that data together. Gen AI is a critical capability within that mix as well. And guess what? Gen AI relies on bringing disparate data together at a point in time to be able to make some of those informed decisions guide through a range of different possibilities for a customer and arrive at a good resolution or a good offering in real time. Those are, I think, primarily the ways that tugs are looking to make this experience for customers just better, much richer, much more informed, much more personalized, but using the vast amount of data that can be in a customer's context.

Guy Daniels, TelecomTV (05:19):
Great. Thanks very much Robert. And Gary, what are operators doing here? What data do they need?

Gary McDonald, Wavelo (05:24):
Yeah, I mean obviously they need a lot of high quality, large volumes of high quality data that is reliable, is accessible and is really relevant to the customer experience or the operations or the business process that the operator is trying to manage. And so Robert touched on the personalization aspect. So having the right information at the right time that can really enrich the personalization and tailoring of the operator service offering to the customer is important. And it's really about the context, right? So looking at the various business processes that an operator needs to support if they're looking to apply greater intelligence and information and insight throughout their business process is what is the relevant information they need at that time in order to be able to enrich that process and offer customers a better experience by better operating their business, the network, and ultimately tailoring that experience to the end customer.

Guy Daniels, TelecomTV (06:30):
Thanks very much Gary. And Gary, you mentioned there that operators need access to a lot of high quality, accessible and relevant data. So what sort of problems do they face in trying to tap into legacy system data for ai?

Gary McDonald, Wavelo (06:47):
Yeah, I mean, when you think about the architecture of legacy systems, it's very long monolithic, very siloed access to data across those applications and systems is highly constrained. As a result, you're dealing with multiple vendor systems which may offer different data schemas, the data stored in different locations, the data might be sometimes proprietary. And so when we're looking at trying to get access to that data to inform these AI models, it's difficult to really pull that data in and unify it in a common repository and a common schema that these AI models can access to apply to the specific operation. There's all sorts of data inconsistencies, duplication in the data and even the process to extract that data whether, and sorry, really it's a mismatch in the extraction of that data between the batch extraction process that happened, whether that be ETL to the real time demands of the business and customers to support the AI model development. So again, you've got this really highly constrained, highly fragmented, sometimes incomplete and inconsistent data foundation. And so while the operators are sitting on all this wealth of valuable customer insights and system data and information utilization of that data support their business operations and AI in particular is stifled as a result.

Guy Daniels, TelecomTV (08:20):
Thanks Gary. So Robert, it's very difficult this, but at the same time it's essential.

Robert Curran, Appledore Research (08:26):
Yeah, I think the first, Gary touched on, I think all the key points there, but reiterate the point about the amount of data that telcos have. That's whatever you want to do with it, whether you want to store it or you want to move it around or process it, that accumulation of data itself is a huge problem. And what we've seen in the past is operators looking to run things like creation of data lakes and so on, which is fine as far as it goes, but that can often be a very, very difficult, very expensive solution to the problem because it takes a long time, as Gary's indicated, trying to get a single model to make sense of all of that customer data and related data is very complicated, expensive, and doesn't necessarily lead you to the right conclusion. So the getting access to the data in time, being able to bring it to where you need it in order to make some of those informed decisions in real time, that's another challenge.

(09:20):
Just accumulating all the data into one place isn't enough on its own. So it's a mix of different problems that we're seeing here. And at the same time we're seeing this rising expectation, this is, we're not just in the analytics space anymore, we're not just looking at customer segments that are tens of thousands. We're trying to get to a model, whereas a customer segment of one essentially. And that's a different order of magnitude of problem in terms of thinking about data. It's not just the data, it is the context. It is the location and the ability to bring it to different processes that might need it. It might not be our processes, it might not be the processes of the systems where the data resides, but that's what we mean when we talk about having agents being able to reach out and tap data that's available to them to be able to do something that's useful and highly personalized for a customer.

Guy Daniels, TelecomTV (10:07):
Okay, great. Thanks very much, Robert. So Gary, lemme turn it to you. What is Wavelo's approach to addressing the issues here?

Gary McDonald, Wavelo (10:14):
Yeah, definitely a bit about Wavelo first. So we are the telecom solutions arm of two cows, which for over two decades, two cows has been building and scaling internet infrastructure, powering connectivity solutions globally. And so as the telecom solutions arm wave load, our mission really is to modernize the telecom stack with a cloud native event driven, modular composable architecture that removes a lot of the complexities and barriers to innovation that carriers have without having to disrupt their existing operations. So we're actually in the process right now of launching our figure data campaign, and really what that's about is enabling operators to employ a more data forward approach to AI and systems orchestration. And so what that means is being able to connect to the data where it lives in the cloud, various systems and applications, connect to that seamlessly across the fragmented systems and silos, convert that raw data from the legacy systems into more of an event driven format that offers consistency and reliability and it's more ready for use across a modern platform that the operators are looking to deploy. So accessing the data, transforming, translating that data and making it available to realtime decisioning to support the training of AI models and from a more static approach to more of a live data feed. And that then powers AI insights automation and innovation.

