The growing role of AI in telecommunications

In this fast-paced episode of the Trending Tech Podcast brought to you by the team at IoT Now, explore the role of AI in telecommunications, especially for telcos and MVNOs, including its integration into IoT services. Your host, Jim Morrish, Co-Founder of Transforma Insights, and his guest, Rupa Datta, VP of Connected Services at Semtech, delve into the benefits and challenges of this fast-growing technology. 

Transcript

Jim Morrish: [00:00:00] Hello, everybody and welcome to this episode of the Trending Tech Podcast brought to you by the team at IoT-Now.com. And, a warm welcome to everyone out there who's, who's listening in, who's tuned into this. My name is Jim Morrish. I'm the co-founder of a company called Transforma Insights.

We're a firm of industry analysts focused on all things related to digital transformation and for this episode of the podcast, I am joined by Rupa Datta, who is Senior Director of Product Management for IoT Connectivity Services with Semtech. So welcome, Rupa, and why don't you tell everyone a little about yourself?

Rupa Datta: Hey, Jim, thanks so much for having me on the podcast. A little bit about myself. So obviously based on the accent, you can tell I'm based out of the United States. I am managing the global IoT cellular connectivity product portfolio for Semtech, and I've been with the company for about five years now, [00:01:00] and although, you know, IoT is a rapidly changing environment that we all know and love, and so along with that, from a product perspective, we're really interested in this space because we are forced to change at a fast pace along with the customer demands.

So it's exciting. It's fast moving, and I'm super happy to be here talking to you about Trending Tech.

Jim Morrish: Great. Thank you, Rupa, and it's great to have you here. So just to introduce the topic of today's podcast. We're going to talk about AI in telecommunications. And CSPs and telcos in general have been quite active in this space in the last few years. You know, various telecom providers, they're seeking to differentiate with things like providing AI infrastructure, data centers and so on.

 Professional services, consulting, AI platforms and AI integrated solutions. And, they're also looking to support and enhance internal operations, you know, including things like brand management and network operations and customer support and [00:02:00] and so on and in many ways, telecom providers, they're kind of, they're just like normal businesses, with opportunities to automate their call centers and marketing initiatives with things like generative and so on.

But, they're also in a unique position, where they can benefit. From network and related operation efficiencies enabled by AI, but, but also they're well positioned to provide an ideal trusted channel to end user adopters and assist them with, the adoption of those new AI enabled solutions.

So that's the topic of today, I'm sure as you've mentioned, that's something that's close to your heart, Rupa.

Rupa Datta: Yes, and it's a mouthful, isn't it, Jim?

Jim Morrish: It is.

Rupa Datta: Lots, lots of buzzwords and keywords that people have been hearing in the industry for a long time. So we're happy to be here to kind of make it simple to understand and give the customers and enterprises a way to understand [00:03:00] how to implement AI, but more importantly, to understand what telcos and MVNOs can do for them.

Jim Morrish: Absolutely and as you say, it is an industry full of buzzwords, although I have to say that I, I hear AI so often that, that every and again, I have to remind myself what it stands for it tends to be. But before we get stuck into AI and telecoms, let's take a quick look at some, you know, a serious tech news story.

This is one that I've found and I'm sticking with the topic of AI, et cetera. But I, I came across this story, which in itself isn't particularly significant, but, but it's illustrative, I think, of a wider trend. So there was a press release, and I'm from the UK, as you can tell from the accent and the, there was a change in the name of the UK's AI Safety Institute to an AI Security Institute, and it just struck me that this was reflecting a change and a shift in the industry.

It's illustrative of that. The [00:04:00] theme and industry and regulators get pivoting away from, you know, the hypothetical consequences of AI and, and focusing much more on, on concrete considerations and it reminded me of a, of the law, which was recently published in South Grid, December, 2024.

Now the draft version of this law emphasised fairness in AI which regulators everywhere seem to do. I mean, this is very much a theme of 2024, but the actual published law does not mention fairness. And, and it seems to me that, that illustrates a kind of a maturity in the regulation of this around, you know, just recognising that, frankly, fairness isn't, isn't possible, in AI, but there are some really significant things which do need to be regulated.

And it just  seemed  the zeitgeist of the industry had changed slightly. Did you notice the same story, Rupa?

