The Rise of Agentic AI

We stand at the threshold of a new era in artificial intelligence—one where AI isn't just following instructions but actively thinking, reasoning, and working autonomously alongside humans. In this eye-opening conversation with Kieran Gilmurray, one of the top 50 global thought leaders in generative AI, we explore how agentic AI is fundamentally reshaping how businesses operate and engage with customers.

TLDR:

• The evolution from RPA to intelligent automation to agentic AI that can think and reason
• Agentic AI creates digital avatars capable of emotionally intelligent customer interactions in 111 languages
• The concept of "one-person, billion-dollar company" enabled by teams of specialized digital workers
• Challenges with AI adoption include skill shortages and cultural resistance to technological change
• The critical importance of high-quality, structured data for effective AI implementation
• Generative Engine Optimization (GEO) as the next frontier beyond traditional SEO
• Businesses must prepare for "machine customers" as AI agents increasingly make buying decisions
• Technology should work for humans, not control them – maintaining human agency is essential

Kieran walks us through the evolution from basic automation to today's sophisticated AI agents that function as digital co-workers capable of handling complex, end-to-end processes. He paints a vivid picture of how these technologies are enabling "one-person, billion-dollar companies" through teams of specialized digital workers that can perform everything from customer service to financial operations with remarkable efficiency and emotional intelligence.

The conversation takes a fascinating turn when Kieran introduces Generative Engine Optimization (GEO)—a critical shift in digital strategy as consumers increasingly bypass traditional search engines in favor of AI interfaces. This requires businesses to completely rethink their approach to digital visibility, focusing on structured data and authoritative content that AI systems can easily incorporate into their responses.

Throughout our discussion, Kieran balances optimism about AI's potential to amplify human capabilities with practical advice about managing risks like deepfakes and data mismanagement. His perspective is refreshingly nuanced—acknowledging legitimate concerns while emphasizing how these technologies can free humans to focus on creative, fulfilling work rather than repetitive tasks.

Whether you're a business leader wondering how to implement AI responsibly, a marketer grappling with the changing landscape of digital visibility, or simply curious about how these technologies will shape our future, this conversation offers valuable insights into preparing for a world where humans and intelligent machines work together to achieve extraordinary results. As Kieran puts it, "We can have everything, and why shouldn't we aim for that?"

Chapters

• 0:00 Introduction to Kieran Gilmouri
• 2:12 The Evolution of AI Automation
• 8:18 Agentic AI in Customer Experience
• 17:56 Balancing Efficiency and Emotional Intelligence
• 24:07 Hurdles and Risks in AI Adoption
• 35:49 The Critical Role of Data
• 42:30 Generative Engine Optimization Explained
• 47:45 Conclusion and Future Vision

Transcript

Sirte Pihlaja: 0:00

Together with AI thought Kieran Gilmurray , we dive into the rise of autonomous AI agents that think, act and adapt like never before and uncover how content strategies must evolve in a world where search engines are no longer the gatekeepers of visibility. Whether you're curious about the future of customer experience or looking to future-proof your digital presence, this conversation is packed with bold insights and practical strategies you won't want to miss. Kieran Gilmurray is a globally recognized leader in artificial intelligence, digital transformation and business strategy. As the CEO of kieran gilmurray. com, he helps organizations harness the power of AI automation and data to drive innovation and operational excellence. A best-selling author of the A to Z of Organizational Digital Transformation and the A to Z of Generative AI A Guide to Leveraging AI for Business, kiran is known for making complex technology understandable and actionable. He is being named one of the top 50 global thought leaders and influencers in generative AI in 2024. And he's a passionate advocate for responsible, ethical AI that truly serves for businesses and humanity.

Sirte Pihlaja: 1:17

Kieran, it's such a pleasure to have you on the show. I've been really looking forward to this conversation. Your insights on the agentic AI and the future of digital work are incredibly timely and relevant. Welcome to the AI Experience Podcast.

Kieran Gilmurray: 1:33

Thank you very much indeed. I'm looking forward to today. I've been waiting for this as well, so we'll have a really great conversation and, hopefully, to goodness. On the back end of this, everybody comes away a little bit wiser, but not in any way more terrified than they already are.

