Webinar: Driving Measurable Outcomes for Telecom with GenAI

Hello and thank you for attending today's webinar, Driving Measurable Outcomes for Telecom with AI, presented by Fierce Network, Microsoft and Amdocs. Before we get started, here are a few housekeeping notes. Additional resources are available in the handouts window and to learn more about our speakers, check out the speaker bios window. If you close one of the windows and need to reopen it, click its name in the top navigation bar. This webinar is being recorded and will be available on demand within 24hrs. And finally, if you have any questions for our speakers, please enter them in the submit questions window at any time. And now I'd like to introduce the moderator for this webinar, Steve Saunders, executive producer at Fierce Network TV. Steve, please go ahead. Hi, everybody, and welcome to today's webinar, which is on driving measurable outcomes for telecom with GenAI, a fascinating topic which is incredibly important to all service providers at the moment, and I'm delighted that I have two real experts to help me discuss this topic today, Ophir Daniel from Amdocs and Gansu Adi Narayanan from Microsoft. Welcome, gentlemen. Hello, everyone. My name is Gansu Adi Narayanan. I'm a partner technology strategist partner CTO with Microsoft. And in this role, I support many software development partners like Amdocs in building innovative solutions on AI and data. So moving on to this topic, I would like to kind of start this conversation with how we Microsoft look at the AI and the transformation at telco industry. Right? So we see it as a kind of a combination of four pillars. Right? The the first to start with elevating the customer experiences for the customer care, sales agents, and marketing agents. The the basic idea here is to kind of deliver frictionless, personalized customer experience to the end user with the infusion of AI. Helps scaling the customer care experience at scale. And one of the great example I can give you is the customer engagement platform that we developed with Amdocs which is powered by our core tools like Dynamic 365, Copilot and AI platform. And moving on the second pillar that we are that we I would like to highlight is optimizing the business process and operations. Right? And this involves reducing the overall cost for telcos and increasing the agility and accelerating time to market. So typically this involves migrating existing workloads to the cloud and building cloud native solutions so that the and these applications can give a faster time to market as well as can be embedded with AI powered smart ops. And moving on, the next pillar that we focus on is modernizing the network. Right? Of course, network plays a key role in any telco industry and this pillar involved enhancing the network reliability, security and scalability by infusing AI technologies. For example, this might involve bringing the various heterogeneous data network data sources to a single place and then building an AI agent on top of it so that you'd be the the proactive and predictive maintenance use cases can be unleashed on top of it. Right? And the last pillar involves building innovative solutions or new revenue growth for telco industries. Right? So this might involve looking at the BSS, OSS business process from an AI aspect and then bringing newer use cases and as well as revenue model. Again, a classical example I can give you is the the Amaze platform that we built together, with Amdocs, which kind of it's a GenAI platform which provides an semantic model on telco industry answering, telco specific questions in a natural language way. Moving on. So we we at Microsoft look at agentic AI as the next wave of bigger telco transformation. Right? And I think this involves, I think, two key we are kind of anchoring this as a kind of a two key patterns that will emerge as out of it. Right? One is we anticipate that there'll be an AI assistant for every user. AI assistant like our co pilot, which is supported for every user, whether that user is a sales agent, customer agents, marketing agent, even developer, IT ops, HR. There'll be a AI assistant specifically tailored for them to improve the overall productivity, to provide personal experience so that their productivity is improved many folds basically. Right? And similarly, we also anticipate that there'll be an AI agent for every business process, especially for telco industries, for every business process like whether it is a billing invoice generation or customer onboarding process or predictive maintenance in the network case scenario, we anticipate that every telco business process will be associated with an agent which will help them in, again, accelerating the overall time to market as well as improving the overall productivity of the organization. And another key area where we see the innovation happening is that the UI. Right? So we see as the the user interface layer being one of the more one of the more innovative solutions that that make Ping An for telco industry. We see that AI could be the new AI the AI could be the new UI for the entire telco industries. By by that, what I mean is that with this new AI capabilities, telco industries or even any specific industry, right, will be going away from