Rupa Datta: Yeah, I did. I picked up on that story and I thought it was interesting. There's a principle of fairness in that. From our customer, and I'm going [00:05:00] to talk about it from our perspective a little bit, in MVNO world, our customers are trusting us with a lot of data. And they want that data to be fairly protected and fairly used so that they know that, unfavorable actors are not going to get access to that data and do You know, unfavorable things with them.

So for example critical data, patient data that's being sent over, cellular, you have lots of security data from some of our, you know utilities, customers, et cetera, how can that data be used in an unfavorable way? This article touched me quite a bit because it really gets to the heart of crime, doesn't it?

Which isn't really where we are, but it does kind of unveil some of the threats that could be there with just access to data as in photographs and personal information that could be out there. On the Internet and, and in [00:06:00] the cloud. So I think security is a key word when you talk about AI, especially in our industries and in any industry.

I think it's interesting that even from a semantic perspective that changing safeties to security actually bolsters the message so much more from an end user and a reader perspective.

Jim Morrish: Yes, it does very much. And, and, and the way that you described it there effectively, you were parameterising fairness, you were roosting it in some definition. And that is something which is very much achievable. Whereas up until recently, the regulators much more started the other end and said, everything must be completely fair to everybody.

And that's kind of unachievable. But this more rooted, grounded tangible approach I think is coming to, as you describe is, is coming to be more widely adopted in the industry. It's more about doing it rather than worrying about it, I

Rupa Datta: Correct. Correct. What guardrails can we put in place for protection? Yes.

Jim Morrish: And [00:07:00] with that, we should probably come to the main discussion about, you know, what, what happens when AI meets telecoms. So, I mean, as we've discussed we've heard a lot about AI, and in fact for years now, but, but why is there so much hype now, do you think?

Rupa Datta: I think there's a hype now purely because we've been hearing about it for so long and we haven't seen a lot of momentum, right? Like, you know, people, as you said, AI, you've heard it so much, you've almost forgotten what it stands for. So, you know, I remember 10 years ago, 12 years ago, I was working for a different company.

And at that time, AI driven meant how do we get proactive notifications to the same set of customers based on datasets. Right. And so that was, how was that driven? That was driven by staffing. You had to increase staffing of data scientists to go look at all this disparate data that was around to make sense of it and say, how do we add value?

What's the value creation for a customer? And [00:08:00] that was the start of AI. I think why are we seeing so much hype now? There's been a lot of investment in AI. You have ChatGPT, AWS, all of the big players in that AI space making AI accessible. In many ways to everyone, to the, to the mass majority of the public.

So people are touching AI, they're experiencing AI, they are understanding AI, so you've, you've almost brought AI to the fingertips of the customer, which is why I think it's more relatable. And therefore more understandable. So hype is there, we see advancements in the AI technology. A lot of it.

A lot of generative AI models. We've had a lot of enhanced capabilities across many, many industries. It's not just in tech anymore. It's down to, you know, you're calling to get your furniture delivered and your call is being routed to the right person because of something you've said. Like that's also, there's some language modeling happening there, right?

So we're seeing it in various [00:09:00] contexts. So accessibility I talked about. And the last one is really around integration. So the tools that are available in the market today, although, they are much more accessible, but the integration within those tools or to those tools are much more simple today than they were five, 10 years ago.

So I think that's where the hype is. People are understanding the value of what they can do with AI.

Jim Morrish: Yeah, absolutely. I tend to agree that and I'll pick out one aspect of that. You mentioned the ChatGPT bringing it to the masses and making people aware. So, so some number of years back I can't remember who it was, but they were asked what the main benefits of blockchain was. They said it gets attention on management boards and boards of directors will listen and understand things that can be done with normal technology, but because it's got blockchain, they're excited and interested in it.

So there's this 

Rupa Datta: That word again 

Jim Morrish: Yeah. the, it's putting it there in front of people and, and underpinned [00:10:00] by real developments, as you say. In, you know, investments, et cetera. But it's not all plain saying. So, what have been and what are, and what might be the barriers to entry for, for businesses to, use AI and AI tools?

Rupa Datta: Yeah. So, I mean, I'll talk about three main challenges that I've seen in the market and that I've experienced across different companies. Number one is really the lack of the skilled expertise in instituting AI in a company. So what are you looking for? Again, going back to my previous comment, data science, right?

