What's Up with Tech?
Tech Transformation with Evan Kirstel: A podcast exploring the latest trends and innovations in the tech industry, and how businesses can leverage them for growth, diving into the world of B2B, discussing strategies, trends, and sharing insights from industry leaders!
With over three decades in telecom and IT, I've mastered the art of transforming social media into a dynamic platform for audience engagement, community building, and establishing thought leadership. My approach isn't about personal brand promotion but about delivering educational and informative content to cultivate a sustainable, long-term business presence. I am the leading content creator in areas like Enterprise AI, UCaaS, CPaaS, CCaaS, Cloud, Telecom, 5G and more!
What's Up with Tech?
Orchestrating AI: Martin Taylor on Transforming Customer Experience
Interested in being a guest? Email us at admin@evankirstel.com
Martin Taylor takes us on a fascinating journey through the rapidly evolving landscape where artificial intelligence meets customer experience. As the co-founder of Content Guru, he's spent decades watching this space transform from simple call centers to sophisticated, AI-enhanced engagement platforms that serve millions of interactions across financial services, government agencies, and healthcare systems worldwide.
The conversation reveals how COVID-19 created a watershed moment for customer experience technology. When face-to-face interactions ceased, organizations were forced to elevate their digital engagement strategies, creating competitive differentiation through superior experiences rather than just products or prices. Today's leading organizations are embracing what Taylor calls "omni-channel, omni-data, omni-automation" approaches that meet customers wherever they prefer to engage.
What makes Content Guru's approach unique is their philosophy of AI orchestration. Rather than building proprietary AI models, they scan the horizon for the best specialized engines—whether large language models, small language models, vision systems, or translation tools—and integrate them through an enterprise-grade platform that ensures security, reliability and accessibility. This lets their customers deploy cutting-edge technology without becoming AI scientists themselves.
Perhaps most compelling is Taylor's vision for the human-AI partnership. He describes a future where human agents sit "atop a pyramid" of automation layers, freed from mundane tasks like data entry and transcription to focus entirely on meaningful customer conversations. The technology doesn't replace humans—it elevates them, providing real-time support, suggestions, and even performance feedback that helps agents continuously improve. The results speak for themselves: financial services organizations slashing handle times by 50%, government agencies diverting hundreds of thousands of inquiries to AI-powered chatbots while improving both citizen satisfaction and employee experience.
Whether you're leading a customer experience transformation, exploring AI implementation, or simply curious about how technology is reshaping human interactions, this conversation offers invaluable insights from someone who's been pioneering this space for decades. Listen now to understand how your organization can harness these same principles to create exceptional experiences that drive loyalty, efficiency and competitive advantage.
An Acxiom podcast where we discuss marketing made better, bringing you real...
Listen on: Apple Podcasts Spotify
More at https://linktr.ee/EvanKirstel
Hey everybody, we're diving into the world of contact center and the intersection of AI and customer experience with a real industry legend and innovator, martin Taylor of Content Guru. Martin, how are you? Great? Thanks, evan. Well, good to see you again. It's been a while, so let's start from the top. For those who may not be familiar, what's the story behind Content Guru?
Speaker 2:from the top. For those who may not be familiar, what's the story behind content guru? Yeah, well, content guru is in the cx technology space, so we came from cloud contact center, ccas, and actually our history goes back 30 years to technologies. Yeah, so before contact center we were all about tv voting. So wherever there was this reality tv wave sweeping through big brother, idol, x factor and all those sorts of things then we would be following along in the networks, taking all the votes and and doing all the competitions around those shows yeah, I remember them well, good times, and so you've been around this space before it was mainstream.
Speaker 1:Before, certainly, ccas was a term of art. So what's your perspective Is now the most exciting point in the CX and CCaaS world now.
Speaker 2:It's a kind of evolutionary watershed really, where I think we move from, I suppose, things that have evolved out of call centers into this whole wider customer experience space. It's not just omni-channel but omni-data, omni-automation. So there's never been a more exciting time to be in this CX space.
Speaker 1:And what are your customers telling you about how they're approaching CX and cloud communications? What are they saying, what are their ambitions, goals and what's the state of the union, as it were, that the customers you're talking to?
Speaker 2:The modern era.
Speaker 2:I, I suppose, could be seen to have started in COVID so five years back now, and I think a lot of well, all face-to-face contact ceased, really didn't it?
Speaker 2:And so a lot more emphasis went into the customer experience that could be delivered digitally and through the contact center, and generally everything that would have been a shop window or a door or a counter has suddenly moved behind a technology layer, and so that put a lot more emphasis on improving that customer experience in order to derive competitive advantage and differentiate one provider from another.