a complex screen based navigations and then more of a personalized experience where the data and the actions are bring to the user and they will be able to do it in a much more faster and efficient way. And combined with an with the tools like natural language interface, the users would be able to complete any command or a task with the natural language interface that kind of opens up and democratizes the entire UI process to many users actually. So with AI as the new UI and Copilot in the forefront, we see that the the the whole the telco industry use cases like BSS and OSS could be transformed. And we expect that in the coming years, there'll be a new paradigm shift in this area with the AI driving the UI factor. And with all this combined, we see this as a kind of a vision for the telco telecom industry. Right? At the bottom of the layer, you can see that the traditional system of telecom where the the topology, external data sources, and infrastructure plays a role here, the the heterogeneous system. And on top of that, situate lines of business application like a BSS, OSS network system. Right? And in the the middle layer is the crucial one where we bring our tools like fabric, which is an unified data platform, which can help in integrating these data sources in real time so that these data can be used to power these AI agents. And also there'll be a that we also see that ARC plays Microsoft ARC plays a crucial role in this, right? In this ARC will play as a backbone for integrating the infrastructure across private cloud, public cloud, on prem as well as the edge, the systems of telco, Right? So once those foundation layer and the integration layers in play in place, if you look at the left corner of this of this diagram and we see that that's the AI platform that we foresee will sit on top of this telco model. To first to start with the Copilot studio, which is a low code, no code environment for building quickly the Copilot for productivity. And we have the Azure Foundry, which is a pro code environment for building production ready AI generative app generative applications. Right? And also, we as part of AI foundry, Microsoft support more than eighteen eighteen hundred plus models across open AI, open open source models, right, where where the customers are can choose this model to build their solution. And, also, all this is powered by Microsoft development tools like Visual Studio, GitHub so that the key requirements of building this AI agents like DevOps, ALOps could be addressed basically. And this will be powering this the with this, you can generate you can build copilot and agentic model which caters to most of the users of telco industry. And moving on, this slide shows some of the key case studies are that that we are already working with many telecom customers in in making sure that they embrace generative AI, and then they are able to improve their productivity and scalability. Right? So you can see, the customers that like Vodafone, they're able to improve twelve percent year on year, the the reduction of calls and so on and so forth. The New Zealand one environment and the PLDT, which are able to reduce the overall call time from fifty percent. Right? So there is many examples. Such examples are available, and we'd be able to share more details around that. With that, I transfer this to Afir to present the Amdocs part of the session. Thank you. Thank you, Gansu. We'll move on to my part. First of all, a bit of an introduction on Amdocs, which is of course a global company. We work together with Microsoft for many years now, We provide services in the essence of products and services for communication service providers all over the world, especially now in the agentic era, there has been such a great momentum happening. And we are actually seeing a great shift as Gansu mentioned, where the CSPs are not staying indifferent to this technology and are adopting it as part of their daily life routines. And that means that they're looking to find the promise of GenAI based on the fact that data driven organizations have been working with AI analytics, machine learning, and deep learning for quite some time now. In order to make the best of the generic technology, you will need to make sure that you have the right foundations for that. And that of course is work that has needed to be done or needs to be done on a daily basis by harnessing data, as Gansu mentioned, to a data fabric, which you can start building on top of it and extract the best of the generative AI. What we have done is we've actually surveyed many CSPs all over the world to find out how AI agents being deployed on their ecosystem will affect the customer experience. And there is a very big impact on that. So we've interviewed many users and also key decision makers, And we found out many insights. So just a few of them right now. First of all, that there is a lot of trust that is going to be happening between the end users and the Gen AI agent. So it's not gonna be something that will surprise anyone. You can see that eighty percent of the customers trust the agent to solve their problems, whether it's customer related issues, or sixty percent will have a positive impact on their brand perception where they'll interact with an agent. So