The future of customer service: More human, thanks to AI

Speaker A: One of the most obvious places a company might look to start transforming itself in the age of AI is its contact center. But what can AI do in that context? And how is that going to end up being good for customers? The answer, it would appear, is to be unremittingly human centered. And at least that's the message delivered to us today by our uh, two guests on the Node podcast. Uh, Ahmad Nizami and Ajmal Mahmoud come From OmniEngage, a UK based company company who are operating very internationally to really unlock companies large and small, um, the powers of AI, not just in helping to understand customers faster and better and to solve their problems quicker, but to help improve the contact center, uh, humans themselves. It's an interesting chat. Stay tuned. Archmal Ahmad, welcome to the Node podcast. We are here to talk about the intersection of our lives, our businesses and AI and the transformation that's occurring. But today, most specifically we're doing that in the context of uh, customer service, customer interactions, businesses and the way that they interact with their customers. So that's every business your company is Omniengage. But rather than allow me to clumsily describe what Omni Engage is, uh, why don't you let me know, um, whoever speaks first gets to go first.

Speaker B: I'll speak first if that's okay. And I can, I can be slightly more technical and I can do a bit of a sales, a bit style. So I am. So yeah. So Omni Engage is primarily uh, um, a software house basically. So we develop software for, as you said, software for the customer experience industry. Um, and that, that's our focus, that's what we do, that is our background. Everyone in the business is a, some sort of background within the CX industry. And that's how we started, um, basically to develop um, agents and supervisor and the staff experience within the contact center, as you said, um, how they basically engage with the customers. And that's where we come in, how we can actually make that experience frictionless.

Speaker C: Yep.

Speaker A: Do you have anything to add to that, Ajmal?

Speaker C: No, that's pretty much bang on. It's always been about the user experience, um, but not only that, it's about getting a company or an organization to a point where they can easily influence that user experience, um, and be able to be agile, um, and be able to evolve really quickly. And Ahmed and I are both from, uh, am I allowed to say legacy, A legacy kind of environment. And legacy is kind of a dirty word. Um, but back in the day where we used to be frustrated by, we're Frustrated engineers and we're frustrated. Contact center agents is basically what you can call us. And it was the way we used to deliver used to leave us with a bad taste because you were giving somebody something, a version of something and you were walking away because that's all that was available. So when somebody wanted to make a change, you kind of felt like they were over a barrel and there was nothing you could do to make them deliver better services to their customers and have a better user experience. And that's been the driver for nearly all of us, um, behind omnigauge.

Speaker A: Right. And so we'll get on to sort of AI specifically in a minute. But I suppose as director of cx, people's experience is kind of key to your role. Largemile I know, but um, where are we at now do you think? Because everybody remembers or everybody knows and still experiences what it's like as a consumer to interact with businesses, businesses that are more critical to your life than others. Depending on who it is you're talking to. Um, it really matters if it's a utility company, it may be, it really matters if it's your cell phone provider or whatever. But um, some things that matter less but so much rides on the quality of your experience and when we start talking about AI, how that gets involved in that process, maybe right at the front end, as in you're literally having a conversation with it, um, but perhaps more so in the back end where it's trying to help solve that old problem where um, you're talking to someone, they say look, can you just, can you just hold for a second? I've just got to switch screens. That's the normal word. As if you know what that means as a customer, but we know what that means. I've got separate systems here, they don't talk to each other. I've got to go and manage that. Um, how is what you're doing now changing that and how has that happened over the last year or two to make our experience as consumers somewhat different in terms of how they interact with your customers.

Speaker C: So I think I'm m going to let AMA talk about the technicalities and what's become different based on all of the newer technology, newer AI driven processes and things that we can do to better serve our customers, to allow them to better serve their customers. But from an experience point of view, working with aws, they bring out these new services. It's amazing. And we live in an echo chamber as a technology people in this um, ecosystem. We live in an echo chamber where Everybody's talking about the most amazing things that are coming along. And we can do things that, uh, felt like science fiction only five years ago. But, uh, there's an element of realism that you have to keep. And a lot of people forget that. And uh, sometimes I get sort of, People do say, why have you said it so bluntly? But you know, when you go, it's amazing someone's telling you about it. Then you go, well, hang on, I've just moved house. And every organization I've called, my experience has been, come on my lab say, can you edit. Do you need to edit that out? No, I'll beep it with something. Okay.

Speaker A: But um, twice in fact, because you said it.