Speaker 2:Because actually, if you look at a lot of the industries we're in take insurance, for example it is fairly much similar products on offer. You've got an insurance policy, you pay a premium, there's an excess, so, other than whether the insurer will pay out if something goes wrong, the rest of it's down to the cost, and then what kind of experience do you get? So, clearly, if you can use customer experience technologies in order to have a better experience and deliver it more efficiently, then that's typically where people have wanted to go, and I think in the last year or two there's a heavy emphasis on more intelligent automation now so that you can provide this very personalized experience in a much more efficient way than was ever possible before. And if you think about what was a contact center, it was a gathering of front office people in a location where data, information and skills can be paired and optimized. So it's the original optimized work environment just now being amped up and taken to another level.
Speaker 1:Yeah, reset. So AI is reshaping everything, not just employee productivity, chatbots, sentiment analysis, on and on and on. So, specifically, how do you guys integrate AI into your platform and what's your approach there? Everyone seems to have a different, unique philosophy, as it were.
Speaker 2:Yeah, absolutely so. We obviously come from having our CCaaS platform storm. It's a CPaaS platform really, for which CCaaS has for several years now been the major application. So we come from having that cloud platform layer and obviously there we are working in an omni-channel environment. It's voice, it's all the digital channels chat, email, socials, internet of things, increasingly. So you've got all of that data coming in together and of course, we look at how do you do AI? Well, it all starts with the data, data, discovery, validation, deployment. So you're in the environment already to do the AI. So we obviously develop some AI tools ourselves. So, for example, around retrieval, augmented generation.
Speaker 2:But in terms of the actual engines large language models, small language models, large vision models, you've got translation models, etc. Those are typically made by numerous other vendors, developers, companies and organizations of all shapes and sizes. Our role really at this platform layer is to orchestrate those. So bring them together so that when a customer of ours has a CX application that we need to put together that, that will find the best AI engines normally plural to support the use case. So it may be that they need a particular large language model.
Speaker 2:It could be they need a particular speech-to-text natural language processor, it might be that they need an image recognizer of some kind or to work with specific spoken languages. So our role is finding, doing the horizon scanning, finding the best components and then surfacing those up with our tools so that they're readily accessible. So our customer generally doesn't want to become some kind of AI scientist and kind of researcher. They just want the best tech that's out there, deployable in a kind of easy to use way through low code, no code tools such as we provide, with all of the assurance and the availability and the security that goes around a trusted platform such as ours. So that's really our role. It's an orchestrator.
Speaker 1:Got it. And as you orchestrate increasingly AI agents and virtual agents and agentic architectures, I mean, what do you see the role of the human agent in this future? How is it evolving and where is it today?
Speaker 2:Yeah. So we always see there's going to be a role for the human. Humans are what create. That's good to hear. Yeah, we'll all have something to do in the future.
Speaker 2:So we see the human agent sitting atop a pyramid, really. So they're at the apex, and then below them are numerous layers of automation. So the human agent is being fed with information about what's going on, who this person is, what they want, what to say. Next, surfacing knowledge management articles, then everything supported around them we like to talk of before, during and after an interaction. So you're a human agent at the top of this pyramid, kind of sitting there controlling it all. Maybe someone's entering into an interaction with our organization, maybe beforehand there's NLP, for example. It's asking what it is that they're contacting us about. It could be on a voice or any other of the written digital channels. So we're segmenting. Perhaps we're able to automate a lot of those inquiries. Others we're going to prepare. Perhaps we'll send a form to complete on your mobile screen and they'll deliver that extra information with the interaction that's then surfaced up to this human agent, and only they're only going to be receiving the cases that really require their attention.
Speaker 2:The rest has been dealt with at these other layers in the pyramid and then they were surfacing up what they need to know. There's life sentiment going on there, how to improve this experience yet further. Perhaps everything is being transcribed for them, the forms are being filled in that they're normally having to type all this stuff into. They're're relaxing, concentrating, engaging with the consumer, citizen, patient, whatever role the individual is coming in as, and they're delivering that great experience. And all of the hard stuff, the typing, the data entry that's all being handled for them. Typing, the data entry, that's all being handled for them.
Speaker 2:And then, of course, after the interaction's over, a lot of what's being considered the traditional post-contact activity, the RAP, that's largely being taken care of as well, and even down to things like the CSAT to gauge the satisfaction levels preparing the quality management, the audit, flagging up to a supervisor, letting the agent know how it went. So everyone wants to improve their game and we don't necessarily just want to wait until the end of the month when some kind of manager comes down like a ton of bricks on them. They want to know, kind of interaction by interaction, how to improve things, optimize them. Maybe it's getting to the end of their shift and their energy levels are flagging and they're being made aware of this fact and just trying to kind of up things a bit to get through to the end of the session. So I'm envisaging this much more supportive environment. Rather than just being monitored, you're being truly supported by the tech.