that's not something that is going to be having a negative effect on us, it's actually gonna have a positive effect, not only on the organization, but also on the people that are using the daily communication service provider services. And of course, that will be something that we'll see the next one or two years being deployed more and more. So agents will become a daily practice for end users and for the developers and for the IT system engineers as they work their way inside the telecom ecosystem. But of course, everything comes with some kind of a fear and some kind of anticipation of risks and challenges, especially around things that are now changing human intervention. And those challenges of implementing agents, we are very much aware of them. So we are aware of compliance issues and regulations and security and privacy, and of course, customer acceptance. And those are things that we have to address from technology perspective. I want to show you how Amdocs has been together with Microsoft has been handling this momentum. And first of all, I wanna talk to you about verticalizing agents. So agents as we know them could do many things. Agent can send you a reminder on your phone and agent can just invoke an API. When we are talking about telecoms, then you really have to make sure that your agents are very much proficient in the domain that they apply to. So if I have an agent dealing with care or an agent dealing with sales or marketing, they all have to base their intelligence on the best practices that we have in the telecom ecosystem, whether it's the right people that we learn from, the right knowledge bases, documents, policies, protocols, anything that we have. We need to train the software to make sure that it has the ability to cope with the decisions that previously had been done by humans. And of course, to that intelligence that we're creating inside the maze, inside our GenAI platform, we're also applying a layer of experience. As you have seen in this webinar that Ganso mentioned, that AI now is the new UI. So you really have to make sure that your experience with the customers is also in tune with your agentic behavior of the software. And agents in MAs are consolidating over five domains. So you have care, sales, home marketing and network. And each one of them has sub agents, which are expertizing in very, very, very specific tasks. So an agent can deal with disputes or payments or retention or assurance. Each one of the agents would have its own dedicated special expertise that will be trained on and support the organization and the end users, of course, of the CSP. Now I want to show you a quick demonstration of how a care agent is dealing with a situation of a build spike of a customer. So let's have a look at that demo. So we can see in this demo two screens. The one on the right is the mobile phone of the user, of the CSP, of the operator, and is getting a bill spike notification. On the left hand side, you will see the backend capabilities of the system, what happens behind the scenes. Eellux, I'm Matt, your AI agent. I hope you're having a great day. I want to let you know that your bill is expected to be higher than previous months, so we wanted to give you a heads up and explain the changes. Okay. Go ahead. You can see that the Amaze platform is using an avatar to communicate with the user to explain them about the bill. And on the left hand side, you see the relevant agent that is taking care of the problem. In your previous bill, you still enjoyed your six months discount of nineteen dollars. You recently subscribed to the amazing cloud gaming subscription bundle for twenty dollars. So the increase is due to the termination of the six months discount and a newly added cloud gaming subscription bundle, summing up to thirty nine dollars difference. Oh, do you have a better offer for me? The Amaze agent has presented two bills, the one from last month and the one from this month, and explained the changes between them. And now the user is asking for a new plan. Last time we spoke, you mentioned you're a Fortnite gamer. Therefore, I recommend our amazing gaming ultra plan, which matches your usage patterns. For eighty one dollars, you'll get two gigabit speeds, a cloud gaming bundle, and bigger bandwidth. This is only two dollars more than your next bill. Any questions? So to summarize this interaction, the user has got a better plan now that was pulled by Amaze from the catalog relevant to the specific behavior of this specific individual. And on the left hand side, you see the entire flow of the system, starting from the prompt via the guardrails, going through the MACE platform, selecting relevant agent, the relevant skills and providing the right reasons to provide a specific answer. And this is done together by Amdocs and using the capabilities from Azure OpenAI services. So after that demo, we can see that we actually have been able to deploy some of those care agent activities together. We'll be working very closely also with Microsoft and Nvidia as part of our services deploying the agents. Of course, for Microsoft, we're using the Azure OpenAI service and Azure ML and other components from the great Microsoft stack. And from NVIDIA, we're working with NIM, their NVIDIA Inferencing