Speaker C: I'll try not to say it again. Um, but there's a realism to it. So when you go and deal with an organization and you think, well, um, so why is it so terrible? And it's because the concentration has been about making an organization slicker, more efficient. So when somebody says, well, what am I going to do? I'm going to use the more agile software processes. And now we're talking about AI and generative AI and agentic and um, AI agents and agent to agent and all of these things. Um, but originally everything was about efficiency. And what does that mean? It means I want less people doing the same kind of thing. And if you were to talk to a big organization three, four years ago and you said, um, and what is it that you're trying to achieve? And they would say, well, we're trying to transform, we're trying to be more agile, we want a better experience for our customers, we want a better experience for our people and we want to be more efficient. I know that if you told them off the record, said, do you know what, I could make everything more efficient. But the, uh, and I'd save you 50% on your capital expenditure or operating expenditure, but the experience is going to be a little bit worse. I suspect most of them would have taken it. So it's what's the outcome that you are trying to drive? And what we stress up front to all of our customers is what's the outcome that you are trying to drive? And most of the time that's the bit where it gets you to the crux of it. Um, and a lot of it might be, it might be a metric of some kind. I want our mps to be higher, I want to do something. But the environment that we're in now, technically things shouldn't be a barrier, but there's also an element of you have to think a bit more holistically about the technology you've got, the industry you're in and where your industry is moving and all of those new things that have come in and I'm going to handover Ahmed, you can talk about those. They're the ones that you look at and go, hang on, I can achieve all of those goals. Um, but you have to be really clear about what you're trying to achieve, where you are now, have you measured it? And where you're trying to get to, and what's my measure of success going to be. And that's the thing that can be a struggle for everybody.

Speaker A: Yeah. When we were having a conversation previously and I remember you talking about the human centricity, I guess, of everything you're doing in terms of technology. And I remember walking away from that thinking, do you know that's true? There are so many. There are so many companies that I purchase things from that I purchase from because of the quality of the service that they provide. Because technology is. Is actually across all of these different companies, becoming a bit of a leveler in terms of what, what they can literally provide. Um, so what is, what's the difference maker? What is it that. And it's actually, I've realized the, the better my experience of dealing with them and the more human they are actually, um, the more likely I am to be loyal to them and stuff like that. But anyway, that's just my rather unqualified thoughts on the whole thing.

Speaker C: No, just.

Speaker A: Just on that.

Speaker C: Just, just on that. Sorry, before you go for there, there is an element of human versus autonomous service decomposition that sounds really. Yeah. Clever.

Speaker A: Sounded really clever.

Speaker C: But, uh, there's a division of tasks between what you want technology to do and what you want people to do. And most of the stuff that you want the people to do is where people want to feel like you've acknowledged the thing that they're trying to do. If someone's trying to tell you about a, uh, bereavement, if someone's vulnerable and you're trying to, you know, your software and AI can work out now, if somebody's vulnerable and what's the first thing you want to do? You want to put them through to a person. Those are the things that we have to identify where. And it becomes much easier then you're trying to fix something. It's a bit more binary there.

Speaker A: Yeah, yeah. And there's the way. So when you're at the front window, if you like, as a customer, and you say, uh, maybe it's your ISP or uh, somebody's providing you with a slightly more technical service and you maybe you're coming to them with a problem that you've already actually partially diagnosed yourself. You know that you need to speak to someone about this. And um, a big question mark then is, okay, how intelligent is this automated front window, front door going to be? It's asking me a lot of questions that I knew it was going to answer me, but actually I'm already irritated because I know I've just got to get through this to the point where I can say, actually just let me talk to a person because I know you can't deal with this. Um, um, that's a real frustration point for me still is that there's this kind of really nuanced intelligence layer there that needs to understand or ask me up front, do you want to speak to a person before we get into this? Because you know more about your problem than I do so far anyway. Um, talking about nuanced intelligence, Ahmad, um, um, you are the chief technology wizard, uh, in all of this. Could you give us a, actually give us a quick rundown of how it is Omni Engage does what it does. I know you're sitting across um, Amazon Connect, um, but if we had 30 seconds in an elevator to describe the kind of the technology footprint that you operate across, how would you describe it?

Speaker B: Yeah. So, I mean, um, probably going slightly, taking a step back. So as I said, we've always been concerned with how we can actually create that user experience, which can actually help because in your question you touched upon um, the one you posed to the um, post Joe, about um, how, you know, how can we reduce that switching and agent experience because at the end of the day that agent is actually dealing with the customer inquiry and the customer has a problem and that problem needs to be resolved. That's the fundamentals of it. Right. So our focus has, have been um, when we actually build on Engage, the first platform that we built on Ningedge Connect as a user experience layer on top of Amazon Connect, was always been how we can actually provide the right information to the agent at right time, at the same time integrating all the right systems in a way that uh, the information or those systems should automatically come up to the agents, to agents tension so they can actually use the right system to actually follow the process and potentially help the customer. I think what AI is now enabling us to do is that agents still had to remember a lot of stuff. They still had to operate from memory, they still had to refer knowledge base to the training when working on these systems. What AI has enabled us to do is to actually bring all that knowledge into one place and feed that into the AI. So basically AI can enable that agent and aid it to actually solve that qquery much, much much faster. And I'm not talking about autonomous park yet, that's, that's obviously the evolution we are going towards and we'll talk about that uh, obviously in our talk. But I think what we have tried is to actually how we can actually bring AI to aid the human who's actually trying to help a customer. Right. So that, that has been focused for us pretty much in last six months to eight months. But I think technology is evolving at a much faster pace and now we are looking into how we can now actually let technology autonomously also help and um, customers directly as well.