That's something that companies need to really look at the barrage of the second point, which is data quality and availability. So many companies, you know, we love data. We love to collect data for everything, every interaction, every customer, everything. We're looking at MPS. We're looking at, you know, how many times have you called it?

We're looking at so much data, but many companies don't [00:11:00] really, categorise and catalog that data in an accessible library and in order to really, get AI on a track where it's beneficial and very efficient. You need to know what your data is and how you're cataloging your data and how you're using it. 

So those two things like you need the personnel to analyse and put that data in the shape that it needs to be in order to actually put some AI goodness in place. And then I think the last one, Jim is really around financial constraints. So when we look at AI development, there are high costs associated with it.

I mean, most companies because of the high cost associated with it are doing the things that you mentioned. Sort of in your introduction where we're looking at efficiencies, we're looking at chatbots, we're looking at things like that because it's low hanging fruit and we know how to collect data and then put it back into that AI virtual machine, right?

So it's really those three things that I think that companies are looking at today with skilled [00:12:00] personnel, the quality of the data they have today, how fast they can move with it and then given the quality of the data. How do we right size a project to kick something off and that's around the financial constraints there.

Jim Morrish: Yes, absolutely. And, and those financial constraints can be considerable, certainly in early days, it's easy to adopt AI solutions if you're a company and you see your peers adopting it and you see them getting returns, but at that point you feel confident that your ROIs can be achieved. But to be the first guy, is, it nerve. 

Rupa Datta: yep, first to market is always great for press, but sometimes we learn from the first to market, don't we?

Jim Morrish: Exactly. But it does seem like in many ways that that dam is breaking in the flexibility of many of these tools and particularly things like, you know, agentic AI, which starts to address some of those data quality issues in terms of actually finding  the right information  

Rupa Datta: Absolutely 

Jim Morrish: wherever it might be There's a real sort of synergy, a virtuous circle going on, I think.[00:13:00] 

Rupa Datta: Absolutely. And I think actually like with the evolution of these AI tools, and as you mentioned, you know, it's getting easier with the accessibility is easier, but the tools are getting more sophisticated. So they are crunching through your data for you and putting them in some sort of format. And in that case, what's going to happen is that skilled personnel piece that I said, that first factor, that's probably going to not be such a big barrier

Jim Morrish: Yes,

Rupa Datta: Because the technology is just going to get better.

And do it for you. Exactly.

Jim Morrish: Excellent, but pivoting this back, we always talk about telecoms a little. So as homes in specifically on telecoms, how are telcos, how do you see telcos utilising AI to support their businesses? Do you see them using it in IoT? Are there any particularly interesting use cases you've come across?

Rupa Datta: Yeah, I think that's a great question. It is particularly interesting and is something that I think we'll see probably in the next couple of years. I think we're all kind of starting from [00:14:00] a level playing field. We talked about the challenges. How do we kind of get started with AI? What I'm seeing in IoT and in telco right now is really lean into how we better the customer experience. And By bettering the customer experience, what we're really seeing is, and I mentioned this before, is sort of the enhancement in how do we manage your call or your tickets better? How do we understand what you need more accurately? And how do we, quickly resolve your issue if there is anything? So in that realm, we're seeing things like AI driven chatbots.

We're seeing the virtual assistants that pop up 24/7 so that a human being doesn't need to kind of interact with the customer. But what we are seeing is the real time issue resolution, which is increasing customer satisfaction scores quite a bit. So that's something that's, you know, a real improvement, especially in the telco space where we [00:15:00] know that, from a, Consumer perspective, we've all been on the phone for many, many minutes or, you know, 30, 40 minutes trying to resolve an issue.

So I think that the emergence of AI and the customer experience piece is going to be helpful. The other thing we're seeing is fraud detection. So there's two levels of fraud detection or, sort of threat detection. One is from a customer experience where a customer may have an application that's hitting the network.

They're using data at an even stream, even clip over several months. And all of a sudden there's a surge or an abnormality in their data usage that could detect fraud. What are we doing with it? So there is AI to say, okay, Jim's fleet of stuff is now all of a sudden throwing terabytes of data. What are we going to do?

Are we going to automatically suspend it and let Jim know? Are we going to automatically throttle it? What are our decision points? So what we're seeing is AI is enabling companies to make decisions or to empower [00:16:00] decisions autonomously so that human beings don't have to be in that mix, and I think that's something again from a fraud and protection perspective.

