From prototype to profit: Executive lessons on scaling agentic AI in CX

[0:00] IQ Era Intro (Video): In every era, there is a need, a spark, an instinct that brings us together to imagine and to innovate. This hasn’t changed. What’s possible has. We create technology that makes our world better. Ideas that move us, empower and connect us. But great ideas cannot thrive alone. They need to be shared, challenged, refined. Through our insight and collective curiosity, we advance and grow. New solutions, new directions, new answers. We do that here, together. This is the age of connected intelligence.

[1:23] IQ Era Intro (Video): Welcome to the IQ era.

[1:42] Dan O’Malley (Engageware): All right. Good afternoon. How’s the sound? Sound good? Louder? All right. Well, I will talk louder then. How about that? We, first off, I am Dan O’Malley. I am the CEO of Engageware, the most used customer AI platform for regulated industries like telecom, financial services and healthcare. We’re a globally scaled platform and have done over two billion interactions with our customers clients. We’re going to be talking about the end of the demo era of AI. An exciting topic.

[2:23] Dan O’Malley (Engageware): As a company who have worked with telecoms, including Net2Know, who’s joining us on stage for over 10 years, we have learned this, that AI vanity metrics are done and that the focus on real business value is here. In regulated industries, after all, you can’t deploy a demo. You need compliance and auditability from day one. So this session is a candid conversation with a telco, Net2Know, who has scaled AI to a great level and an analyst who has studied what’s worked in telecom for many

[2:57] Dan O’Malley (Engageware): years.

[3:03] Gilbert Miniones (NetUno): We’re joined by Gilbert Minionis, who is the chairperson of Net2Know in Venezuela, who has launched an AI agent called Bruno that has led him and Net2Know scaled their customer base by almost 250% while only scaling headcount by 15%. It’s really an incredible story. Looking forward to hearing Gilbert talk about it. We’re also joined by Diego Anasini, an analyst and vice president at IDC who studies the global telecom market. We’re going to talk over the next 25 minutes or so about three things.

[3:38] Gilbert Miniones (NetUno): First, we’re going to start with the end and we’re going to ask Gilbert to talk about the results, the end ROI that he has been able to achieve with Bruno. After that, we’re going to talk about breaking through all the barriers there are between the demo stage and actually scaling. And I think that’ll be a really interesting part of the conversation. And then we’re going to wrap up talking about lessons that we can take away. Also a little bit about the future.

[4:06] Gilbert Miniones (NetUno): So that’s our agenda. We’re going to try to have a little fun along the way. We were talking about that as we were prepping.

[4:12] Dan O’Malley (Engageware): And let’s start with quick introductions. Maybe Gilbert, you can lead us off. Hi.

[4:16] Gilbert Miniones (NetUno): Thanks, Dan. My name is Gilbert Miniones and I’ve had the opportunity to lead Net1 in Venezuela for the last 25 years. We’re originally a cable operator converted into ISP. Very much focused over the last five and a half years, deploying fiber to the home and abandoning everything that was HFC and adding value added solutions over the network that we have. Diego?

[4:49] Diego Anacini (IDC): Hi, I’m Diego Anacini. I’m with ADC. I’ve been for 20 years. I’m based in Argentina. And I oversee the analyst team in Latin America. And also I’m part of the global networking team with ADC that has been recently since January this year.

[5:08] Dan O’Malley (Engageware): Awesome. All right. Let’s dig into it. We’ll start with you, Gilbert.

[5:14] Gilbert Miniones (NetUno): When you launched Bruno, which was way back in September 22, so you had been operating it for a long time. Where was Netuno in terms of AI maturity before that launch? I

would say that before that period, we had extremely little AI use over the data that we had. The information that we had all came from traditional systems, how many phone calls would come in, how many calls you got, how many calls you could get. And we had a lot of data that we had from the Internet.

[5:45] Gilbert Miniones (NetUno): From traditional systems, how many phone calls would come in, how many calls you got, how many repairs you need to send out, how many truck rolls. They’re all in stovepipes. Everything was not integrated. And anything that we needed to do was coming in through different channels. So we had people taking care of chats, people taking care of phone calls, people taking care of WhatsApp messages, people taking care of SMSs. So it was multiplexed. Of course, the information we had was there, but it was not unified and it was not crossed-referenced, let’s say.