Speaker 1:Brilliant. Well, that's something we can all get behind. You mentioned Omnichannel a couple of times. So many channels it's amazing, overwhelming, to say the least, for the customer's journey, and not just, of course, the usual voice and messaging, but we've got video and web chats, and apps are increasingly channels, and on and on RCS. How do you support all of these channels and make it seamless for the customer not to be overwhelmed?
Speaker 2:So customers want channel choice nowadays. So if you think of the first stage of going digital digital transformation, it was all about channel shift. So typically it was seen through the lens of cost saving. So an agent interaction costs me this much, I want to maybe hide the phone number and make them go through this potentially tortuous web process. So that's potentially good. If you are the provider of that service and you're just focused on reducing those immediate costs, perhaps it's not so great for improving your CSAT or the customer loyalty. So we're seeing consistently something like 10% uplift in CSAT where channel choice is offered.
Speaker 2:So channel choice means meeting the consumer, citizen, patient on their own terms. So different people use different types of interaction mechanisms at different times for different purposes. So sometimes we just want to check something, check a balance or something like that. We're not going to expect to have to speak to a person to do that and the objective is to get the information quickly and then move on with whatever else needs doing that day. Sometimes it's going to be a complex stroke, urgent emotional type of interaction and the human voice is what we're looking for and specifically we want to talk to a Human person as well, because this is so complex or emotive. So it's about presenting the right channel at the right time in the right situation, and I don't think we want to lay them all out there. Why don't you contact us through all these 10, 15 mechanisms?
Speaker 2:However, I did see a stat last week that the average interaction uses 2.4 channels. Now, so during a single interaction. So maybe I've called up on my mobile, I've spoken to say what it is that I want. We've then categorized the organization has categorized my inquiry. I'm told that to wait on the line, and I've been sent an SMS message with a link in it. I click on that link. I'm then filling in a form, perhaps name and address details, and then sending that form in, so that's being submitted then with the call and that information is coming through to the agent with the call delivery. So we've used three mechanisms right there in a relatively straightforward interaction. I think the key is to make it very simple for the user so that they're not having to consciously think about navigating through the sort of sandbars of all these channels and you're adding to their stress. We want to make life easy for them, even if it makes it a little bit more complicated for the organization, in order to service them in this convenient way.
Speaker 1:Yeah, well said. You have so many customers and partners, deployments, any success stories or anecdotes, customer stories, that kind of capture. What Content Guru is all about?
Speaker 2:Yeah, I mean we obviously we like to focus on customers, using some of the latest tech, of course, because that's always more exciting. So one I like is called Together. So they're a financial services organization. They make loans to property developers, so they're quite big loans. So you're always going to have a spoken dialogue because you're talking about hundreds of thousands, millions of dollars per loan.
Speaker 2:So the person at the front end, the agent, would have this conversation and then they would need to then talk to someone in the back office who's the underwriter, who then approves whether the loan is going to happen or not. So the goal there was to optimize that interaction and try and take some stages out of it. So we've used transcription and summarization primarily to try and kind of eat into this long back and forth process between the customer and the agent and the agent and the back office underwriters. So what we do now is we automatically transcribe and categorize the information from that initial customer interaction they call it an assessment call and then we drop that into various customized assessment templates, uh to categorize the information, uh into applicant case notes. But normally the the contact center agent would have had to create those case notes. So those are then passed straight to the underwriting team for approval.
Speaker 2:So gains there is we cut the average handle time at favorite contact center stat by 50% and also the wrap time for that after call work. We've taken 275 seconds out of that, so it means we're saving six minutes per interaction. And then the verification process that happens after that we're saving nine minutes. So they're quite simply able to get through more work in a day without adding to a user base. So that's one I love from financial sector. Government has also got some great use cases. So in the UK we have a kind of national DMV we call it with the driver and vehicle licensing agency.
Speaker 2:So everything to do with vehicle licensing, or down to license plates or your actual driver's license, medicals on drivers, all of that sort of thing. So the presumption there before was this is all about voice. So you'd go in at the top of the pyramid, probably after waiting a while, you'd talk to someone and you'd work through what it is that you want to do. So what we've done with the DVLA is that we've put in not just a new cloud contact center, but we've also replaced what was a quite derided chatbot with an AI-driven one, which is just fantastically better. So the chat is quicker.
