T-Mobile’s Business Push, Technology Advantages, and 5G Slicing with Daryl Schoolar

Full Transcript

0m10s Speaker 0

Hello, and welcome to the two hundred and sixteenth episode of the week with Roger, a conversation between analysts about all things telecom, media, and technology from Recon Analytics. I'm Brett Clark, and with me as always is Roger Etner. How's it going, Roger?

0m23s Speaker 1

Hey. It's going great.

0m24s Speaker 0

Good. And also from Recon Analytics, we have analyst and director Mitch Clawson. Welcome, Mitch.

0m29s Speaker 2

Oh, thank you for having me on.

0m31s Speaker 0

And lastly, our guest of honor this week is Anthony Guna Telica. Anthony is group president, technology, and head of strategy at Amdocs. Anthony, welcome to the podcast.

0m41s Speaker 3

Thank you for having me, guys.

0m42s Speaker 2

Anthony, thank you for joining the podcast This Weekly with Roger. Back in September, I had the privilege of coming to Amdocs for the Up Close Summit, and thank you for the invitation for that. And it was a great event. During this summit, there was a central theme around AI. Um, this is currently a lot of hype around this in the market.

1m1s Speaker 2

With all this hype, where do you see we are really in the AI adoption cycle? And how do you see this evolving over the next two to three years?

1m9s Speaker 3

You know, I think you're right. We are, I would say, smack bang in the middle. And depending industry you talk to and who you talk to, I think they'll give you different spots in terms of where they would place themselves. I would say, look, this is probably the most material technology change we've seen in decades. And it's not because there hasn't been any other technologies that hasn't had a large impact like this, but it's the acceleration, it's the expediency of how you can deploy this technology and use it and get results pretty fast.

1m43s Speaker 3

That's really, I would say, been the game changer. Again, depending on who you ask, if you ask us where do I think we are in this cycle, I would say we are like a little bit kind of past that huge acceleration. And maybe right now, I don't want to say when the trough of disillusionment, but I would say if you ask a lot of customers or a lot of CFOs, for example, everyone is going, Okay, fantastic, great technology. I see what you can do. Show me the results.

2m12s Speaker 3

Right. Like what a famous movie like, show me the money. Like, this is the time. Right. So I think the whole industry is in this spot going, Okay, fantastic.

2m21s Speaker 3

Love the technology. We see the results. We see what you can do. Now, show me whether it's revenue growth, whether it's driving efficiency, driving acceleration in what you do. Show me that translated to whatever my business KPI is.

2m36s Speaker 3

And that's kind of the area we're in, I think, in the next twelve months where everyone is like, you know, we've kind of gone past the experimentation, the innovation, almost like every group is doing something generative, even internally within our company. I just had a meeting today and our head of marketing was kind of presenting his AOP plans. And, you know, like 75% of his meeting was dedicated to all of this stuff. He's using generative AI and utilizing it. So I think everyone has embraced it.

3m6s Speaker 3

But I think the next twelve to eighteen months is really about showing business results that impact the KPIs that the business is focused on.

3m15s Speaker 1

So, Anthony, could you remind our our audience who might not be as familiar with Amdocs as we are, what you are actually already providing to your clients in terms of GenAI support?

3m32s Speaker 3

GenAI for us, I would say, is split into a couple of different categories. Primarily, all of our products are now rolling out starting from January. 2024 was rolling out with all sorts of copilot assistance with our products. So you get out probably you get something like the enterprise catalog. You don't have to even know how to use the application.

3m53s Speaker 3

You have a guided copilot that will give you information. You'll just type it in your natural language or even even say it through a voice input and we will kind of create the marketing plan that you wanted to accessing all that type information. That, I would say, is the first part of it, which is around the copilot for our products. The second part, we call it kind of type B, but this is really the agentic experience, right? This is really the next generation of employees.

4m20s Speaker 3

This is the next generation of agents. And we have seen some tremendous results like we shared in our last analyst conference. You look at operations and call centers and things like that. Like we can reduce impact your primary KPIs by double digits, not just double digits in the low sense, but double digits in the 50%. So things like average handling time, first call resolution, transactional NPS, you know, all of these criteria can be impacted with a single technology, which is really what really excites us.

4m51s Speaker 3

So this is, I would say, the second part. And the third part is really what our CEO calls the infrastructure or the plumbing. A lot of this stuff still needs to be done, right? So you need to normalize your data. You know, your stuff is only as good as what you put into it.