[6:17] Gilbert Miniones (NetUno): So we had no way of really understanding what was happening in a global form. By the way, our company today is 300,000 subscribers and we have 1,300 people. So just so you can size a little bit the company of what we’re talking about. So things were a little messy. Channels were separate, disparate. Was there one moment that made you say or one problem that made you say, like, we need a customer AI agent to help solve this?

[6:52] Gilbert Miniones (NetUno): Yeah. What happened is for those that are in the telecom world, the introduction of FDTH and dramatically larger speeds and responses and customer experience dramatically increases. I had launched FDTH in Brazil a few years before and it just took off. So I knew that we were going to launch it in Venezuela. That was the same behavior it was going to have happened. And we were not really prepared to handle two and a half times or three times,

[7:26] Gilbert Miniones (NetUno): a number of subscribers in a short period of time. Therefore, we needed to really prepare the company. Otherwise, you’d just have to triple everything and you wouldn’t have any profits. So that was the core driver. So after we introduced the agent, we developed Bruno, we worked, we trained it for about six months and we began implementing different functionalities. And today, we’re very happy to say that we have 64 functionalities in use, extremely successfully. And one of the interesting things that we have with the Gager is that the system is organized around modules.

[8:14] Gilbert Miniones (NetUno): They’ve got some like 13 modules. And of these modules, we use 10. Most people use one, two or three. They take very, how do you say, careful steps. Because of the urgency that we had in place, we tackled many things simultaneously. So we really jumped on it. And if any things would come around, we would just focus and put resources and fix whatever needed to be fixed. So the implementation was very smooth. We didn’t have any serious surprises anywhere.

[8:55] Gilbert Miniones (NetUno): And today, it’s incredible. About nine months ago, we decided to launch a new product, a new value added product on our system. And we negotiated the launch with a supplier with about eight months. And over a period of three weeks, we prepared Bruno, our agent, to be able to handle this. And in three weeks, we were up and running with it. So it is extremely easy. Once you become familiar of how to use it, implement it, the time to market is incredibly short.

[9:31] Gilbert Miniones (NetUno): And so we’re extremely happy. The results we have are incredible. I don’t know if we’re going to talk about that later. We want to talk about it now. Why don’t you go ahead and hit it? It says, so these are fairly important numbers, which to us are critical. And I wanted to share that. And of course, not everything has to do with Engageware, right? A lot of it has to do with the product that we’re launching.

[9:54] Gilbert Miniones (NetUno): But we wouldn’t have these results if Engager were not there. So we increased subscribers over the last three years by a factor of 225%. So it’s 2.3 times X the number of customers. Today, we have 300,000 customers. We had something like 100 and some of thousands of customers before. Revenue grew 250%. And our payroll, the number, the headcount, the person’s increased only by 15%. So imagine almost tripling the number of customers with only increasing the number of persons

[10:28] Gilbert Miniones (NetUno): taking care of the triple load by 15%. Are you at that margin? And that’s very much the result of the product increased by 13 percentage points. So we’re typically in the order of 30 plus or under 30% even that margin. We jumped at the 40 plus percent even that margin. So it’s a transformative technology that did things for us. And it was able to be implemented thanks as a result of this. Our phone calls dropped from 63% of the contacts that we have with us to 2%.

[11:04] Gilbert Miniones (NetUno): Our digital interactions went from 37% to 98%. So we used to have a chat bot, whatever, it’s gone to 98%. Bruno today handles 84% of our interactions autonomously. That means it takes the request, whatever field it came from, adding a new server, just making a payment, doing whatever you want to do. 84% completely gets closed, the interaction by Bruno with no human interaction at all. We always have the ability for the customers to request a human.

[11:48] Gilbert Miniones (NetUno): We think that this is one of the critical success factors in the adoption because people get fed up not being able to speak with a human. Some people just don’t want to do bucks. So from the get go, we tell them anytime you want to speak with a human, just let us know and it will be transferred to a person. But as the years pass, people realize this thing really resolves things. And today, 84% of the interactions get handled by Bruno directly.