Speaker A: Yeah, so I guess uh, maybe this is a question you can't really answer but you've got lots of customers and um, some of them are big and some of them are small. Some of them are critical, mission critical type things for customers. Some of them I guess are less so. In terms of generally speaking your experience of how, how mature uh those people are in terms of AI adoption where I know you can't talk specifics but where are we at? What kind of success levels are people beginning to achieve?

Speaker B: Yeah, I mean I could talk from the technology and barriers that I face and then um, obviously as we can talk about business processes and I think it's a combination of the two. Obviously business needs to go through a certain transformation to adopt the technology, change the processes, embed the new technology in the process. So there has to be a degree of flexibility and um, change there. But um, from the pure technology uh perspective deployment of the technology, I think um, AI is as good as the knowledge that it feeds on and the systems that it connects with. Now if you have a whole bunch of systems that do not talk to anything outside, you still have a problem. You still have to build some sort of orchestration layering between where AI can actually tap into those systems. You still have to somehow bring all the knowledge and system, uh, all the knowledge from various places together and feed that into AI in a manner where AI can actually provide a good solid response back to the, back to the agent or back to the Dominion. So what we see in some of the barriers is still um, not having that knowledge available for example in one place or having, or business having to basically go and do a, carry out an exercise to actually bring that Knowledge together and then we can feed that to AI. So from the technical barrier perspective, these are the things that we still have to deal with, uh, businesses but we do guide them obviously we help them through our consultancy and PS exercise that. These are the steps that you're going to have to basically uh, go through in order for AI to actually do things effectively for you. Atmol. From the business perspective, do you have anything to add to that?

Speaker C: Uh, well yeah, because the readiness question of organizations, of course there's um, a way to adopt it and the way Alma talks about it is really crucial is the fact that you've got the data, it's clean, uh, it's available, it's organized, you've got to be able to call on it. It's not replicated anywhere. It's not, it's all in sync. Um, and most organizations don't have that. Um, and you know, the age old garbage in, garbage out still applies. Right? Um, and, but there's also the bigger, the organization that it's slightly more difficult but it's, there's, it's more of a minefield because you're going to suddenly have an AI governance process. You can't just throw in um, an AI solution to do something without a lot of questions being asked and answered, which is, you know, ah, the large language model, um, are the guardrails in place? What happens if um, there's some uh, hallucination that's come back, um, those kind of things, how do we detect those things, how do we report on it? But also there's a lot more work being done around what are we trying to fix? And Ahmed um, will tell you, he's always said to me, from the, ever since we worked together, you know, what problem are we trying? Give me a problem statement. What are we trying to fix? And at the midpoint, at the three quarter point, at the end point, you keep asking yourself are we still fixing this problem? And that's the bit where you go, has it been a success? Whereas there has been a trend of must, we must have AI because it's the thing that we, but that's not the outcome. The outcome isn't we must have AI. The outcome is we must be better and are we going to use AI to do it? Um, and that's the difficulty with organizations at the moment. Small organizations think, yeah, turn everything on, let's see what happens. The big organizations know we have to go through these steps and be able to say we're consuming AI services and these are the outcomes that it's driving.

Speaker A: Right, because that's another thing of course is that AI costs quite a lot potentially. And so um, you've got this dual need to make sure that you must invest in it but without necessarily having a clear idea of what the return on that is going to be. And it's very easy to secretly around the corner somewhere say well the return is less people on the payroll. Um, but that's not the message that I received. Um, actually that's not the message I generally received. But it certainly wasn't the message from the conversation we had before which was actually these are tools that we can use um, to make people just a lot better and quicker at their jobs so that companies overall are more productive and that's where the return is. So um, just to go back slightly into something you both mentioned I think, which was there are big companies who have got certain amounts of considerations and then there are small companies who might behave slightly differently. Um, you have got kind of different products I guess or different services for different sizes of business, haven't you? I think, because that's particularly interesting I think is you can easily imagine bigger companies with more resources being able to go uh, to exploit the advantages of new AI tools and experiment them and get away, and get away with it more than the majority of companies which is actually much smaller companies. How are you addressing smaller companies? How are you providing services to them uh, in a way that um, well first of all they can afford but then secondly that definitely results in something for them. How are you getting them into the AI game?

