How BCG, Naturgy, and Konecta Are Deploying AI Agents in Production | ElevenLabs Summit London 2026

[music] [music] Hi everyone. How are we doing today? Good. [cheering] Pretty nice. It's amazing to see London being our global headquarters. So many nice and good faces, familiar faces here. Thank you again for coming. Um we have a unique session right now because what we wanted to do is like we we see that like like AI is getting adopted in the real world and who best to actually bring three like leaders in different industries to actually tell us more about that. Um this is not about the level labs we're in levels event it's more about like what is actually happening in the world and that's what for me is like the most interesting piece of it. Um Juan welcome thank you for coming. Juan leads um AI, generative AI um across BCGX. Um he's he's fantastic. We've been working together for over a year. We also have Terrell who leads like um um AI across Nuri, 18 million customers globally, 24 countries, one of the largest utility companies across across Europe. Oscar, thank you very much for coming as well. Oscar heads generative AI for deployment for connector the six sixth largest uh BO in the world. Um thank you guys again for coming. Um for me it's important to start like uh on Monday we announced our collaboration strategic partnership with BCG. It's been a fantastic honor to actually work together very closely over the past months. Um, and it's one of the key things Mati was talking earlier about the fact of like going deeper and helping the world deploy AI and agentic systems. And that's what we think like collaboration between both both companies is super key because you have the expertise across a number of industries uh big small customers all of that stuff. We have the technology. So there's really good synergies in there. Thank you so much for picking us um and and helping us. Um Juan um what are the trends that you see in the industry? Well, you know, first of all, I think we are entering into amazing times. You know, people are really and organizations are adopting Genai and Agentic AI. Uh we are seeing numbers above 40% in terms of adoption. But companies are doing basically three things. Okay. companies are implementing final users applications like you know transcribers uh summarizers for the dayto-day of the employees. The second thing that we are seeing which is a heavy trend and 80% of the companies doing AI are doing is reshaping an entire business unit transforming the customer service transforming the work the the way their engineers work uh transforming the way they do digital marketing taking a process reshaping it completely um and identifying it. And the third trend that we are seeing very preliminary but also um with huge potential is the reinvention you know companies that are trying to find new revenue streams even creating green field autonomous organizations from scratch. So these are the three things that are happening in organizations today. Huge penetration in reshaping to capture quick value. I really like it. Um Oscar like what are you seeing like you lead one of the biggest like deployment for one of the biggest like uh BPOS in the world like what are the trends that you're also seeing what are the industries where you seeing highest adoption tell us a little bit more uh so we actually seeing uh quite a lot of uh transformation across the board industry wise we are seeing a lot of work on uh industries that are process heavy that are uh that have complex activities that they need to go through activities around um banking and uh and insurance for topics around back office processes. Uh also a lot of engagement with customers uh on the front end retail quick service restaurants for order taking or for some of the processes that are heavy and repeated processes are also something we are we're observing a lot from a uh tech from a region perspective we we saw a surprising uptick in Latin America as one of the hot beds of innovation. It's been a lot of pressure from the customers coming to us first say hey I see that this is happening I want it now with all the complexities also from a language standpoint a lot of the models started very heavy in English with a little bit more of complexity on on latam languages so it has been that's been an interesting evolution and we are seeing a lot of growth also in Europe and in the rest of the regions as we as we go with a lot of pressure on the compliance side so we have a lot of ad hoc processes and hog ad hoc products that we can deploy. The reality is that enterprises need an enterprise ready solution to have in place and that allows them to manage, orchestrate and and uh and control whatever is deployed in front of their of their customers and what they can what they can have. It's so funny actually like I was thinking uh and we're going to go off script like all the time. So but I was actually thinking like there is like multiple like velocities in deployment across the world right and I would love to like hear your thoughts Juan as well fundamentally because like what we seeing is like Latin America, India specifically some countries in Latin America, India like Spain, UK, Germany like they're adopting like Nordics are adopting generative AI much quicker than many other countries right so you end up having the two tiers. Is it something that you're also seeing Juan? Yes. Yes, we are seeing and also the things that they are doing are different. For example, um in Asia, we are seeing a huge innovation in terms of automation. They want to manage the customer interface. You know, huge adoption of customer interface management. In United States focus is mostly in two main dimensions. Accelerator acceleration of their engineering departments. We see heavily that. And the second thing it's augmenting the capabilities of their professionals like real-time copilots solutions that support the telephonic agents or the chat agents. So geographies are doing this because they have different efficiencies and also different cost structure. You know it's very different the cost per call in India which is near 25 cents to the cost of a call in US which is near $3. So that also changes the equation of where you will invest you know the next AI dollar you know. Yeah 100%. Terrell, how are you deploying agents across? Yeah, I mean we we've been having a a kind of a a we think it's a unique approach because we started this process back in 2021 where AI wasn't even in the even heard of and and we basically redefined our IT entire IT stack and and we and we we started a transformation process with uh people first because we wanted to be the