[12:14] Gilbert Miniones (NetUno): These sessions went from 84,000 to 350,000. That means more people are doing more things. Reducing services, reducing speeds, increasing speeds, whatever is required. Office visits drop by 50%. So people no longer come to the office as much as they used to before, less squares, feet, meters, whatever you measure to be present there. Our response rate on the chat is 97%. Of course, it’s 24.7%. And like I said before, we have today 64 functionalities. Today, you can pay with two clicks through WhatsApp.

[12:59] Gilbert Miniones (NetUno): So you don’t have to go into a website, put in your account number, your password, none of this. You WhatsApp, Bruno, I want to pay. It recognizes who you are. Boom, you pay your input. Two clicks on WhatsApp, which is worldwide, probably not in the States. The States is a lot weaker in its adoption of WhatsApp. They like using SMS. So we have it on SMS as well. But normally in the rest of the world and in Venezuela, WhatsApp is the most used application

[13:29] Gilbert Miniones (NetUno): for communications. So those are tangible results that go straight to the bottom line just by integrating. You can reach Bruno through all the channels, SMS, Facebook, Instagram, WhatsApp, email, whatever system that you feel comfortable in using, they are all integrated.

[13:58] Dan O’Malley (Engageware): Awesome. Awesome. It’s a very nice story. Obviously, a very successful deployment of AI. You said a couple of things. I’ll just put a pin in for now. We can talk about later before we switch to Diego. The autonomous nature of your agents, like autonomous agents, we probably had 10,000 conversations about them here at MWC.

[14:20] Gilbert Miniones (NetUno): The fact that you are 84% autonomous with Bruno is outstanding. Got really a testament to the work your team put in. So we’ll talk about that more in a bit. Don’t worry.

[14:32] Dan O’Malley (Engageware): In the US, RCS is coming and we’ll be able to provide that same experience your customers have in the US as well. Couldn’t help mentioning it. Diego, let’s turn to you. You’ve analyzed the performance of many, many telecommunications companies. What challenges do you see when telecoms are looking to deploy AI today? Great. I would like to first put in context the challenges around AI.

[15:02] Diego Anacini (IDC): IDC performs surveys on a regular basis, an annual basis around what is the agenda of organizations in general. Of course, telecoms are among those. When asking about the different challenges around AI, the lack of trained professionals stands out. So the need to partner with strategic partners comes in a very important stage here. Then scaling AI around different areas and different initiatives. That is another important challenge that organizations as a whole face right now. And then the concern of using internal and private data to train third-party models.

[15:51] Diego Anacini (IDC): So those are the three top concerns, the three top challenges for organizations. If we add the challenges of using agents, AI agents, the concern about the return of investment is also very important for organizations and also the data in the agents’ governance. That is a very important topic that organizations are looking to work on in terms of challenges. And when we talk about telcos specifically, we could add the complexity. Complexity in terms of organization, complexity in terms of technology.

[16:38] Diego Anacini (IDC): Then there are cultural items like change management and the alignment of the different initiatives with the business goals. That is another important challenge around telcos. And then the need to modernize the architecture. To think on more open architectures to be ready to adopt AI agents and other type of intelligence. That’s an awful lot to think about. An awful lot to work on. How do you coach or

[17:17] Diego Anacini (IDC): In this space, we suggest to break the different use cases and separate them and group them first in terms of the impact. So there are use cases that will have a short-term impact in productivity like sales, marketing, and also in software development. And there’s also a second group where we could put the use cases that are more strategically or will impact on a disruption in the business. So that would be the second group and that would be new product offerings that what you

[18:02] Diego Anacini (IDC): were mentioning, customer service and those type of use cases. The second activity that would be recommended would be to create timescale. To put those use cases in a different time horizon to address them in an organized way. So first, we would suggest to put the use cases that would create an incremental impact in terms of productivity. In those cases, we would set the base for the next stages. Then a second stage would be to put use cases and this would be two or three years horizon.

[18:45] Diego Anacini (IDC): The use cases that will have focus on disruptive innovation. So that will also create or will train the organization to face the third group that would be those that would be placed on a strategic innovation. So first, incremental innovation. Second disruptive innovation. Third, strategic impacts. That would be the recommendation. So figure out what initiatives might fall in each category and making sure you’re pushing on all three really.