5m5s Speaker 3

At the end of the day, you need to make sure you have all the integration to your real time systems. You need to make sure your governance is in place. You need to make sure that, you know, you have all your security criteria in terms of which large language models you're using, how are you using it? All of, I would say, the hygiene stuff. That's kind of the third part.

5m23s Speaker 3

And so we kind of come to the table with kind of one, two and three. And of course, customers are picking and choosing and taking different flavors of the three.

5m32s Speaker 1

You know, and our frequent listeners are probably sick and tired of me saying that we run the fastest, largest, most agile customer inside service, both on a consumer and business customer side. And what's really interesting, we are measuring component NPS in a like 16 to 20 dimensions, including customer service. What we found, when telecom providers are announcing that they have rolled out Gen AI assisted customer service, in the first month, maybe two, the customer service NPS drops. But then it accelerates out and gets better than it was before. Because the Gen AI has learned from a lab to being trained on real customers.

6m20s Speaker 1

Do you see the same kind of things that we are seeing?

6m24s Speaker 3

Yeah. No, we we don't actually see that. And I can give you the reason why. I mean, the last production pilot we rolled out, like we saw positive GNPS results from day one. And the reason for that is our models are pretrained, right?

6m36s Speaker 3

So we are in this industry for forty years. We know the data. We know the systems. We are fine tuning it. We are pre training it.

6m44s Speaker 3

We know the what the expected results should be. So we are not we are not experimenting, right? Like we're not putting it out and looking at the results and looking at if it's acceptable, not acceptable, and then bettering ourselves. We are already coming to the table with, like, I would say, pretty much a level playing field by utilizing all the knowledge we have, the systems, because we build products, right? We're not just a services company.

7m6s Speaker 3

So we bring this deep product knowledge that's already incorporated into our large language models. So I think that allows us to achieve results on day one rather than having that kind of catch up time where you're doing inferencing and training and where the models get up to speed.

7m22s Speaker 1

That's very impressive.

7m24s Speaker 3

Yeah. We're excited about it. It's a game changer for us.

7m28s Speaker 1

Do customers actually realize that they are interacting with Gen AI?

7m33s Speaker 3

The pilots and the production trials we're running at the moment, what customers choose to do is not enable it directly to the consumer or the enterprise. What they will do is they will still have an agent in between and the result is given to the agent and the agent will convey it to the customer. It's just that added level of safety and security before they go full ball and let it out. So either way, the consumer has no idea. But here's the difference, right?

8m2s Speaker 3

So we have probably, you know, in all of the inferencing we've done on data, out of every call that comes into an operations call center, probably about 30 something plus percentage is because of some type of bill inquiry. You know, what's this charge? My billing is too high. I didn't have this charge yesterday. My bill is different.

8m21s Speaker 3

Some type of pricing billing inquiry. What we are kind of able to do is when we benchmark agents responding to this, it takes someone between twelve to fourteen minutes to respond. And they give some type of answer. Why? Because they need to go into their past history, look at what happened, look at what charges were put on, look at what proration dropped off, look like if there was an early termination fee on a device.

8m44s Speaker 3

They need to kind of comprehend all of this information and then provide an articulate answer. We can do this in forty five seconds and provide the most a better articulated answer than any human can provide. But we have this human in the middle just to kind of validate, make sure there's nothing crazy going on. And they will just read the answer to the consumer and we get the resolution done in under two minutes. So you take a fourteen minute transaction delivered in under two minutes.

9m12s Speaker 3

But not only that, when you benchmark the NPS, you get a better NPS score. Why? Because the answer is very articulate. It is empathetic. It responds to exactly what you've done.

9m23s Speaker 3

It's clear. It's the same versus you can imagine, you know, call center employees. I mean, there is a it's like hot seat sometimes, right? People are being changed out and stuff like that. So the level of answer you get from a human is really it varies.

9m38s Speaker 3

Like, there's a huge spectrum. So you get consistency, you get better results. And of course, you know, the future dream is to give this directly to the consumer.

9m48s Speaker 1

I use about $10 per minute of customer service time as my rule of thumb of cost. Yep. Going from fourteen to two minutes is a major cost savings for operators. And the consistency is priceless.

10m5s Speaker 3

Yeah. Yeah. Definitely. I mean, this is why I say it's a game changer. Right?

10m8s Speaker 3

Usually, when someone initiates some type of business initiative, you know, they would have an impact on maybe one of the key KPIs or two of the key KPIs. But to have an impact on all three primary KPIs is very rare. And here we have a technology and a solution that can have that impact. And that is really why, you know, all the way from, you know, we spend a lot of time working directly with Nvidia on a lot of our roadmap where they're putting stuff in and we are including stuff in our solution. I mean, they are telling us, you know, this kind of agentic experience is a game changer.