best customer service uh utility in Europe and and we still that's still our goal. Uh mainly because that's that's where we think our differentiator in the market lies, right? And and so this um this has allowed us two things. First is to have people on board from the beginning because they know that we're in this transformation project and and and in in this journey and and it's either we do it or we are out of the picture, right? and and second is that our our systems are AI ready from from from the beginning. So then what we did is we started doing a lot of pilots and and and PC's uh to make sure that we understood what what we could achieve with with the technology and and we know the right partners to to go along in this journey because that's also uh pretty pretty important. So right now we are we are on a on our beta uh almost and we're almost ready to go into production and put this uh solution in front of our customers which is something that we're really excited about. One thing that we see a lot and I would love to get your take, right? Is is the fact that like there's like a very like like sea level wants to deploy generative AI across a number of areas, right? And you're saying like you like you're now going live, you've tested and so on like but there's like two sides of this equation. One is like what we want to do as enterprise or as businesses. The other one is like the team that we have supporting us as a business, right? How is the team transitioning? How are you helping the team transition so that like they all like are engaged with this big transformation that is happening in the business versus against it? Yeah, I I think it's it's it's really important to have uh sea level of sponsorship on these projects because and and when I mean sponsorship is not just because they they want to have a press release and say that you're working with AI. is because you really understand the the technology or it's this the sea level executives understand the technology and where the value lies. Um that that's that's pretty important but also uh to get everybody on board from the beginning was was really important because uh people tend to see AI as as a threat uh at the beginning, right? and and and the way that that that they lower that that nervousness around AI, I think it it it tends it works when you're working with it, you understand what it's capable of and what your relationship with it will be in the future and and that and and that makes you a believer in the technology and and that's what what the our journey has been in the last uh 12 plus months. Makes perfect sense. Um you brought us a demo, right? Yeah, I did. tell us. Uh well uh we're working with B VCG and and and and with you guys on on on as I said our our beta test. I just did a very quick call because we don't have that much time and our calls uh are are tend to be pretty pretty long but I have a demo that that that I want to show on on how you can how the interaction with this voice agent will be uh in the very near future for our customers. Excellent. Let's play it. foreign. No. Solo installation. No. Yes. Let me tell a detail of this because I think it's very interesting which is we've worked with 11 labs also and they they they've supported us in something that was very new to us which was human voice cloning. So the voice that you are hearing um we did an exercise of recording 1 hour and 30 minutes of a person. Next day we had a humanlike voice that the one that you are hearing in Spanish. Yeah, that that's that's one of my a person from my team at Natur that it's her voice and and we did it just to test as as Swan said the the the voice calling capabilities of of 11 Labs which is pretty impressive and as I mean it was really excited the I I I didn't do because I didn't want to make a very long uh conversation but this this demo is is also uh grounded on our system so it can it allows us to do on behalf of the customer and because of how our systems are built, everything that the customer can do in his uh private area on the web page, the the voice agent will be able to handle as well. So that that would be a pretty exciting to to SCA to deploy this and and and achieve we think a a very a very impressive customer service. I think like uh so uh uh like Antropic put it really well uh in Davos where like last year it was all about proof of concepts for the big enterprises right this year is about a proof of impact right it's like how are you generating actual value for both your customers and for the organization so I think there's been like a massive shift like towards like showing it because everyone is testing it but like who has it in production how much actual ROI you're getting on it right um What is the biggest issue Juan that you see with organization trying to deploy uh agentic systems? Well um there are three things that organizations are challenging today. Um the first one is focus. They don't have focus. Every time that I land in an organization I see that they are doing 120 PCs uh but impacting zero dollars in the P&L. You don't need to do 120 things. You need to do 10 that impact the P&L. The second thing is you know people think that this is a tech project. This is a complete business project. tech and as we like to say 10% of the effort in a project and the impact relies on the intelligence of the model 20% on the IT infrastructure platform etc. But 70% of the effort and impact will come from changing processes changing people and making this capillarize along the organization. So this is the first challenge focus. The second challenge that I see sorry the third challenge that I see is not focusing value and focus in technology. You shouldn't just think about which is the best LLM which is that's secondary. What you should pursue is value because today technology is very mature. We hear that voice for the ones that understood Spanish. That woman took a breath in the code. She basically laughed a little bit. So we are in a level in terms of technology that it's already sufficient to start bringing heavy impact. Let's take that as a priority instead of just improving one extra point in accuracy. You know that that's my view. I fully agree. Oscar like what are what are the biggest challenges that you're seeing when deploying because like you're in this like really interesting space in the middle between companies and technology overall right so what are the some of the challenges that you're seeing right now so [clears throat] uh taking on on Juan's point I think the the element around focusing on the value and how you are transforming the process of the of or the experience that you are creating is