[19:28] Dan O’Malley (Engageware): Gilbert, you had a remarkable success with Bruno, but I’m sure it was not easy to get done.

[19:35] Gilbert Miniones (NetUno): What were the challenges that you faced in pushing Bruno out? Well, you know, when you tackle the implementation of such a broad area that touches many parts of the company. Because once you interact with your customer, you’re making installations. You’re delivering trouble tickets. You’re analyzing counts payables. You’re doing multiple aspects. The coordination of all those elements and making sure that everybody understands that this is the driver of the company. It drives your ticketing. It drives all your statistics on how quickly do you resolve the issues, etc.

[20:16] Gilbert Miniones (NetUno): Is always a challenge. So we need, honestly, I was quite surprised at the adoption. And of course, as usual, when you have a success story and you’re growing, that helps people accept things. Most people associate everybody’s scared. You’re going to implement AI. You’re going to lose all these employees, blah, blah. That’s not our case. You know, we had clarity that we’re in the growth mode and that we needed to have a truck that we were going to suffer a tremendous amount of stress in the company if we didn’t

[20:58] Gilbert Miniones (NetUno): have the right tools to be able to implement this growth. So it was extremely quickly adopted by everybody. So the support was everything. We had a little more resistance with people, the customers using it at the beginning than we had expected. But very quickly, word of mouth is the best advertiser. One customer tells to another, you know, I tried Bruno today and I was able to make the payment quickly. Oh, and I don’t have to speak with somebody.

[21:33] Gilbert Miniones (NetUno): I don’t have to wait for anybody online. I just did this. And slowly things, you know, you build momentum and it grows towards you. So those initial challenges that we’re very fearful about were overcome reasonably quickly. And as we, you know, go down the list of things that you went to implement, like Diego said, you know, you go from the immediate needs to the ones that are more strategic over time. And right now our next stage, of course, is to try to take all the data that we have across

[22:06] Gilbert Miniones (NetUno): and use it in a more predictable form. So our next stage is to instead of just capture and improve efficiency, is to be able to predict when things are going to happen so we can be one step ahead of them. And that requires a lot of data analysis, of course. But the data is now there. We now have three and a half years of information. And that makes it a lot more secure that the information that you have is not just somebody’s

[22:33] Gilbert Miniones (NetUno): opinion, but it is backed by true information. A tremendous opportunity to keep improving the quality of the agents that service your customers. Yeah. Our company has always been focused on enchanting the customers. We’re a small operator compared to the big guys. So we know that our differentiator has to be excel in customer experience and excel in technology. So we deal with the huge guys that have deep pockets, they can do whatever they want, you know, they can advertise.

[23:06] Gilbert Miniones (NetUno): So our differentiator is to enchant the customer. We’re not okay with satisfying customers. We need to make them be in love with us. We believe, the whole company believes that if a customer is in love with us, they’re not going to look around the sites. They’re not going to have a frequency to listen to other pitches. And that’s how we’ve been able to survive. Yeah. I love it. Obviously, your customers do too.

[23:38] Dan O’Malley (Engageware): I’ll ask you a quick response. If you could give one piece of advice to somebody who is about to start rolling out customer AI or is like rebooting a customer AI that they rolled up, it didn’t work. What would be the one piece of advice that you would give them?

[23:55] Gilbert Miniones (NetUno): Do more than you thought you were going to do. Do more. If you thought you’re going to pilot one or two items, pilot six items. Jump in the pool. Don’t sit there and a little toe in the water, two little toes in the water, a foot. You dive in. This is no turning back. The whole world is going that way. The faster you go that way, the more competitive advantage you’re going to have versus your competitors.

[24:28] Gilbert Miniones (NetUno): I love it. You rolled out many use cases. You rolled out many use cases. 60 integrations, I think, between Bruno and your systems.

[24:41] Dan O’Malley (Engageware): Diego, talk to me about the use cases that matter most that you see in Telcom.

[24:47] Diego Anacini (IDC): IDC measures the AI and GenEI use cases around the world. If I were to rank the top ones, I would say that there are two very important topics. Very engagement and risk reduction. That is mantra for AI and GenEI among organizations. Of course, Telcos are not the exception. Use cases go around, they go around customer service and self-service, the contact center, guided selling, and also fraud detection and analysis. In terms of Telcos, those are also very important.

