Rick Lievano: [00:00] I can’t tell you how many times an IT executive has said, “Hey, I’ve stopped deploying AI agents until I can get governance under control.” There’s a big hold-up on how weak AI governance is and the overall organizational readiness to start deploying hundreds of AI agents. We’re not talking about one or two agents. AT&T alone has already deployed over 400 agents. How do you manage that?
DR: [00:22] Yeah.
Announcer: [00:28] This is “Telco in 20,” a podcast that helps telco execs achieve a competitive advantage with AI and the public cloud. It is hosted by Danielle Rios, also known as DR. Today, we’re talking to Rick Lievano, the CTO for worldwide telecommunications industry at Microsoft.
DR: [00:47] Hi guys, I’m DR. MWC26 is just around the corner, and you know what everyone’s going to be talking about: agentic AI, autonomous networks, sovereign cloud, and network APIs. But telco execs aren’t flying to Barcelona to hear more hype. They want proof, real use cases, P&L impact. That’s exactly why I’m excited about today’s conversation with Microsoft. Microsoft has been working with T-Mobile on a network that heals itself when hurricanes hit, with SK Telecom, an agentic AI that feels more like a digital concierge than a teFnetlco app. Real deployments, real numbers. They know what’s working and what’s not.
[01:29] Today, I’m talking with Rick Lievano, Microsoft’s CTO for the worldwide telecommunications industry. We’re going to dive into what telco execs will be looking for at MWC this year, where he sees telcos getting breakthrough results with AI, and the toughest question of all: what’s holding operators back when it comes to AI adoption? So, let’s take 20.
[01:54] Rick Lievano is CTO for the worldwide telecommunications industry at Microsoft. Hi, Rick. Welcome back to “Telco in 20.”
Rick Lievano: [02:02] Hey, DR. Always great to be back in “Telco in 20.” It’s my favorite telco podcast, as you probably know.
DR: [02:08] Woo!
Rick Lievano: [02:09] Super excited to jump in again. Let’s talk about what’s really happening in telco.
DR: [02:13] Yeah. MWC Barcelona, our biggest show in our industry, is just around the corner. What do you think everyone’s going to be talking about? What are going to be the biggest topics?
Rick Lievano: [02:22] Oh, man. I think we’re going to see plenty of talk about AI everywhere.
DR: [02:26] Yeah, for sure.
Rick Lievano: [02:26] Being a telco-focused event, we’re going to see things around non-terrestrial networks. I think that’s going to be huge there. Your buddy Elon made a big announcement with Starlink recently, right? They’re pouring some pretty serious money into Spectrum and they’re pushing hard to become part of the mainstream mobility ecosystem. Agentic AI. Come on, that’s got to be everywhere.
DR: [02:46] Yeah, for sure.
Rick Lievano: [02:47] We already got a hint of that at MWC25 last year, but I think that was like a teaser. I think this year is going to be about agentic proof points, real deployments. Network autonomy I think is going to be huge there as well.
DR: [02:59] Yep. I believe so, too.
Rick Lievano: [03:00] And maybe this rebirth of the edge. Let’s see. What are we going to call the edge this time? Maybe let’s wrap it around this shiny new sovereign cloud label. So we’ll be seeing a lot of that as operators try to localize and differentiate their own cloud offerings and workloads. And if we’re lucky, DR, I think maybe someone will unveil the latest 5G-connected AI-enabled bird feeder. I know you’ve got one. Maybe we’ll see an even better version of that, because once you turn 50, you know you become a birder, so…
DR: [03:29] You become a birder. So what should telco execs be watching for from Microsoft at MWC?
Rick Lievano: [03:34] Yeah, good question. If you’re a telco executive, there’s going to be a few things that you should be keeping your eye on. First, our network automation and autonomous network vision, that’s going to be clear at MWC. We started a project that we called internally NetAI. It’s work that’s now maturing into the Network Operations Agent framework or NOA, and that really is giving operators a path towards real measurable autonomy in their networks.
[03:59] I think another piece is Copilot. Who doesn’t love Copilot? Look, it’s not just an AI companion for your employees. It’s really a platform with real telco customer stories behind them. We’ll have strong aging governance capabilities and you’ll see us lean hard into optimizing network workflows with agentic AI from field operations to customer care, network planning. So watch for some cool demos that enable regular business employees, not just Stanford wizards like yourself, to create, deploy, manage their own AI agents for BSS functions and automations like product management to manage order fallout. You know order fallout is a huge issue for telcos. Let’s automate that process, and then we’ll demo how you can do that with AI tools like Copilot Studio.
[04:44] And then finally, I think we’re really helping operators take all of this innovation to market. It’s all about the top line, delivering AI-powered telco capabilities so that their enterprise customers could benefit from this, and it opens up new revenue streams for the operators that go beyond connectivity. Let’s help them focus on running smarter, moving faster, and then monetizing AI in ways that actually scale.