Microservices and more elements from their AI Foundry stack. So that collaboration is really bringing something that is innovative and very comprehensive to the telcos coming from the three companies together. We also have it as part of a CEP, the customer engagement platform that is dealing with commerce and sales and marketing and care throughout the entire ecosystem of the telcos, working together in a collaboration together with Microsoft services and Amdocs products and solutions to provide a very holistic, comprehensive solution for the B2B and the B2C domains. The last thing I want you to leave you guys with is actually the great achievements that we had with some of the POCs that we did for our customers together. So you can see that some of the KPIs we'll be able to achieve are really talking about cost reduction of tokens and thirty percent improved accuracy and down by eighty percent of the latency of the inferencing. So that really shows us that it's not just about a new technology coming out there to the market, it's really impacting call center, customer experiences, average handling time and other KPIs. And those are numbers which are very, very impressive for new technologies. And we're happy that we can share them together with you guys. And lastly, will leave you with some of the feedbacks that we even got from agents using this. Like for example, that the information was accurate, it was good to have a very right set of information, it was powerful. So you can see that care agents that were using the technologies we built together have really a great outcome and really improve their day to day work with the agentic technologies in place. And also of course, as I mentioned, on average endemic time and first cone resolution and Net Promoter Score. So these are great outcomes that we see from multiple POCs that we're doing together. And we hope to continue and see those more and more as we go along. And they really bring something new to the market with the agents and the software between the two companies. Thank you. Very cool. Thanks, guys. I have a few questions here, actually. Ofir, can you talk a little bit about how the technology from your company interoperates with Microsoft's technologies to create this customer experience? I mean, where's the dividing line and how do you join them together, I guess? Okay, so sure, Steve. So it's actually, we're working on two fronts. So first I'll talk about Amaze, that the collaboration really tries to bring the best of both worlds, right? We have the Telco native intelligence bringing from Amdocs, we have the Microsoft enterprise grade GenAI infrastructure. And I think using Microsoft Azure OpenAI and Azure ML and other components from the Azure Stack, we're actually being able to embed GenAI at scale. We make it more secured, more governance, and most importantly, we make it very telco specific. So you have the telco know how and you have the agentic framework and infrastructure coming from Microsoft, including their cloud that brings together the best of both worlds to create a new experience for CSPs. And as I mentioned, on top of that, we also have the customer engagement platform that is using the Microsoft collaboration and productivity ecosystem like the Microsoft three sixty five, the Power Platform with additional tools that we bring from Amdocs. That combination together again, brings productivity and telco focused specific insights in AI across all of the customer journeys and touch points. Very interesting. Gansu, I'm wondering whether you can bring this to life a little bit for me and the audience. The idea of a Gen AI agent being able to act proactively, perhaps even faster than a human agent would be in solving a customer issue. Can you give me an example of how that might work? Sure. Thanks, Shu. That's a great question. So, in terms of, right, I think the technologies that we provide as Microsoft, the platform, we provide the tools necessary to integrate into the data and the systems that I'm not sorry, any telco partner would have to to identify those patterns or the or how to reach out proactively to those users so that that there is a satisfaction as well as improving the overall go to market productivity aspects. Right? I can give you a few example to start with. One is, for example, see, one of the use cases could be say, if the in the customer care, if if we want to proactively identify the customers who are who might have some issues, for example, billing issue network issues, right? Using some good using some predictive model and proactively reaching out to them to make sure that, hey, you have this change in the billing. How do you like to address it? And then providing the either the absolute or the causal motion with them. Right? I think that one use case is the that we are already exploring with while partners and the customers. The other use case I can I can think of is for to think of is in the areas of is in the area of customer onboarding? So you see, if you are looking at a customer b to b onboarding, of course, you are looking at thousands of customers, small, medium customers. One of the challenge we see is that the how do you kind of automate this this overall sales onboarding experience so that you can serve many customers. Right? So