Sirte Pihlaja: 1:50

I definitely hope so, too. If we start by your background, would you like to a little bit tell us about what inspired you to specialize in AI and digital transformation, and digital transformation with, I mean, with the rapid advancements that we have in artificial intelligence, what is the thing that really excites you most about the future of this field?

Kieran Gilmurray: 2:12

The key bit is the elevation of people. That's the bit that's always excited me. How do you put an amazingly well-trained, emotionally intelligent person and that's not always easy, by the way, because statistically, only 36 percent of us are emotionally intelligent, so, uh, two-thirds of us aren't but how do you put amazing people who've been really well trained, really well developed, coached and mentored and really beautifully managed? How do we give them the technology and the direction that they need to allow them to do more? And by say, do more, look, I'm do more. Look, I'm an absolute fan of business, I'm an absolute fan of economies, growing businesses, growing people making profit, and I'm an absolute fan of humanity. So how do we use technology to elevate us as a human race? And sometimes those two things are at diagonal opposite.

Kieran Gilmurray: 3:00

It feels like an and or, but mine is technology, and AI in particular, allows us to have everything. So back in the day, when they went, would you like it at low cost? Would you like it at lower high cost? Would you like it, you know, at speed, and would you like it at quality? And someone said, right, pick which one or two that you want, but you can't have all three. Now we can have everything that we want and then some, but that isn't without impact. We want and then some, but that isn't without impact, because now remember our careers. Our careers were built on us going to college of some sort or university learning something and gradually accumulating knowledge over time for which we extracted value. Now the danger is that we believe that AI can capture all of humanity's knowledge, and therefore, what will humans do when we're no longer rewarded for what we know, when AI can fill in that gap for us? So we live in interesting times, but not the first time. Tech has excited us, delighted us and frightened us.

Sirte Pihlaja: 3:56

That's for sure. That's true Seems like a very high purpose you have there. You've written about the rise of agentic AI. For our listeners, who are maybe unfamiliar with the term, how would you define agentic AI and what makes it especially relevant for transforming customer experiences? I mean, how is it different from automation businesses you already use to transform or automate CX and yeah also sorry, you mentioned in your books or your upcoming book actually the agentic AI maturity level, so maybe you can open up those a little bit for us as well.

Kieran Gilmurray: 4:36

Yeah, let me try to so. Afterwards, by the way, I'll share some of the chapters for free with your readers so that they can have those items and learn a little bit more about this, and I'll give of the chapters for free with your readers so that they can have those items and learn a little bit more about this, and I'll give away the book for free as well. It's being released week by week, chapter by chapter, until it goes up on Amazon, and then, when it does go up on Amazon, I encourage people to buy it, not on the basis that I am a Harry Potter author and that I will make millions, that the book itself, like the two other books that I've written to date, are written for two purposes. The first purpose is, I believe, that all boats rise in the floating tide. In other words, if we share our knowledge widely and widely, everyone will benefit and the economy will grow. So those books were paid for by me, hopefully to allow everybody to grow their knowledge, to get better at learning, digital transformation, generative ai and agentics, and the money from the books goes to social housing charities and charities in ireland as well. So please use the chapters, learn, build better economies and businesses and then give a little money to try to be really cool.

Kieran Gilmurray: 5:41

Look, there's sort of been a continuum. So let's talk about two continuums. Continuum number one is automation isn't new, except this is a different form of it. And if you go back in the day or I say sort of back in the day, but let's move from RPA, which was about 2010. That is RPA was. You know, I sort of follow instructions exactly as they're given to me, so that was more task automation than end-to-end automation. But it was a wonderful example and there was automation before that for anybody who was asking. It eliminated manual effort on wasted, repetitive tasks. So you had to execute structured, rule-based processes, perform repetitive digital items, you know, you had very defined boundaries and you had very exacting step-by-step procedures that worked up to a point. And then you'd intelligent automation that allowed you to, you know, a little bit of coping with more of the messiness and more of the complexity of humanity.