DR: [05:05] I agree with that. I think at MWC, people are going to be looking for not just how does this work and what can you do with it, but how does this make my business better?
Rick Lievano: [05:13] Agreed.
DR: [05:14] Show me the money. On your NetAI and NOA framework, that was a big topic at TM Forum Ignite that was in June of this year. Tell me a little bit more about this product and how it’s working.
Rick Lievano: [05:28] Yeah, good catch. Microsoft runs one of the largest networks in the world, 600,000 kilometers of fiber, and our biggest fear is just a man with a shovel-
DR: [05:37] Wow.
Rick Lievano: [05:38] … because a fiber break or an outage, they’re just daily occurrences for us, but we need to keep that entire network humming at all times and we can’t do that without AI.
[05:47] NetAI is really our big leap forward in letting networks run themselves safely, reliably, and at a massive hyperscale way. You can think of it as a collection of intelligent, well-trained AI agents that are going to handle all the tough, repetitive parts of network operations, anomaly detection, spotting issues, diagnosing issues, even really being able to kick off repairs. In fact, we expect 10 times the volume in four years, and network traffic, it’s huge. How are you going to scale that? Are you just going to continue adding headcount? No, we’re going to leverage AI.
DR: [06:20] Yeah, for sure.
Rick Lievano: [06:21] So a couple cool innovations around NetAI, this concept of stateful workflows, curated engineered prompts, specialized roles. Agents will take a role of a NOC manager or tier-two support engineer or a fiber repair specialist and they’ll be able to automate the whole lifecycle of an incident from detection to resolution. Each agent really is constrained to do only what it’s supposed to do. It has guardrails, for example, read-only defaults or pre-approved repair actions, really tight RBAC controls with regards to role-based access.
[06:55] Going back to curated prompts, super important. These are engineer-vetted prompts or queries that these agents can use to make sure that their actions are deterministic, which you know that’s not the norm. I know you love stats, so with NetAI, 40% more incidents that humans can handle, 80% faster root cause analysis, that’s huge, 25% reduction in time to repair. All of that is baked into something that we call the Network Operations Agent framework, NOA framework. All this goodness is now available to telcos to build their own NetAI capability.
DR: [07:35] Yeah. People are getting into Waymos. You’re not even in the front seat. You can’t even grab the wheel. I hear all the time, “Oh, how can we have AI take over the network?” and I’m like, “It is possible.” You got to do things like what you’re talking about, curated prompts, the guardrails, the checking. Waymo wasn’t built overnight. It took a lot of testing and neural network analysis to make that safe. We absolutely can do that for the network, and we’re going to have to do that for the networks around the world.
Rick Lievano: [08:05] 100%.
DR: [08:06] Let’s talk about real telcos using AI. I think everyone’s heard all the use cases about customer support. I think that’s a snoozefest, but can you share some examples of telcos using either NOA or some other Microsoft AI products in other ways?
Rick Lievano: [08:20] Yeah. Customer care, that’s where the party started, but it’s morphed and it’s moved to other parts of telco as well. I’ll talk about some recent things that we’re doing. I think a really interesting one for me is SK Telecom, our good partner in South Korea, they’re doing the agentic workloads, which they built with Microsoft Foundry. Microsoft Foundry is our developer AI platform. Instead of just really plugging AI into a single task, they’re really using it to re-architect how customer interactions and internal processes outside of that basically flow end to end.
[08:57] They’re being able to now create these multi-step AI-driven experiences that really deepen that customer connection in ways that feel more like the digital lifestyle concierge than a traditional telco app. But what’s really interesting there is that it’s not just like ChatGPT and the call center. We’ve heard that a million times. It’s really about building new AI-native digital experiences that leverage this new agentic AI capabilities. So they’re finally moving into becoming the techcos that we’ve talked about, and that’s a story that most people still haven’t heard.
DR: [09:22] Yeah, for sure.
Rick Lievano: [09:23] Another cool story, it’s with T-Mobile, our Un-carrier friend in the US. T-Mobile has decided to stop treating their network data like the messy garage full of priceless stuff. They really teamed up with us to roll up their sleeves, unify their entire network data state. They went from, “Hey, we think we know what’s happening in the network,” to processing multi-petabytes of data. They’re ingesting almost a petabyte a day, so 800-plus terabytes every single day of network data, handling more than a trillion transactions daily. It’s numbers that blow your mind.
[09:55] Suddenly T-Mobile has this crystal-clear real-time view of how their network behaves. Where is it humming? Where is it straining? From a planning perspective, where should we deploy new features? Powers, for example. This move to the cloud to Azure didn’t just make things faster for them. It really changed the entire rhythm of the company. Product teams can now collaborate in the cloud.