We're going to see coming out in telco. We're doing that today at Semtech. We're actually detecting the fraud and proactively notifying customers if we see abnormalities in usage, and I think that's been a game changer for us. The third thing is around predictive maintenance, and again, it's around.

How is your equipment behaving? So Jim's got an application, Jim's got a device. How is that device and application behaving on the network? Again, if we see failures, if we see spikes, what we can do is actually generate some proactive maintenance even. So can we reboot the device? Can we do, what can we do to get Jim's device back to normal?

And those are things that we're working with customers on today as we You know, detect that abnormality on the network and then enable the customer either [00:17:00] to make a decision or enable the customer to give that decision to us to say, here are three things to do and to set that really with AI.

Jim Morrish: Yeah, absolutely. I'll just pick up one of those things that you mentioned that that real time issue resolution. Now, clearly, that kind of customer support of customer engagement support exists across many industries. But One of the key things, as you mentioned a little later, is the real time nature of telecoms networks, the real time emerging problems.

So The extent and the landscape from which a customer service agent or an engineer may need to draw information to say, ah, this is what's going wrong. All that and navigating to that in real time is something which is kind of peculiar to telcos, peculiar in a good way. [laughts] 

So focusing a little down onto Semtech.

How do you guys view the utilisation of AI and what can enterprises expect to see coming out of Semtech under an AI brand?

Rupa Datta: Yeah, I think enterprises in general from an MVNO and from Semtec, [00:18:00] they're looking for first and foremost, why do they come to an MVNO? They're looking for reliable connectivity. And so some of the things that we talked about before in the fraud detection, predictive maintenance, and also from a network optimisation perspective. What we're doing is we're putting AI algorithms and some anomaly detections in place to really make sure that we are seeing any abnormality in the network based on a customer application or across our customer base so that we can keep that uptime going so that we can keep the integrity and efficacy of the network intact for the customer.

So that's number one reliable connectivity. That's table stakes for us. I think on top of that, we deal with certain different sizes of enterprises. So we have small, medium, very large enterprise customers also. And so we've seen some of our customers grow from medium sized businesses to very, very large sized businesses.

And so what, what are they looking for there? They're looking for [00:19:00] scalability. So scalability is paramount for them to know that, Hey, I'm a mid sized business. I can come into Semtec and as I grow, the service isn't going to change. I'm still going to get the same customer experience. I'm still going to get the efficacy of the network.

I'm still going to be, you know, protected with the security protections. So I think, you know, we are doing things for smaller, the same thing we're doing for enterprises. Any customer gets the same sort of goodness from the AI solutions that we're putting in place. I think for us, the other thing that we're doing is we're looking at cost effectiveness.

So that's the last thing that I want to touch on from an MVNO perspective. MVNOs, we of course offer multiple networks in multiple countries. And the most, important thing for a customer is how do I get market pricing in all of these countries right? It must be cost effective. We know that when we're roaming in different countries, oftentimes it's either not cost effective or we're not meeting regulatory requirements.[00:20:00] 

So from an MVNO perspective, it's really important for us to be able to optimise our networks and optimise the offer for the customer. To be able to use, a single SIM and go anywhere in the world and by utilising AI and other sort of business rules within that offer and within our network, we're able to optimise the network utilisation from a cost perspective, we're able to optimise the customers, network access to the SIM level also, in order to maintain those things, reliable connectivity and security.

So, all of it's really driven by the genesis of AI, and it will only get better. Today there is, of course, some human intervention in there, but from a network optimisation perspective and the SIM enhancements, those are all done automatically. And we are putting some of that automatic autonomous sort of Enhancements within our solutions.

Jim Morrish: Yes, absolutely.   As we've mentioned, you know telcos being particularly complicated in real time businesses, you know, there's a lot of [00:21:00] optimisation that can happen in those networks and it does lead to real financial results. I like to mention the, anomaly detection as well.

You can do quite a lot of that today. But, but just having AI agents, you know, sitting on top of these connections and saying, hang on, something different's happening here, it can be a real value add and actually highlighting problems or issues or changes in context, to clients that may not in fact be related to the solution. 

Rupa Datta: Exactly. Exactly. And I think the interesting thing will be down the road when we're able to, able to do the, if the, if that, then what, so if this anomaly is detected or abnormality is seen, then what, and the what can be defined by the customer, right? And then we can have some automagic decision making put in place to simplify everything. 