Speaker 2:And we've introduced working up that pyramid idea again more and more use cases that are steadily more complex. So now over 10,000 chat sessions a day are being completed in the chat. So it's trimmed out something like 30% or maybe even slightly more than 30% of all the call traffic by directing these cases into the chatbots, and every month or so we're adding a new use case or two into that chatbot. So that has also then freed up the human agents who are actually there on the phone to deal with the even more complex cases. So those are handled quicker and then 300,000 plus a month are diverted, and successfully so, into the chatbot.
Speaker 2:And that's actually changed not just the perception of the DVLA, but also what it's like to work there. So, of course, if we think of automation and AI, there's always this scepticism around. What does this mean for the worker In the case of a DVLA? We used to have a political sitcom here that was very popular, called yes, Prime Minister.
Speaker 1:Oh, good show. Yeah, it was excellent.
Speaker 2:There was a character called Bernard Woolley who was the principal private secretary or something to the prime minister and he was speculating as to something was going to go really badly wrong with this new policy. And I'm going to get sent to the DVLA in Swansea, which was sort of civil service death, whereas now they're polling not just ahead of their directorate, the Department for Transport, but ahead of the whole civil service as a place to work. So that's almost a kind of unexpected outcome, but it's a great place to work now as well, as it's processing a lot more inquiries a lot more efficiently.
Speaker 1:What a fun story. Very cool. I'm going to go rewatch some of those old episodes. It's a good time.
Speaker 2:Yeah, I know it's timeless.
Speaker 1:Indeed. So the contact center space, as you know, is on fire. It's probably 10 times bigger than it was a few years ago. I think you and all your peers seem to be doing phenomenally well growing. It's sort of this rising tide, but you do have quite a few uh, big tech competitors. I won't call out, but you know very large global companies diving into this space. How do you see that is there's plenty of room for for you and all the other players in this space, and how do you navigate this world of the tech giants getting all of a sudden interested in CX and beyond?
Speaker 2:Yeah, it's true. I mean it's an attractive space, it's high value, the customers tend to stay a long time 10 plus years is normal. Margins are good if you look at it as a business. So why would you not want to be there? Why you might not want to be there is that it's hard to do so, it's hard to get this stuff right.
Speaker 2:It's a very unforgiving environment and I think particularly if, like us, like content guru, you focus on the kind of higher end of the market, the large enterprise, the government and the requirements of reliability, security and actually the sheer complexity of what you need to do is off-putting. It doesn't lend itself to kind of stamping out millions of them. These are long consultative engagements that haven't historically suited the kind of scaling that some of these very large tech companies go for. So I see that they will have the most influence probably in the kind of small C count SMB type businesses up to mid-scale, where the requirements are fairly standardised and lends itself more to almost a kind of self-service ordering type process, whereas I think if we look at some of the kind of things we do, getting through the FedRAMP high uh, that was 421, I think it is separate security procedures that all have to be tested, independently accredited, and even things like sock 2, type 2. It's continually assessed. That's hard and these are. They're not table stakes, but these are expected standards at the higher end of the market. Things like PCI DSS level one to take the payments securely, again, that's non-trivial and you've got to keep it up once you've got it. Lack of any tolerance for things going wrong, that will be the barriers to entry to these markets.
Speaker 2:We're expected to deliver a 99.999% SLA, just generally, but in our emergency services practice we're talking about a 100% SLA. We run a form called Storm ESP for those what we call blue light customers. So we're doing police. Every emergency ambulance call in the UK, for example, goes through the Storm ESP platform to find the best available call handler anywhere in the country to take that call and dispatch the resource from local to where the requirement is. So I think it's a case that commoditization, as ever, will build from the bottom, but I think there's always going to be a place for specialist integrators and people who are going to put together not just for technology technology but the layer over the technology, the systems integrators, the consultancies. There's plenty of room in the market to play because it's an expanding market, because, if you think about it, we're replacing more and more functions that would have been provided by other parts of the organizations who are our customers.
Speaker 2:I was at a dinner last week with a big UK local government customer.
Speaker 2:We were at a conference and they got placed at our table.
Speaker 2:I wasn't expecting to meet our actual customer there and these were some people from that customer who I'd never met before. They'd barely heard of us either and it turns out that their whole function, which is to do with social care, looking after old people in the home or perhaps putting them into care homes, that's now being merged into the contact center team and environment and a lot of what they're doing is now going to feed into our platform, using Internet of Things devices and so on, to create services that aren't even traditional CX but they're leveraging those technologies that the customer has in place scale some of this exciting IoT stuff that's only really been proven out at a smaller scale level or in pilot type environments so far to bring those to the mass scale, to plumb them in with all the relevant systems of record, other information systems that are specialized to those areas. So the scope of what goes in to this new front office is just ever increasing, so there will not be any time soon a shortage of things to do Absolutely.