DR: [10:20] That’s awesome.
Rick Lievano: [10:21] Less friction. They can release updates faster. They can experiment the fast-fail much more quickly. But now that they’ve got their data state in order, now they can do some really magical things with AI. I live in Florida, so hurricane is just a part of my life. It’s just another Tuesday for me. Hurricane hits, trees are down, power lines are out, cell towers take a beating, and then for most people, the network either just works or it doesn’t. But behind the scenes, something remarkable is happening with T-Mobile. They’ve built this AI-driven, self-healing network, fully autonomous.
[10:52] As soon as the cell site goes dark because of an outage like a hurricane, the surviving towers, they’re not going to wait for an engineer to just show up. They can instantly re-tilt antennas. They can adjust power levels. They can stretch their coverage to fill the gap left by the cell sites that are out. That’s autonomy.
DR: [11:09] Yeah, that’s cool.
Rick Lievano: [11:10] That means that, hey, I can call 911. I can call my loved ones and say, “Hey, I’m okay.” That’s the magic of this transformation, but it requires that data foundation to be trusted, and then you can do amazing things with AI. It’s moments when communications matter the most, it’s not just technology, it’s your lifeline.
DR: [11:27] Yeah, it’s when the network shines.
Rick Lievano: [11:29] Exactly.
DR: [11:30] Yeah. You mentioned this at the top, giving edge a new name. Microsoft has been announcing telco partnerships around data centers and data sovereignty. So my big question, asking you at Microsoft, what is different about a sovereign data center versus an Azure region? Are you partnering and setting up an Azure region in that country with some extra bells and whistles to make it safer a little bit, or is it truly a different thing?
Rick Lievano: [11:56] There are some differences. A sovereign data center is really all about controlled jurisdiction and trust, and it’s built specifically for a country, for a region. They have final say, really, over how data is stored, how it’s accessed, how it’s governed. In this sovereign cloud model, all customer data stays within the region. Operations in a sovereign data center are exclusively performed by in-region personnel, EU residents, for example, and that gives telcos and governments and other regulated industries that assurance that sensitive network and subscriber and other metadata never crosses borders or really falls under foreign legal reach.
[12:33] The easiest way to understand it, is a sovereign cloud is the same as an Azure cloud, but it’s with all those guardrails turned all the way up. We announced this concept of the public sovereign cloud, the private sovereign cloud, which is essentially between your four walls, and then something special which we call the national partner cloud. Really in all cases, you have the ability to have your data stay in Europe, for example, operated and handled by European personnel only. Your encryption keys can now live outside of Azure in the partner of your choice, so now Microsoft wouldn’t even have access to the keys themselves.
[13:08] One example is Bleu in France, where we partner with Orange and Capgemini to essentially run a sovereign Azure environment that’s completely run and managed by a third party. There’s no Microsoft employees that are involved in the day-to-day running or access of the Bleu data centers. The beauty of this is that, for operators, it really unlocks a new role for Orange. They’re not just network providers, they can actually be trusted digital stewards who can offer now sovereign cloud services to governments or regulated industries or enterprise customers that are airtight. It also means that telcos can run sensitive OSS/BSS workloads, even these emerging AI use cases under really, really strict data residency and legal jurisdiction guarantees. So a lot of cool stuff happening.
DR: [13:55] But what do you do about the LLM level? You can’t control where Claude is running or ChatGPT. So is that forcing their LLM choice to be some weird model?
Rick Lievano: [14:06] That’s a great question. We have this concept, Foundry Local, which enables you to push an inference to the edge or to a sovereign cloud, so you do have the ability to do inference of some LLMs.
DR: [14:18] Yeah. You guys are working with some big telcos on these AI implementations. What’s slowing them down? Is it, “We don’t have the people that know how to do it”? Is it a cultural battle? Is it fear of things going haywire and off the rails? What’s holding them back?
Rick Lievano: [14:34] All the above. I work with telcos all the time on big AI programs. They do struggle in very much the same places. The obvious one is fragmented solid data. Every telco already knows this. It’s a pretty big blocker, maybe the biggest. I’m not going to belabor it. It’s boring. The biggest challenge I’m seeing really emerging, in fact, I can’t tell you how many times an IT executive has said, “Hey, I’ve stopped deploying AI agents until I can get governance under control,” there’s a big hold-up on how weak AI governance is and the overall organizational readiness to start deploying hundreds of AI agents. We’re not talking about one or two agents. AT&T alone has already deployed over 400 agents. How do you manage that?
DR: [15:13] Yeah. A demo, one capability, no problem, but when these start talking to each other and making decisions, and I’ve been blogging and talking about this a lot, managing context at scale. This is where Totogi is really focusing on ontology, where we’re arguing you can manage that context in a foundation layer and then all your agents can read from it is going to be a very important thing for telcos to be thinking about as we move from experimentation to deployment at scale.