Speaker C: So there's so just to touch on the larger um, organizations, uh, the things that we try to instill in them is that they need sort of agility as a core capability. So if you've got modular architectures and you can make changes really quickly and with the AI based services that are available now they're so consumable so you can grab something and go I want m to try that really quickly and but if you're set up as a smaller organization you're going to have a cleaner environment. So if you're a telco for example, they're going to be made of lots of different companies and those companies haven't been fully assimilated. So they're not going to have a single CRM and single um, knowledge, um base. They're not going to have um, a single product provisioning um tool. So all of those things are complex. And not only that, there's um, this sweating old assets. So there'll be stuff that's you know, um, um, a uh, center based, data center based solution, it uh, might be installed on desktops. You can't always easily leverage new technologies. Whereas a smaller organization, they are much quicker at being able to adopt these things. But they, but it's harder for them to have a clearer, uh, um, measure that they want to enable. And so you have to have a more of a turn it on. What's going to happen? Are we getting the benefits that we thought we would get? Yes, we are. No, we're not. Balance it against the cost and because it's so consumable and you're not locked in, you can just turn it on, turn it off and you go, that didn't really do what I wanted. Uh, but the bigger organizations, yes, the benefits are huge, huge. But there's all of this. If they're big, you know, their technology, their financial insurance, they're going to be regulated as well. That's a uh, different element which we'll probably touch on a bit later. So there's lots of different ways to look at it, but with the way that it's so consumable that AI services are, or AI based, it's such a loose term now, uh, it can mean so many different things. But because they're so consumable and the way that we structure it, we want them to be able to turn it on and turn it off, but we also want them to know exactly what they're turning on and what they're turning off and what they're going to get for it. Would that be fair, Ahmad?

Speaker B: Yeah, so I mean, and I'll take a slightly different angle on that, Ravi, if that's okay. So smaller businesses and technologies like agentic AI can really help them with the customer service. Again, we're going to focus on customer services today because that's definitely too. And so from a smaller business perspective, having lot of people just to answer emails and chats and WhatsApp messages, all these channels, um, obviously sometimes can be cost prohibitive and that sometimes for them means lost revenue, lost competition and not having enough people to actually respond timely to the customers. I'm picking a very, very basic use case. So that's where actually agentic AI can actually enable them to be more available to their customers on all channels and um, any channel of choice for the customer, whether it's, you know, um, you know, digital or voice. So I think agentic, in that case it can actually um, uh, allow them to offer a lot better service at a lot less cost. Um, from that Perspective because they, they cannot invest a lot into having a lot of people just handling those interactions, inquiries, bookings, whatever that that may be. Um, so that in that sense it can actually save them a lot of money as well and offer a better service back to the customers.

Speaker A: So let's, I guess let's talk about the future a little bit. Um, it feels to me as if we're still at the very beginning of this journey if you like. And um, but uh, I'm not sure entirely how true that is anymore. It feels almost like actually there are some people that are quite a long way down this journey now. But where are we going to be in two, three years time with this? What do you think the impact is going to be on people who work in this kind of context in particular?

Speaker B: Yeah, um, I'll handle that first if that's okay. Ashbal. So from the tech landscape perspective things are evolving at, really at speed. So probably predicting something in two years is going to be slightly uh, difficult. Um, but what we are seeing in terms of the again customer services industry is that we definitely going to start seeing um, kind of a mix of AI and human um, on various channels and depending on the use cases humans um, will probably be handling more complex inquiries, maybe voice um, channels while digital channels can actually be handled by AI. So um, I've literally um, just published an article on my LinkedIn few days ago and that was about the future of contact centers is going to be an execution center where you actually have uh, humans actually handling the interaction on one side and AI actually working in parallel uh, and in collaboration with humans and then having a kind of a management layer on the top which is kind of feeding into both. So you can actually see how well are humans performing versus how well are your AI agents performing and you can actually provide feedback to both um, based on what you're seeing. Um, so what we, what I really think is that that's going to be you know, a mix of AI and to be honest I say it will be but we already working on it and I'm happy to kind of take you through that as well where we are already introducing um, the, the AI agents and that's what our next focus for next six to eight months is really to actually introduce those AI agents along with the human agents in the same platform.