Kieran Gilmurray: 6:39

In the last couple of years we've got degenerative AI. So from following instructions to I can create based on prompts and prompts are just instructions that we write and they can be verbal or written. But suddenly generative AI got popular because of its ability to create things. Of course we create cartoons, a bit like we got Google Maps. We didn't travel the world, we lived in our own back garden, but you know it allows us to suddenly become more productive. You know so. It's a productivity amplifier, for want of a better phrase. You know it allows us to suddenly become more productive. You know so it's a productivity amplifier, for want of a better phrase.

Kieran Gilmurray: 7:12

You know accelerating creativity and analytical work. You know it assists with lots of different tasks. You know writing want to give it, but it goes away and figures out actually how to achieve them. And rather than you being a passive recipient of outcomes or actions from RPA or generative AI, now you're a collaborative actor. You know the need for constant oversight isn't necessarily there. It acts as a virtual you know co-worker that can complete end-to-end processes. It learns, it self-directs, it coordinates multiple tasks and, to a degree, done right, this can cope with all of the things that RPA and intelligent automation promised but never quite delivered in the scale or to the degree that it needed because of human messiness.

Kieran Gilmurray: 8:07

Now we've got what I would call, you know, to a degree, digital workers sitting on top of large language models, allowing technology to think and reason and learn like a human and act like a human in a self-directed way. Now, it's not artificial general intelligence, which is probably your next layer up, and it's not artificial super intelligence, where, you know, at some point, allegedly computers can outperform human intelligence. And again, we have to be really careful about all these terms, because those can frighten the life out of people or we can give AI skills and qualities that it doesn't actually have. So I think we're a long way from super intelligence. That said and I don't mean this diskindly we have to be careful what we mean by intelligence, because you remember our friend Gary Kasparov. He went well, if a robot can beat me at chess, it's intelligent. A robot beat him at chess and, of course, it's not intelligent.

Kieran Gilmurray: 9:01

You know, there's's degrees of intelligence, but this is an exciting time where technology could really help us. And when I say help us, you know we're at a stage where there's far less people being born in most countries. There's a middle group that are going to be the, the key group to fund. You know, an elderly group that are getting to live older, where there isn't necessarily the money in most countries to pay for them. So now we need, you know, more than ever, amazing people who are given productivity and personality and business amplifying tools. And that's where agentic AI, agi, generative AI and all these tools should help humanity. At least you know, run and keep up up.

Sirte Pihlaja: 9:43

If not, run and catch up right if we think in more practical terms. Uh, in what ways do you suggest that agentic ai can reshape the traditional customer journey, I mean from onboarding to support and loyalty and all that?

Kieran Gilmurray: 9:59

yeah, well, we're. We're already seeing some of this, you know so. A lot of people listening is going out, my goodness, you know another new shiny technology that promises the world that doesn't deliver. But you have lots of examples of this in place at the moment, you know so. For example, let's be honest, some of the things that we actually do in customer services are not necessarily that complicated. So a couple of things here.

Kieran Gilmurray: 10:18

I go on a website, traditionally now I click point, click point and hopefully buy, or what might happen is I decide not to buy and then I dropped out of the basket. What I can now do is have, you know, digital avatars, agentic enables, where someone that looks pretty much like you or I and again, you can really drive this home to make someone that looks identical to us, that sounds like us, that looks like us, identical to us, that sounds like us, that looks like us. You know, the deepest fake technology allows us to create digital twins that genuinely, you know, have shadows and mouths and moves and they don't suddenly look like robots, so they can go on the front screen and have what might be described as an intelligent conversation and something that's akin to an emotionally intelligent conversation. In other words, I can put in the inflections, the ums, the ahs, the pauses, except when I'm having a conversation with you as an avatar and I can use screen technology to see what you're wearing. So there's a great example from a retailer. I'll not name any of them, but someone's ringing up, they're on the screen or dialing in and the agent is able to have, you know, a human-like conversation, see what they're wearing, see the color of their hair and then start engaging around some outfits that they may want to to bring to a wedding and just have a really engaging conversation. So well, that's OK. You know, I could jump on a screen as a person and do that and video call in the same way that we're doing now. But actually this is on steroids, for want of a better phrase. So now I can do that in 111 different languages. Now I can do that 24 seven, with genuinely no loss of enthusiasm or the sound of enthusiasm, no loss of productivity.