[25:33] Diego Anacini (IDC): I would add the AI provisioning. Telcos use AI to provision themselves for AI. All that has to do with threat detection and analysis. Telcos are also very concerned about cyber attacks and how they would impact their customers. Those are the top ones. Would you agree with the use cases?

[26:07] Gilbert Miniones (NetUno): Yeah, except we are a risk-loving company. We are not risk avert. We are the opposite of risk avert. I think that if you are small, the world belongs to the daring. You have to be in the front. In such a challenging environment, in such a fast-moving environment, like the one we live in these days, everyday things is faster and faster and faster. More people know. If you are not at the front of being the changer of things, people are going to change you.

[26:43] Gilbert Miniones (NetUno): If you get changed by somebody else, you are reacting. You are not acting. Our organization is focused on acting. On developing, knowing where is the future. We think we understand a little bit of the market and our customer needs. We try to be upfront. Just to give you an example, where the telecom is. The last product we launched is a health product. What the heck is a telecom company selling a health product? Well it so happens in Venezuela that health is in dire straits.

[27:19] Gilbert Miniones (NetUno): Things do not go well. When you look at what a customer values most, health is up there in the list. If you feel sick, you can’t go to a social security hospital or an urgent care area. We teamed up. We did a strategic alliance with an urgent care company, private company, telemedicine, and home care if there is an urgency as a first touch. So you say, is this important? Well, it has nothing to do with telecommunications, but it is one more thing that you will deliver

[27:59] Gilbert Miniones (NetUno): to a customer that they really value. And at the moment of changing, when somebody else gives them a pitch, I am now selling you 10 gigs for two cents. I can call a doctor with these guys. So I am going to stay with these guys. So to me this is critical. We love risk. We have a lot of faith and trust in our personnel. Our people are aligned. We work our values extremely intensely. Our core values of taking care of our people.

[28:41] Gilbert Miniones (NetUno): We have an ability to move very quickly. And therefore, we jump on things and we do them. So let’s keep double clicking on risk. And let’s talk about autonomy. So we have talked about the fact that 84% of Bruno’s transactions happen autonomously. That includes payments. That includes service changes. So you could literally turn off someone’s service accidentally or lose their money if Bruno doesn’t act appropriately. That’s significant risk. How did you and your team get comfortable with how to manage that risk with an agent?

[29:18] Gilbert Miniones (NetUno): Well, you know, it is very easy to talk about what we do. But for that to work and be comfortable with it, there’s a whole bunch of other things underneath that one doesn’t see. So the papal traders are there, the audits are there, there’s areas that check to see things are there, the conciliations are there. We have phone calls. I mean, people make calls on a human cause on a regular basis to confirm that things actually occurred.

[29:55] Gilbert Miniones (NetUno): You know, all the bird formulas on probabilistic and so on. So we think this is so much probable. So you make calls to the equivalent of the risk that is possible to happen. And that’s the way we get. We got running the assurance folks. We got all these things that get compared in the data that you end up collecting. And some things become distorted. You go take a look at them. Yeah. Yeah.

[30:25] Dan O’Malley (Engageware): Diego, why don’t you give us some perspective. First is Gilbert’s experience and actions normal for the industry. And then why don’t you tell us about general best practices for kind of managing risk, especially around autonomy? Yeah.

[30:39] Diego Anacini (IDC): I would link recommended perspective on the different challenges that we discussed. So to overcome agents, adoption, I would say that the modular yet customized platforms are very important. There’s adoption support beyond technology and also the complexity and the organization management and change management. The orchestration and the reliability of the platforms is very important to overcome risk. And also very important, last but not least, align the technology and the platform with the business results that are expected to overcome.

[31:33] Dan O’Malley (Engageware): We are just about out of time. I’ve got like a million more questions that I want to ask you both. But I’ll just wrap up. Thank you for all your thoughts. Much appreciated. Great conversation. And we’ll do another one of these next year and share the updates. Great. Thank you very much. Thanks guys. Thanks guys. Thanks guys. Thanks guys. Thanks guys. Thanks guys. Thanks guys. Thanks guys.