Speaker C: So if you go, if you go a little bit further into the future it will be so um, the AI agents living under, uh, maybe even for an organization living under an agentic process that leverage during workflow, leverage multiple Agents and those agents might go off and talk to other A.I. agents. Um, and they'll all have human like qualities that we've attributed to them like um, um, identity, access, management, all of those roles we've given them. And they can only access certain things like, you know, they can't give uh, the wrong type of refund or things like that. There'll be a limit on what they can do. Um, but also when you think about from a consumer or a customer's, ah, customer point of view, if we're looking at our customers, then it might be that you'll just have a personal assistant, a digital personal assistant and that then has a bunch of agents working underneath it. So for example, you might just say to the assistant, um, um, I want to, I don't know, increase my home insurance cover to whatever amount of money because you've got all of those agents that sit under that personal assistant. You might just go and say, right, there's an agent for that. That agent will talk to the insurance company agent and make that transaction happen. And if there's an element where you've breached uh, a threshold or something like that, it might say, actually you need to speak to somebody and that agent will do the preamble for you, identify you, take, um, the intent and pass it across. And when you speak to that person, they already know who you are, they already know what you're trying to do. So that little interaction with a human will be really slick and not just a repetition of everything you've done before. So those are the, that's the sort of outcome or the utopia that we talk about. But going back to what I said about, you know, when your phone the insurance company or you phone the bank or the waterboard or whatever, and it's still a bit rubbish, but when you look at it from a Persona based, um, outcome. So what is it that I'm trying to drive, or what's the Persona that I'm attaching to that and deconstruct that into a service. It becomes a bit easier to work backwards from there and go, I'm going to create an agent and this is what they're going to deliver. Um, and you don't want to throw tons of AI services and compute power and all of that. And it might just be a much simpler process of a little bit of automation rather than saying, well actually that's AI or that's an AI agent or that's an agentic. Sometimes these things are banded around, but sometimes you can distill it down to Something a lot simpler.

Speaker A: Yeah, because, uh, I mean, that was going to be my next question really, which was, you know, in terms of what your focus areas are in the next few months, um, which I think you've answered, really. And agentic is, you know, obviously must be a really interesting area of R and D more than anything. But, um, I can't help myself going down this kind of mental journey that ends up seeing me sat in one of those hovering chairs on the spaceship in Wally, though, you know what I mean, Where I'm just, you know, like, I've got agents. The people I'm buying things from have got agents. Whether it's groceries or cell phone service or Internet or um, deliveries, whatever. I don't actually have to communicate or talk to or do anything to anybody anymore. I can just send my spiders off into the world to go and just have these things brought to me. I don't even need to get out of my chair. This is not realistic, but that's how my head works.

Speaker B: But I suppose two years, that's not happening in two years.

Speaker A: Um, definitely three years, two and a half. No, but that's a sort of almost dystopian vision of how this might work though, because sometimes it feels as if agentic AI, you can see how these things can patch together and form this jigsaw of convenience that ultimately ends up feeling rather soulless because humans don't seem to actually be involved apart from just consuming stuff. But that's definitely not the way it'll go because humans just don't work like that. Um, especially the ones who work in the businesses that are providing the services. I don't think you, you know, you can't. So it's going to be interesting anyway. But, um, I suppose just uh. We're about up with our half an hour here. But, um, what can you tell us about what Omni Engage is looking to do? Obviously we're all in the UK right now, but where are you, uh, operating and where are you looking to grow in and do new and interesting things in the coming year or two?

Speaker C: Do you want to do that one

Speaker A: or I guess commercials?

Speaker B: Yeah, absolutely, I can talk about that. And yeah, so I mean we definitely, um, venturing into new territories and um, new areas, new countries. Uh, one of the key venture that we kind of key country that we're trying to venturing in is us. Um, we already had uh, very good discussions going on there. We're working with AWS team over there as well. Sales team has actually uh, had few tricks as well. Um, and um, we are kind of looking at a few verticals, especially the healthcare side as well. And um, with the recent announcements from aws where they have actually, um, provided, uh, an integration in terms of the agenti layer, uh, with the US healthcare system, I think that is something that we really want to look into and that is really interesting and probably so that's something we want to explore more, uh, with the clients in us.

Speaker A: I was about to try and wrap up, but now you've got me interested in asking another question, which is obviously something you mentioned was, um, some of the industries you're operating in are highly regulated. And then there's all sorts of questions that pop up then about how you're treating, how you're collecting data, how you're using it, who you're transferring it to. Um, I may have all my spiders from my floating sofa, but that doesn't mean that they're not going to bounce off data privacy walls when I need to, um, go and get something or whatever. Um, how are you dealing with that? Because, you know, I do, having worked in that industry, I know how hard that is, um, from a, to manage, navigate regulatory requirements and so on. Um, it's, it's obviously a massive opportunity, um, for things to improve in a really general sense. Um. Ah, but how, how, how do you see that playing out? And is it different in the us say versus the UK Europe?