for example, you can identify a location based service. If a particular customer is in a particular location, you can identify is there a specific services that are provided on the location and proactively reach out to those customers. Right? That's another area in the customer onboarding where we can see the patterns around that. The last area I I just want to quote is is on the network area. Right? So of course we talked about a bit of in the part in earlier as well. So the network is is offers a great area where we can apply some network the the gen AI models as well as the AML model to identify predictive maintenance side going on. You can identify some of the issues that that may come up in the future and proactively reach out to the user saying that, hey, we we we anticipate that there is an outage coming up. Right? Do you wanted to have some additional data package to address your issues? Right? Those kind of use cases, we are seeing that is exploring and they are, we are trying to build with partners like Amdocs. Very, very interesting. Ophir, how do you find your customers measuring the benefit or the efficacy of Gen AI agents? Is it mainly cost saving or is it money making or is it customer satisfaction? How are they measuring the success of these programs and what's most important to them? Well, Steve, I can say it's always a mix of everything, right? All the parameters that you mentioned, right? You really have to see how the business is being driven by this. And of course you want to, first of all, try to ease up on your call center, for example. So if you are predicting that you're gonna have like twenty thousand or thirty thousand of your customer base are going to be impacted by a bill spike, then you want to act upon that. So one of the things you can do is the care agent can be proactive here and ignite the conversation. And just like we see in the demo with all of those customers. Now what's going to happen is that less calls will come to the call center. So you already have a KPI in place where you have less calls coming and then you're having less load on the call center, that means maybe less people that need to maintain that call center, right? Or for example, if you're using an agent to reduce the time of the average handling time, right? Or the first call resolution, Again, you could use agents in the back and measure those parameters to really provide a reduction on the load. In essence, also for the synergy that you're looking between a human to AI. If you're talking about a CSR now that will work independently, it might take more time for the CSR, the customer service representative to respond as if they had an agent, maybe they could do it in less time, more effectively, get Q and A from all the documents that they have much faster. So really having another way to kind of make your organization more effective. And like anything that ties down basically into revenue, you can really accelerate your revenue. So not only about making more money, but maybe making the money faster and getting those revenues faster when you have AI working in the back and help you kind of speed up your operations. Interesting. Gansu, last question for you. What's a super agent and what can we expect super agents in the telecom space to do? Yeah. Super agent is a very interesting terminology. It's an evolving space. Right? So, of course, agents would the view we have is that the agent will evolve this into a super agent. By definition, you can think of super agent is a kind of is of a complex agent which automates many of the the task of the transactions across many system, right? In a telco example, if you take a like, if you want a transaction to or a task to span across, say, BSS, OSS, network system combined together with the required human in loop concept. So that's what we call it as a kind of a super agent. Right? So super agent is very early stages. Right? Right? Right now, we are looking at more of the agents that we are building is focused on a particular domain or a particular task. And once this agent technology becomes more mature and the adoption becomes more mature, what we see is that super agent will be the one taking over. And when I say taking over means the organization's key task or that use cases that will help in navigating the complex systems of organization so that you can see more productive use cases around the super agents. Very interesting. We're really at the beginning of a journey here and the agent capabilities today seem quite super. I'm hoping that perhaps the super agent, because of course, GenAI, you can use it to write music. So perhaps the super agent could also do something about the hold music in customer service centers as well at some point in the future. But anyway, guys, it was great to talk to you. Thanks so much for taking us through all of the detail here. Gansu, Ofir, thanks for joining us today, and thank you to our audience as well. See you next time. Thank you, gentlemen, and thank you to our audience. We received a lot of great questions, and we will do our best to follow-up after today's webinar. As a reminder, a recording of this session will be available on demand within twenty four hours. Thank you for joining us and we look forward to seeing you next time.