the first big hurdle is technology it's been proven works you have significant ways to uh to implement and putting in play in in in place solutions that really feel natural that really are driving the conversation. Ensuring that you are focusing on the right part of the processes on solving for the friction points that a customer might have and then addresses all addressing all the different silos that an or an an organization has to actually get you there is where usually a lot of the time is is taken. And then also creating that feedback loop or that intelligence to listen, understand, share an intimacy with our clients to explain, hey, we're observing this. This is impacting your business. Let's focus on how do we change it this way and iterate much faster than what we were used to iterating uh back in the day with a traditional service where you have to train the agents, you have to spend time getting them ready to follow a new script and then you can accelerate that feedback loop and training capability. And that's uh that's that's an area that requires some change in the behaviors of our teams and the client teams, but it's it's really the the point that can make it or or break it. And do you see some interesties that are like much more like forward looking towards like this like technology stacked with like old building? We've seen uh quite a lot of of uh of interest in as I was I was mentioning before on on pro on industries that are heavy on uh on some of the processing back office uh or industries where there is a lot of interaction to provide basic information to to customers or where they need to take several steps to complete the the activities. Uh Terra was mentioning uh on the uh utilities industry all the processes that you can do in your own personal page. That's a perfect example of right now being 100% reactive. It has a lot of opportunity to transform into being a more accessible and being more proactive into on how you engage the that area. That makes sense. And and I'm actually quite curious like how did you pick your stack your tech stack because like there's plenty of providers out there. There's like 11 labs. We power pretty much 90% more than 90% of anything that is AI boys in the market. But there's plenty of solutions out there, our agents and a bunch of other ones. How did you pick your solution in your stack and and and everything? So our main guiding principle is being open and interconnectable because at the end the reality of an organization like ours is we are working with the technology that a lot of clients already have. So the theoretical I can go and implement whatever I want into the client environment is nice to have in a in a lab. It doesn't really work in in real life. And then the focus is looking at best in breed solutions for each one of the the elements and that modularity of how do we put everything together uh to deliver the value that we are looking that we are looking for starting from the process and the value we're trying to create not so much the technology technology has to be an enabler is critical but the focus is what solves best the problem that we are trying to to get into. Okay, that makes sense. You brought us a demo as well. I did as well. Yes. What what you'll see is a brief demo of a financial customer in of financial client in Latin America. Uh the focus in there was to transform one of the processes that is taxing for human agents which is cash collection which is basically calling people and say you've especially the first 30 days post delinquency you have not paid for your credit card your this uh we should negotiate a a payment plan. And what we built or what we worked on building there is a as an agent together with BCG that would help us turn that not only in into a notification but making sure that there is a negotiation that we can get to to the to execute and it actually is performing amazingly well. We have managed more than 350,000 calls and the reality is that it is performing at the level of the best uh agents that we have working on those on those services focusing on additional business that the customer either the client in this case couldn't address before implementing implementing this. That's excellent. Well, we'll talk about like ROI and performance and everything in a second, but uh let's see the demo. Awesome. Perfecto, William. nice. Let me let me say a fun fact about this because I think it's an interesting project, right? which is when we were making it. I mean, you know, collections, collections is a hot topic, you know, not easy calls and you would assume, hey, people are going to hear a robotic voice, which is not the case at all. And by the way, 11 Labs was not only the best Colombian voice, but also one of the only ones that exist in the market, okay, without a a cloning. Okay, but the fun fact is that the NPS of the experience of the deptor improved. So you would expect a much lower uh what we call RPC which is real connection because a body is calling. That's not the case and Oscar can give more details here but when you look at the funnel and as Oscar mentioned the body is performing equally to a human. We we see it a lot in Brazil for instance like where like um it ends the conversation like with uh people end the conversation talking to agents saying via right which is like equivalent of like have a good day but like in a very similar like in in a very specific way for Brazil. Um, and for me it's I'm always like amazed because like fundamentally I'm like really people forgot that they were talking to an AI agent and then like thinking they talk to a human because that's the natural thing and that's why I believe like once you get that you've made it right you've made it in the sense like people have exactly the same experience or better that they were having before. Um, how are you thinking about the ROI success metrics all of that stuff? Yeah, for for us as I said the main success metric is to to really have the best the best customer experience and and and the happiest customers in a utility which is which is a big ask because nobody loves its utility company. Um so that that's our driving force behind uh the project that we're that we're doing and we are really excited about it because if you think about we we have over 800 agents in four different locations across two continents serving our customers which is do a terrific job but it's it's it's really hard to have a a homogeneous experience for our customers depending on who is is who are they talking to. So we feel that if if we if we can handle most of our conversations through an AI agent that are our more day-to-day things that we treat