10m47s Speaker 3

And I really think we are going to see look, everyone needs to reduce cost, right? I mean, it's no secret. Driving efficiencies in operations is like the number one goal in most telecommunication companies today. In order to do that without reducing a service level or without depending on, hey, I have to offshore this, you know, getting subpar service or whatever, or different time zones or anything like that, Being able to provide an output that is better than you would have normally provided in a shorter time with a higher satisfaction, this is the exciting part.

11m21s Speaker 1

Yeah, no. You know, we analyze the operator's financials, and some operators like AT and T this quarter, all of the increase in revenue in mobiles dropped straight down onto the profit line. Yep. I don't know if you had anything to do with it, but the cost containment, the pressure there across all operators is intense, right?

11m45s Speaker 3

Yeah. I don't think it's any secret. And this is a global phenomenon, right, across the communications companies. I think they are looking at how do you positively impact your balance sheet, how do you kind of change the model. Because it's a very it's not that it's a bad industry, but there is a very heavy capital cost structure on the industry, right?

12m6s Speaker 3

You have to invest in five gs. You have to invest in fiber. And you don't do it once and then live for the next ten years to just recoup it. Like every three or four years, you need to do equipment upgrades. You need to roll out bigger pipes.

12m20s Speaker 3

I mean, the capital cost structure is so intensive that you need to find other ways that you can be more efficient. And so this is why I think the two dovetailing.

12m29s Speaker 1

And I think the industry is getting a bad rep. Show me an industry with you know 50% gross profit margin and I show you a healthy industry.

12m37s Speaker 3

Right.

12m38s Speaker 1

And every time we show them growth like FWA, it gets like poo pooed and like yeah, but that doesn't matter.

12m45s Speaker 3

Yeah. Right.

12m46s Speaker 1

And it just doesn't matter. So all the good things that the industry does gets pooh poohed. It's like people don't want to be happy about this industry. It touches everybody's lives, and typically for the better, you know, and companies like you opt to make it even better. When something goes wrong, companies like you, with your GenAI services, make this all better.

13m11s Speaker 1

We all know that there's room for improvement.

13m13s Speaker 3

Definitely. I think this is the look, and some of the stuff that we're doing interesting interestingly enough just translates across verticals. We are doing a pilot in financial services at the moment. You know, the interesting thing that we are finding as we train these models and fine tune it to the agentic behavior is humans are humans at the end of the day. Yes.

13m35s Speaker 3

Like, I would say it's a lot more about behavior interaction. And when you're calling in regarding a service, yeah, I mean, there are some nuances that are related to industry, but a lot of these things can translate across industries. So this, you know, I think, is another opportunity for those that succeed in this space.

13m54s Speaker 1

Oh, absolutely. We all interact with customers, and they typically don't call for happy reasons, right? Yeah. To tell you how happy they are with the service, but that something went amiss. And, you know, it might be financial services, it might be online services, you name it, and your solutions can certainly help there.

14m15s Speaker 3

Yeah, and the other, I would say the other big change, like if I go back, you know, a year or a year and a half, I think is the level of accuracy. I mean, we are achieving at the moment around 96.4% accuracy in production. I would dare to challenge anyone that a human can achieve that level. Because if you think about making mistakes, policy mistakes, not understanding what you can tell the customer, or the number of times, and I'm not talking about just telecommunications here. A number of times I've just called someone saying, hey, like this is charge and they're like, oh, Okay, yeah, I don't know what it is.

14m51s Speaker 3

Let me just refund it or let me just offer you a discount next month just to get me off the phone and do fiscal resolution without really understanding or knowing what's going on. So I think there is like some added benefit that comes here that you're just communicating with your customers better. So this is a fun part about it, right? Like this is also a tool Like, leave aside the efficiency part and all the cost savings you get. This is also a tool to potentially take your customer experience to the next level.

15m20s Speaker 1

Well, with that, I think we're running out of time. And I really appreciate you being here and telling us about Amdocs and and Chen AI.

15m30s Speaker 3

No worries. Glad to be here.

15m32s Speaker 0

Alright. Great. Well, thank you, Anthony, for joining. Thank you, Roger. Thank you, Mitch.

15m36s Speaker 0

And I hope everyone has a great week.

15m38s Speaker 3

Thank you.