Speaker C: Uh, obviously there's different regulations. We're not that different from obviously the eu. We might be outside the eu, but the regulations, we still operate really closely and the US is a little bit different. But, um, it's still not a free for all when it comes to AI and um, the regulation, because like we said earlier, all of those virtual agents, um, AI agents, whatever it might be, they still have that identity and access management and they're treated like a human would be because they still, they still have limited access into all of the systems and what they can do with your data. So it still lives in your account, for example, as a catch all. Um, but there's also the way that we look at a lot of that data and make it actionable. So you know, when you're talking about the virtual coaching, um, but also the regulation that you mentioned, uh, if you talk about telco, for example, you know, you have to um, register every complaint and the, you know, Ofcom needs to know how many complaints you've had. But if in the new world you're looking at, instead of monitoring calls, you're doing three to four calls per agent per month. And sudden suddenly, with Amazon Connect, for example, and I'm sure it's similar with other platforms, you are now monitoring, recording, uh, transcribing and evaluating every single call. It would be fair to assume that you're going to capture more people who are making complaints in that process. But if you then go to the regulator and say, the regulator says, well, why do you have so many more complaints? You go, no, I'm not. I don't have more complaints. I'm just better at evaluating what I do. And also with like, you know, our latest tools that we have is a virtual coach that takes all of the data that's available. Again, it's actionable. Like, I'll keep going to that point. Data is great, but if it's not actionable, it's pointless. You might as well just throw it away. Um, but what we do is we make it actionable by incrementally making each agent better, giving them a little bit of feedback every day. But you don't batter them with it. You still take into account there's a cognitive load, uh, element that you have to consider. So you tell them what they're doing really well. You tell them, here's one thing you could do to make your overall performance better. Um, but ultimately, going back to what we were talking about, what we at Xomnie Engage, do we still make it that you can accelerate the adoption of all of these new technologies and you're in a space as quick as possible where you turn things on, Is it working? Turn it off, it isn't working. Or, um, productionize it to say, I want to put it across my organization because the pilot was so good. Um, and I think for me and Ahmed, um, might have a slightly different view. But for me, that was always the thing. The applications that we developed, the processes that we did, and the way that we leverage all of these new emerging technologies, it was always about, doesn't matter what size of organization you are. You can adopt them really quickly and move into an ecosystem that allows you to have agility as a core capability.

Speaker A: Yeah, you're the folks that people come to if they understand that there's something exciting and that they probably need to be doing it in order to continue to compete effectively. Um, but they've got to move quickly and hopefully competently. Yeah, they can come to you.

Speaker C: There's a little bit of, um, sometimes you've got to tell people to kind of stop as well and go, hang on, why are you doing it? And it goes back to me saying, if the why you're doing it is because I want AI, you need to go back and find a different question that we can help you answer. Right.

Speaker A: I think we've all got personal experiences of that. I think still AI is still at the stage now where it's really cool and it can do these incredibly powerful things. So we try and make it do those things sometimes without really thinking about why we're getting it to do those things. In fact just this week I had an example of that where um, we had a range of documents needed to be ah, rebranded as part of a process, let's just leave it at that. They needed to be submitted, uh, and they all needed quite a lot of time consuming work done to them. Claude can definitely do this. That was the thought. Right. So then we spent all week prompting, reprompting, Claude can definitely do this. We sailed straight through the time that actually would have just taken us to do it. Still trying, it never managed. So uh, there's a point at which, and that's not because in this case Claude was bad, it's just that there's a time and a place actually you don't need it sometimes what you need is humans. Um, so it's interesting that you can have those kind of conversations with customers as well, which is to say, or potential customers, which is to say, okay, what do you expect AI to do? But how is that actually going to help? Um, from an expert point of view, here's the way that you can start doing things that makes a difference.

Speaker B: I think it goes back to again as you gave that document example and I mentioned it, um, already as well in terms of the context, um, you can apply AI to his particular problem, but you need to tell AI exactly what is it that he needs to look at from the inquiry perspective, from the context perspective and then basically provide the best answer. And sometimes where we actually see the problems are basically the context is not of good quality. And that's where obviously uh, and what's probably dangerous about LLM is basically they still provide an answer. They're not going to say no, they provide some sort of answer, how valid that answer is to be seen depending on how good the context is. So LLMs are always going to provide you some answer, uh, and how good context is being provided to LLM for that answer to be really relevant.

Speaker A: Right, yeah. How you have to be confident that the answer you're being given ah, are ah, correct. So for example, when you give a batch of document to an AI to just rebrand and what it sends you back is I've also changed all of the words for you. Isn't that good? Well actually this is thousands of pages and no it's not.

Speaker B: What branding information did you provide of that uh, to Claude in that case?

Speaker A: Do you not remember me telling you to not change any word at all? It's just the branding. Did you not remember that?

Speaker B: No.