Speaker 1:And speaking of things to do, you're a global concern. You have deployments across every continent. It seems like AI and Gen AI are purpose-built for multiple languages, dialects, translation and all those services that multinational enterprises need across regions. That must be a huge opportunity for you as well.
Speaker 2:Yeah, it is. So, again, another of the great features of AI is speech to text. It's great for doing all the transcription, but once you've got that transcription, you can, for example, translate it as well. So we are doing a pilot for the UK National Health Service around real-time language translation. So again, the use case there is to supplement, augment the human translator. So a human translator has a high cost, but you've also got an agent that's tied up for several minutes while they find the translator and then they have to stay on the line.
Speaker 2:It's a very long call, even if the issue that they're dealing with is quite a straightforward one. So we developed a front end which is kind of like a walkie-talkie. So you speak one language, it comes out another, they speak their language, it comes out back in your language, and then we produce simultaneous transcriptions of the spoken languages and the translation. And of course, you've got the cool recording there as well. So that's great. But obviously, if you then apply that to a highly sensitive environment like health care particularly this is urgent health care as well it's not just calling the family doctor or whatever.
Speaker 2:This is something's wrong. You're calling an urgent care line. It's got to be safer than waiting and talking through the human translator. So that's not just plug it in. Let's see how it goes. We have test scripts, test actors, a pre-alpha stage, an alpha stage and working through to limited beta using lower risk patients or service users. Maybe they've got dental problems, for example, something you're not likely to die of immediately and then we kind of progress up to a full live implementation. So it's not just for the multinationals, though they're very important as well it's also just serving diverse communities within a multinational environment, such as London. So if you look at the London healthcare language use case, it's like a pie chart. There is no kind of single language. We've got the global language, English. Which version?
Speaker 1:Yes.
Speaker 2:Exactly so. You've got Arabic, as many dialects of Arabic. You've got the Indian subcontinent languages and something like Bengali that then has a number of sub-dialects depending on which bit of the Bengali sphere the speaker is from. You've got all the European languages, the Eastern European languages. The Western Europeans tend to have a good stab at speaking English anyway, so we're not worried so much about those, which is a shame because those are the best of the translation engines. So we're continually working through it. Best of the translation engines, so we're continually working through it Again that's the multi-engine environment in order to be able to tackle patch two or something and how we're going to do that. So we're searching around always to find the best engine fit for applications like translation.
Speaker 1:Amazing, so I'm almost reluctant to ask what's next. You're sort of living in the future already in many ways, but what emerging tech trends or technologies are you excited about as you look forward?
Speaker 2:to the rest of this year and next.
Speaker 2:Well, one thing that I've been finding in recent months is actually how useful small language models are, a lot of the focus being on large language models. But yeah, we've been finding actually getting a good small language model and then applying it with retrieval augmented generation means you're getting a very nice, tight outcome, cutting out the possibility of hallucination, and that's been enabling us to work in some quite difficult environments, like the nhs 111 service in london, for example. So the presumption before was you wait in a queue and then you'll talk to a human, but a human behaving in a robot like way, going through a kind of flowchart algorithm, a bit like a Visio flowchart. Instead, at the top of a call you're asked why you're calling. And then we developed a small language model and a retrieval augmented generation grid that drops you into one of 180 or so categories and depending on that category, you can be not just categorized and streamed off in different directions, but also prioritized.
Speaker 2:So when we set out on that project, I'd assumed it was a large language model type of project, but it turned into a hugely successful small language model and we've been able to roll it out really without any problems. We're now working on some quite complex, multi-categorization work within that program. So it's some of what I see coming is that kind of multi-model deployment really, and that's why I think I'm glad we went with our orchestration way of doing things, because there is no one AI engine maker that's got the lot engine maker that's got a lot, and that will decreasingly be the case in the future. New models and providers are popping up the whole time, and whoever's going to be the leader in 10 years? If there is a leader, they might not even have started the business yet. They're probably in their college dorm room kind of working out how to duck their next exams.
Speaker 1:Yeah, really fun, fit for purpose. It's really the key. Well, really enjoyed our chat. I learned a lot, as always. Thanks Martin for joining. Absolutely Appreciate all the insight and thanks everyone for listening, watching, sharing and be sure to check out our new TV show at techimpacttv now on Fox Business and Bloomberg Weekly. All right, thanks everyone. Thanks Martin.
Speaker 2:Pleasure Bye.