Rick Lievano: [15:40] I think that’s what you’re going to love to hear about Microsoft Agent 365. We’ve been listening. We did announce this at Ignite. Agent 365 is like this air traffic control tower for your enterprise AI agents. No matter where those agents come from, it’s going to give organizations one place to go and govern how those agents behave, what data they have access to, what actions they’re allowed to take, which tools can they access. So nothing flips out of sight.
[16:09] It’s not really limited to Microsoft’s own agents or Copilot or Foundry. It can really orchestrate and oversee agents built on Bedrock, Vertex, really any other platform as well, it’s not a Microsoft-only game here, so that every decision path, every trigger and every outcome is monitored, it’s accountable, which is super important, and it really means that you can start scaling AI confidently without losing that visibility and control.
DR: [16:35] That’s super awesome. Well, last time we talked, you were playing a little bit of pickleball in Florida. You and I are now connected on the DUPR app-
Rick Lievano: [16:42] We’re DUPR buddies.
DR: [16:43] … which is the way that you keep track of everyone’s skill and what tournaments and how people are performing. I found out this really interesting fact that I want to share with you, which is, did you know that the founder of Microsoft, Bill Gates, has been playing pickleball for 50 years?
Rick Lievano: [17:00] Yeah, that’s amazing. He’s also a really big tennis player, right?
DR: [17:03] Yeah.
Rick Lievano: [17:04] He’s 70 now, so I think he’s probably shifting a little bit more towards pickleball. But not a surprise. Pickleball was invented in the Seattle area, Bainbridge Island.
DR: [17:11] Yeah, for sure.
Rick Lievano: [17:12] Who would’ve thought the cloud and pickleball have the same home?
DR: [17:14] Yeah, no. I found this blog or something he has called Gates Notes where he wrote about how he’s been playing pickleball. He knew the founder and played in the original court-
Rick Lievano: [17:23] Wow.
DR: [17:24] … that was first built.
Rick Lievano: [17:25] I didn’t know that.
DR: [17:26] As everyone knows, I’ve been getting into pickleball, and so that was a really interesting fact that I want to share with you since you work at Microsoft.
Rick Lievano: [17:32] And congrats on your tournament win.
DR: [17:34] Thank you.
Rick Lievano: [17:35] You’ve inspired me, so I joined the league, definitely following your lead. I’ll see if I can catch up to your DUPR, though. I’m a lowly three, four, but I might be sandbagging a bit, too.
DR: [17:44] We got to help DUPR with their machine learning algorithm-
Rick Lievano: [17:46] Oh, my goodness, yeah.
DR: [17:47] … because it’s terrible.
Rick Lievano: [17:48] It is terrible.
DR: [17:49] Rick, this is an awesome conversation about what’s going on in Microsoft and AI, and you guys have a great plan for MWC, so I can’t wait to see you in Barcelona.
Rick Lievano: [17:58] It’s going to be a lot of fun. See you there for some tapas.
DR: [18:00] Awesome. Thanks. Stick around. We end each podcast with a “Telco in 20” takeaway. I’ve got two minutes to tell you something you need to know.
[18:16] Rick just told us that telcos are finally deploying AI agents at scale, but operators like AT&T are staring down 400 agents with little visibility into what they’re doing or what they have access to. Some telcos are even hitting the brakes on AI deployments because they’re scared of losing control, but there’s a better way to manage this problem: use an ontology. Think of it as a semantic foundation that teaches AI how your business actually works. Instead of 400 agents each guessing what subscriber or churn risk means across your different systems, the ontology establishes one canonical definition that every agent uses.
[18:54] This says two things: it prevents AI hallucination by ensuring semantic consistency and it serves as the operating model that tells agents which systems to access, in what order, and with what constraints. When you need to make a change, you change it once in the ontology and every agent follows the new rules instantly. This is exactly what we’ve built with BSS Magic at Totogi. Our telco ontology manages the control layer so your agents can act with confidence. The control is in the ontology, not scattered across individual agents. One of our tier-one customers actually accelerated their deployment to production because the ontology made it so safe. They skipped months of testing and compliance review and jumped straight to value. The best part, it works on Azure and anywhere else you want AI deployed. So don’t pump the brakes. Just unite your AI with Totogi’s telco-specific ontology and scale with confidence. Want to see BSS Magic in action? Visit Team Totogi at MWC in Barcelona from March 2nd through the 5th. We’ll be in Hall 2, Booth 2G51.
[20:01] Connect with me on LinkedIn or X @telcodr to set up a time to meet. Until then, tune into more “Telco in 20” episodes. Like and follow and leave us a five-star review. Don’t forget to sign up for my amazing email newsletter on telcodr.com and check out our rockstar YouTube channel and hit that subscribe button. Later, nerds.