Guy Daniels, TelecomTV (12:04):
Okay. That's interesting. Thanks Gary. And can you elaborate a little bit on as to the primary benefits of the approach that you've taken?

Gary McDonald, Wavelo (12:13):
Yeah, I mean the intent is really to help operators innovate their services, launch new services that take into consideration the constraints of the legacy systems that they're operating, enable them to work with those legacy systems to handle the real-time demands of the business and of customers and help 'em launch new revenue models that are more dynamic and responsive to the market, all underpinned by a cloud native ai, native orchestration architecture. And so we can really meet the customer where they are in terms of their legacy system modernization. We can help them along their journey to really kind of enrich their existing business processes and systems with a more intelligent and automated AI native solution using this kind of event driven cloud native solution. And as I mentioned a little bit earlier, because of just the nature of event-driven architecture, we can do this without disruption to the operator's existing systems and operations.

Guy Daniels, TelecomTV (13:19):
Okay, thanks so much Gary, and let's talk now about event-driven architectures in telecom. Robert, let's start with you. Is this a new development or they're now in more common use?

Robert Curran, Appledore Research (13:31):
Yeah, it's really interesting guy. Event driven architectures are something that's relatively new in telecom, but incredibly well proven in other industries that have been digitizing, digitalizing for some time, particularly finance, banking, retail plus your typical hyperscalers, the Netflixes and so of this world, we are moving from a time when we use so-called service oriented architectures, SOA, that was pretty commonplace for the last 20 years or so as this sort go-to architectural choice. But that was really created in a sort of pre-cloud, certainly pre-digital first context that we have today. So event-driven architectures are certainly becoming much more relevant and telecom is just behind the curve basically. The examples from other industries are very, very strong. Just again, the question of scale comes into play here. Telecom likes to think it's an industry that has a lot going on. Some of these companies are processing a million events a second, a second, it was phenomenally fast.

(14:31):
And they're using that to feed data pipelines that in turn feed ai that's looking for all kinds of things simultaneously, not only for things like dynamic pricing and choices and preferences and options, but also things like trying to identify fraud and so on. There's a lot of different ways to use this architecture, these principles to solve problems. And we are seeing it becoming more, certainly more a feature, more relevant in telecom, more examples. Wave is a good example today of where that's being applied in the real telco world, certainly more relevant and that I'd say working backwards from the customer back from that market of one is certainly drawing more attention to what event-driven architectures can really do.

Guy Daniels, TelecomTV (15:14):
Yeah, thanks Robert. This is absolutely fascinating. Gary, can you elaborate more on event-driven architectures and what they can do for telecoms?

Gary McDonald, Wavelo (15:21):
Yeah, I think Robert touched on it. I mean, telecoms are perhaps a little bit late to the party and not unsurprisingly, there's a lot of risk involved in terms of the size of the operations that operators are managing. So there's perhaps a bit trepidation about employing a new type of architecture rather than the traditional kind of peer-to-peer API type architecture. But as Robert mentioned, there's a number of high profile companies in the digital domain that have employed this and use this as part of their core operations, whether that be Netflix from the recommendation engine and monitoring Uber from real-time analytics, pricing, Airbnb from event-driven messaging in the booking operation. And so we're starting to see more and more employment of event driven architecture in the telecom sector for use cases like real-time network management or more AI based customer experience management. But just I think given the scale and the size of the applications and services that the operators are running, again, there's a bit of that risk aversion, but it doesn't have to be that way. As I mentioned earlier, the good thing about event driven architecture is you can effectively run a different or parallel process system to your legacy stack without interrupting the ongoing operations or system processing that's happening in your legacy stack. So I expect based on some of the interactions we've had, there's a lot of interest how to go about deploying this in the telecom space, and I expect we'll see more and more on that in the years to come.

Guy Daniels, TelecomTV (17:03):
So final question then to you Gary, and wrapping up our discussion. What does this approach mean for telcos in terms of their operational efficiency?

Gary McDonald, Wavelo (17:11):
Yeah, I mean, you can immediately go to the cost equation of that, whether it's offering more reliability, greater performance, but ultimately all of that impacts the customer at the end of the day and translates into loyalty and perceived value on the network. So certainly operators can and are using AI to help apply that intelligence to their existing operations. I mentioned the network in terms of network performance, network management, self-healing of the network, traffic shaping, all those kinds of things. We're starting to see more of that in the customer experience domain as an augmented care agent experience, offering a more tailored service to their end customer, even with applications like Next best offer, powering sales and marketing with the right intelligence about the particular customer base that they're pursuing, how they can put the right offer at the right time in front of that customer. So these things all amount to being much more efficient within the telco operation. And of course that translates into cost reduction or cost optimization, and of course the cost optimization would be passed on to the end customer at the end of the day, and that adds to the perceived value they're getting from the operator.

Guy Daniels, TelecomTV (18:38):
Great. Well, unfortunately we must leave it there for now, Gary. And Robert, great talking with you and thanks so much for sharing your views with us today.