Speaker C: See it's a prompt engineering. Prompt engineering is going to be the skill and it takes me back to uh, I mean you've probably seen the TV show, uh, Silicon Valley about the developers and it's years ago, so early AI still in television. And he releases these AI into all of their systems and he types into a uh, prompt window, remove all the errors from my code. And it promptly deletes all the code and says right, all the errors are gone. And then he says order me cheap burgers. And it has these big vats of meat delivered because that's the cheapest way to buy burger meat. And it's. So you have to be very specific. And the prompt engineering is going to be, become more and more important because it's uh, at the end of the day it's a machine. It doesn't understand context like a human does. It doesn't understand hm, empty like a human does. So there's an element that you have to modify your behavior but it also tells us why humans are still required at the end of the day. For us, our most effective, at our most effective is where the human is in the loop. Um, and it's those two things working together. There's an autonomous AI driven self service element that can deal with complex queries. But at some point you're going to need a human because you need the human reaction and context and empathy that only a human can apply.

Speaker A: Absolutely. Because from in a CX context there is very literally always a human in the loop and it's the one that's giving you money. So you better treat it with respect uh, and meet their expectations and ideally meet them with humans when they need it.

Speaker B: Right.

Speaker A: But um, but empowered ones, I suppose that's, that's where we're at with, with all of this.

Speaker B: So just to add one point around prompt engineering and how we are basically handling it, um, in our products we, this, this is something we saw and I'm just sharing an experience here that um, when you, when you pass for example a story to a developer and say look, this is what AI should do, and then he goes and he just writes a prompt and that prompt is probably sometimes not really what business outcome should be. So that where we saw some gaps that developers were writing all sorts of different prompts and they were not optimized prompts that the product manager envisaged. So we changed that process and we said we have to record each single prompt that we use anywhere in the system, whether they are system prompts, where there are prompts to actually do something. We should, we should have a library of all our prompts documented and they should be reviewed time to time and testing teams should actually be test the outcomes based on those prompts and we keep optimizing them. That's how important they are. You have to keep your library of prompts optimized all the time because yeah, again whatever AI does is as good as the prompt that it gets, uh, or what you ask it to do.

Speaker C: Yeah, we haven't even got into the use of gentic AI AI agents, uh, in how we use it in the delivery of our services, uh, in the background. So that's a whole, that's probably another session.

Speaker A: Well you can't just leave that hanging there. Um, give us a little bit more.

Speaker B: So that's where we're using. Platforms like AWS lend themselves really well to kind of you know, having all the APIs available and basically um, having an agentic AI layer on top, consuming those APIs and actually go and do provisioning. So we are basically uh, we have actually now developed a platform which will be used by our deployment um, team to actually deliver these migrations really quickly. So you basically capture customer requirements, you turn them into something which is configurable and then you pass that agentic layer and say look, can you go and configure near Amazon connect uh, instance in this region and these are the agents that needs to configure, these are the queues and this is how they're all linked. And basically then we have multiple agentic AI agents basically. So there's a master agent, the multiple agents, one agent responsible for just creating routing profiles, one is responsible for creating agents and they're all doing their own individual tasks and then reporting back to the master agent. So that, that's how basically we built this whole agentic layer, uh, or an army of agents to actually go and provision systems. And we think it's basically it's going to pass a lot of savings back to the customers because we're not going to basically spending days and days just to deploy um, their instances. Right. So we are looking at basically doing it in most efficient, less human error prone way and Also pass some savings back to the customers to say look, you don't need 50 days, this can be done in 10 days or 5 days for example, fully tested.

Speaker A: Wow. Is this something you're already doing or is it something you're. Yeah, it is.

Speaker B: We are, yes we are. So we are actually enhancing the capabilities. So there's, as you can imagine there is a lot of guardrails and security consideration because what you don't want is to, for agent to actually go and do something that you do not want it to do, especially for the production systems. So currently we're just doing it for the initial deployments, not actually for the changes in production. But that is also the plan. So as part of our managed service we would be able to give customer an access to an agentic layer to say look, if you want to, for day to day management of your um, contact center, you want to go and create a user, just talk to your AI and basically it will go and create that. But before we get there we need to make sure we are confident in the um, in the security and guardrail side, uh, of the platform.

Speaker A: Yeah. Especially if what you're, if you're doing this in the context of something like the healthcare. Healthcare industry.

Speaker B: Absolutely.

Speaker A: Because it's just in the healthcare industry what you're, what you're describing there is an ability to be really, really rapid um, in terms of deployment which is exactly what healthcare customers always need.

Speaker C: Right.

Speaker A: Because they're always doing mission critical stuff and the last thing they want is a big IT transformation that takes them offline for a month or whatever. So the fact that you can deliver really quickly is amazing. But they're also going to have all of these questions about guardrails and how is this working, how do you know what's going on, who is in the loop and is it us and so on. Yeah. Wow.