Jim Morrish: So picking up on the automagic decision making and it is not too much of a blue sky question, where's all this going in the future? What's the future of AI and telecoms? 

Rupa Datta: So really [00:22:00] around the future, like today we're talking about, you know, taking that crawl, walk, run. I think we're kind of at that crawl to walk phase and everybody's doing kind of the same thing. What we see again, I talked about network automation, I think what we're going to see is AI is going to increasingly automate that so that we can have some self optimisation within the network.

And again, reducing human intervention, Jim's fleet is doing this, someone has to do something on the back end, that's going to go away and that's going to be beautiful because all of that, again, to your point, the real time resolution comes to life. Right in a very tangible way. I think we're going to see AI at the edge.

So processing data closer to the source reduces the latency. And again, looking at real time analytics. So you're not waiting to do some sort of great analytics once your data goes over to the cloud, AI is going to do some, you know, a really good analytics at the edge while the thing is running so you can get the real time metrics off of it.

We do see [00:23:00] some of the enhanced customer personalisation also, which in a global economy and across different applications and industries, we see different applications require have different needs. So, some of them have different application behaviors and require a different reaction from the network. I have a battery powered device, I'm going to go to sleep for 24 hours and then ping you.

And I hope I have a network when I wake up, right? How do we optimise that versus I have a 5G, sort of video surveillance use case where I have to be on all the time. How are we optimising that? Because you're using so much bandwidth on the network. So I think all of the personalised services around what type of application do you have?

How does the network need to react to it? And then, you know, mobile versus fixed assets. Also, how does that network need to react to those different applications? I think that's really going to shift the game for customer engagement and satisfaction.

Jim Morrish: Yes, [00:24:00] so, moving television is much closer to the source, much more real time, much closer to real time responses, more personal and, and, and I hate to say as a service, but a network much more as a customised and personalised 

Rupa Datta: Correct.

Jim Morrish: Excellent. Well, it's been great to chat about AI and telcos.

But we ought to move on now to the promised “what the tech” segment of our podcast, where we talk about, interesting and or amazing stories in the world of technology. And I came across one last week, which made me both smile and frown, effectively is Rent a goon.

So there's an application, which has been released. It's a mobile app. And what you can do is you can, on your iPhone, obviously with an app, you can hire armed bodyguards to escort you around either New York or Los Angeles, complete with a chauffeur. So it's around about a hundred dollars an hour.

 And you can choose, you know, how these chaps are dressed and it just strikes me, I mean, how could that possibly go wrong? aside for the potential for [00:25:00] bodyguards to end up chaperoning drugs jobs and, or caught up in the middle of gangland warfare or something. But, did you notice that one, Rupa?

Rupa Datta: I loved that one, Jim, and I loved the fact that when you pick your bodyguard or goon, it's the same guy standing in the same way with different clothes on. So, it reminded me of that book when we had, when we were little, where we could have the guy and we could choose his clothes for work and play when we were like five.

So it was a little bit, you know, , it took me back to the younger days, but yeah, the guy doesn't look very friendly. I'll tell you on the app, but the alternatively, I think it's interesting that you have rent a goon here in US, LA, New York kind of thing. And alternatively in Japan, you can rent an elderly person for the day to help you go around Japan and eat the right food and take you on the train so you don't get lost.

How different is that?

Jim Morrish: Really?

Rupa Datta: Also done on an app. Yeah,

Jim Morrish: It's a different cadence of society, I think.

Rupa Datta: Absolutely!.

Jim Morrish: [00:26:00] Fantastic. So Rupa, it's been an extremely interesting discussion. Thank you for joining me and, thank you for supporting the discussion.

Rupa Datta: Thanks, Jim, for having me. It was really great.

Jim Morrish: Cool.

Rupa Datta: Thank you very much.

Jim Morrish: And with that, I think we should draw this podcast to a close. And just a reminder that you can subscribe to the Trending Tech podcast wherever you found us today. And thank you for joining us, of course. We're delighted to have you listening in, as part of our growing audience.

And we'll be back with another edition of Trending Tech soon, focusing on another aspect of digital transformation. In the meantime, please keep checking IoT-Now.com and VanillaPlus.com where you will find more tech news videos and much more. Thanks again for joining in. Bye for now.