our customers and we can give them a very homogeneous experience which is which is I think really impressive what you can achieve with AI and leave our our agents to handle the most important and and most complex parts of the of the operations with the customers. then you you can we will achieve uh a much better customer experience from what you get today and and as Matty was saying at the opening uh reduce weight times uh have 24/7 support uh multilingual uh support for our customers uh and and so on and so forth. So that's that that that's how we are looking at ROI. We're we're not in in a very financial type of situ type of of um approach to it right now. Juan, are you seeing like similar metrics, KPIs, success, all of that stuff across different organizations, different industries or does it vary a lot based on the industry? It varies a lot based on the domain, use case and industry and also the strategy that the company has. It's very different to have a cost optimization strategy than to have a you know growth strategy and depending on the market you know you will decide where you want to invest your your your buckets you know uh today most of the projects are focusing in cost optimization having said that the business case becomes much more positive when you focus in topline impact of course a lot of people think that agentic is about you know let's cut here cut there in reality The way you should think about this is imagine scaling your organization without any limitation. Imagine contacting any, you know, every person living in London today with no limitation. So that's not cost efficiency. Yes, it's productivity, but it's about topline impact. It's about scale. Agentic will bring into companies the scale that they didn't have and the speed that they didn't have because we were focusing in human capacity you know that makes sense um like we were talking backstage about agentic organizations fundamentally right and you were telling us Juan like what you're seeing right now in the market like give us like 30 seconds yes so this is related to the first uh topic that I was saying which is some companies are really creating a green field where autonomous organizations are being created and we are not very far away from seeing you know a first autonomous telco or a first autonomous sales distribution channel for a utility or even an entire autonomous BPO right the limitation is not in the tech the limitation today is mostly in regulation um for example what Terrell was saying the experience is delimited by regulation but from a techn technology perspective you are fully capable today to create these type of organizations and we will see it very soon Oscar what's the what's the future of BPOS right like probably what everyone is thinking here like how are you seeing all of this and how are you like getting ahead of it and what are the trend and all of that stuff so it definitely changes the positioning that we have as a as an organization we traditionally were working in an industry that was based on providing seats to address processes that companies wanted to outsource that moved more into outcome based activities. So focused on hey how much of that can you deliver and I'll pay you for this and it is focusing up the chain on how do I get to the outcomes of the business to uh to Juan's point more focused on what is the top line and what is the benefit for the customer as we as we evolve we see it as an increasingly tech uh tech focused activity. So we need to ensure that we not only operate the activities that are happening but we also are leveraging the best in breed that we that we can find and also maintaining that hybrid mode of of working we still have and for the for the next uh next years we foresee that there are parts of the processes parts of the activities that we will want humans to be part of. So we will want uh humans to be managing, addressing, engaging with customers in certain situations where you want to where you want to have that closeness and that and that empathy. Whilst there are many cases where uh you can automate and remove that part of noise from the from the activity by making the whole process uh hybrid and that orchestration that positioning on how do I manage the right approach? How do I make sure that my the customers uh the customers of my client are getting the right support in the right place at the right time through the right channel? Optimizing the channel is a critical part of where our our industry will be evolving because we're going multi- channelannel across the board, right? Like uh I mean I like I hate when like a companies like force you to actually interact with them in one single channel in that specific channel, right? And I think like that's one of the big benefits of agentic systems that like essentially you end up like plugging in multiple channels and be where the customer is versus like be where uh only where you want to be. Uh very briefly ter like uh what's next for your deployments like anything that you can share like you don't have to reveal any major. Yeah. So, so our our focus is to to deploy this on second quarter this year and and we're as I said we're really excited on on on on putting this in front of our customer and seeing their reaction. So yeah and and uh and also I think that that we are starting to to look ahead on what's what's coming next to to us because we were really focused on this project for for the last 12 months and and and and we think that there there is a a lot of other places in the organization where aentic AI will will will will help us deliver better experiences for a customer. Will you focus more on the next ones on internal optimization of the organization? Will you focus much more on customer uh I think I think I think both Carlos and and and I mean the the the possibilities are almost endless right now. Uh but we we want to have we we want to have focus and impact on on what we're doing. So those are the two main criterias that we're following on on how to approach uh our next projects. That's excellent. Well, I mean for me it's it's been great having you here guys. Um seeing like the impact that like your organizations and the deployments are having across the world. Um what what we wanted to do fundamentally is like just share a different perspective of what is actually happening right because like sometimes what we see is like we only are in our own vertical our own industry and I think it's important to actually see what's what are the people actually doing. So I really really thank you guys for for coming here on stage and being with all of us. It was an absolute pleasure. Thank you everyone.