Speaker C: That's just on healthcare there's a little um. Because people worry about the automation and self service and things like that that exist in, that might exist in future in healthcare. But I think the focus certainly at the moment is about leaving the healthcare professionals to deliver healthcare to people. So they're not doing the admin tasks, they're not taking phone calls, they're not rearranging appointments, they're not doing mundane, uh, repetitive tasks that a machine or an AI program or an AI agent could do or a process could do. And they are delivering um, services, healthcare services to people who need it. And I think that when you think about, when you go into a hospital, certainly with the nhs, you go to the um, see a doctor, you go to the health center, wherever it might be. The amount of times that a uh, thoroughly trained healthcare professional is spending on tasks that aren't delivering healthcare to people, that's the bit that needs attention. And I think if we could make those people just way more effective in delivering that healthcare, I think we'd be winning.

Speaker A: Absolutely. And it's of course, especially in the US actually, but when, if you are a normal person and you're engaging with the healthcare system, the sheer amount of the number of people that you deal with and the number of processes and touch points that you deal with as a patient, that has got nothing to do with anything clinical because it's admin is 90% of your time is spent dealing with things that aren't actually your health or your clinician. It's everything else. So the opportunity there is absolutely enormous to streamline that. Especially because every healthcare system in the world is dealing with exactly the same problem, which is, um, rapidly aging populations, all of whom are walking around with massive amounts of chronic diseases and so on, overflowing the system at all. There are not enough uh, medics and clinicians and radiographers and all the rest of it coming in at the front end. So this kind of technology isn't just really helpful, it's arguably completely critical. It has to happen and it has to work in order for healthcare systems everywhere to meet the challenges that they confront. So it's kind of exciting I guess that you're you know, looking into that, that part of the uh, world. It must be quite exciting. But anyway, with, with that, it's all quite exciting. Talk about dynamic. It must be. I don't know how much sleep you get, but I bet it's, I wonder if it's worth it.

Speaker C: You have to kind of limit, uh, sometimes just going off, you know, coming up with stuff that you wouldn't even see on Star Trek because you're, you've got the scope to do it, but you have to make it deliverable and it has to be, you look at your existing customers, you look at potential new customers, they have to have an appetite for, has to make a real positive impact on their customers. But there's multiple elements of it. It has to be deliverable, it has to be regulated, um, and it has to give them ah, a benefit to the bottom line as well, all of those elements. Uh, and at the end of the day you still want, for us, we still want that human in the loop, but we just want them to Be more effective.

Speaker A: Yeah, absolutely. Anyway, talking of being more effective, um, where can people go to find out more about Omni Engage?

Speaker B: Um, I mean, first of all is our website. Uh, obviously, as always. Um, but then I think we would love to talk to anyone who's particularly looking at anything agentic AI to transform any business processes. I think we. We would rather prefer having a chat, don't we, Ashmole, than someone just, uh, making inquiries. Um, obviously all those channels are available LinkedIn website. Uh, but I think what we would really welcome an opportunity. If someone really wants to explore it further, be happy to talk to them.

Speaker C: Yeah. And you can go to the website. You can look me, Ahmed, Dan, our CEO, Matt, our CRO up on LinkedIn, message us directly. Uh, we're at events, we'll be at the AWS Summit. Uh, we'll be at multiple events throughout the year. Keep an eye on our company page on LinkedIn, where we post where we're going to be and how easy it is to contact us. But also the newer trends that are coming out that we're talking about, we're deploying, we're experimenting with, yet we want to hear from as many people as possible.

Speaker A: And if people do reach out to you, they can talk to you or your agent. Um, are you going to have. Are you going to have your, you know, personal agent?

Speaker B: Not yet. They will talk to us, at least for timely.

Speaker C: How do you know you're not talking to an agent now?

Speaker A: I don't. I don't know. This could. And what would I feel like. How would I feel if I found out that actually this was, uh, all completely artificial?

Speaker C: Yeah, we're not really here.

Speaker A: I would probably feel. Friday night, we're out. Am I real? No, this is getting. You can tell it's a Friday as we're recording this. Anyway, um, I think that, uh, that wraps up for tonight, unless you've got anything else you'd like to add. Um, what we will do is, uh, we'll go away and put this on, uh, our website, Node magazine, and include lots of links to the things that you were just talking about, especially events. I guess, if people can, uh, see where it is that you're going to be wandering around in the world and hope to meet you in the actual flesh, and then they'll definitely know you're not an agent. In theory. Um, uh, we'll see if there's any more content we can provide alongside that as well. But, uh, in the meantime, I do hope you have a wonderful, uh, rest of 2026. Because, you know, by the time we get to this time next year, I've got a feeling this conversation will be completely different again. The sounds of it.

Speaker C: Absolutely.

Speaker A: Excitingly, I think maybe M. Yeah.

Speaker B: In an exciting way, I think. Yes.

Speaker A: All right, thanks very much, guys.