Kieran Gilmurray: 11:57

I can now bring in all of the personalization elements that I want. So as long as I have a history of you or a bank of data around you as a particular person, then I can really personalize the conversation at scale. So now where you and I, would you know, use our emotional intelligence to look at the person in front of us. Ask us the question. Some of those we don't need to ask, because if we sufficient data and, let's be honest, we pour out our heart and soul online then I can go and scrape the internet and find out about you from looking at your email address or your phone number or, for example, there's software out there at the moment where, if I put in your LinkedIn profile, for example, into an engine, I can get a full psychometric profile of you that tells me whether you're optimistic, pessimistic, what language to use or not use. I can do a full psychometric evaluation, to a degree, of you and I can put all that into the sales or personalized call.

Kieran Gilmurray: 12:51

Now you sort of go with this. That sounds a little bit creepy, but your best agents and your best businesses have been collecting information about you for years. Their job is to train agents to do exactly like they're training a digital worker to do or an agentic AI to do. And, let's be honest, if only 36% of us are emotionally intelligent the fact that a digital worker sounds emotionally intelligent every single time I actually engage with them and they're available for me when I want, in any language that I want, then I'm more likely to go with them, you know.

Kieran Gilmurray: 13:23

So that's one example where I can put that in front of an e-commerce website and do exactly what we're doing at the moment in an intelligent or hyper-intelligent manner.

Kieran Gilmurray: 13:32

You know that information can come in internally and all of a sudden I can have a gentic labor picking up the order, having transcribed the entire call, updated the CRM, updated the ERP, processed the order, had another agent coordinating all this project, managing all this, a finance agent, you know, issuing the invoice, taking away the finance instruction, another agent placing an automated call to keep me updated in terms of emails or texts or voice or something else, all in service of the client, to make sure that I have the most wonderful purchase and continuing to deliver experience that you might possibly imagine, until such times as I get my product, get my invoice, pay my bill, and then you're back in the loop where I can now suddenly upsell, cross sell and do a whole host of other things, with more information on you now gathered, because we've had further and deeper interactions, because there's a lot there there is, there is all happening at the moment, all real technology, and it goes back to the article that I wrote a little while ago.

Kieran Gilmurray: 14:37

It talks about the one person, one billion dollar company where I know, for example I'll not name them because I don't like naming brands, because people assume it's a recommendation. It's not. I want people to find, work out their business opportunity or challenge and then go and find the tech, but I have a company that I'm dealing with, that I'm talking later on to, and they built a one-person insurance brokerage company with, you know, an agentic actuary, an agentic underwriter, agenticic service team, agentic claims team. I know, all of a sudden, this one person business is one person plus a multitude of digital workers, and even Jensen Hang said look, he wants to be a business of whatever. It was 50,000 people and you know, 150,000 agentic workers. So this is real tech that can suddenly help elevate businesses and their own performance at a time when we desperately want and need it and, dare I say, expect it, because we know what the technology can deliver and therefore we want personalized, deep, affordable, 24 7, multilingual, emotionally intelligent experiences that get us everything we want when we choose.

Sirte Pihlaja: 15:40

We want it with all of this technology around? How can organizations balance efficiency and empathy when deploying these autonomous AI agents?

Kieran Gilmurray: 15:51

Yeah, well, let's talk about efficiency. So efficiency is the part easy bit, because this is the interesting bit. We look at RPA, we look at intelligent automation. We look at, you know, 20, 30 years ago ago, I had robots running around a database in a business providing law clients with full disclosure of their legal cases, text updates, email updates. So there's nothing new about most of the technologies that we've got. Maybe you know large language models and generative ai. It's still ai, it's another iteration of it. But where we've reached the phase is that everything is quicker, cheaper, better, faster and more affordable than ever. So the technology's always been there, so we could have done a lot of this before, but now it's so affordable it's unreal. And now, because of, you know, masses of data, cloud computing and cheap ai, the operational efficiency bit is a given. Where you're paying your money at the moment is and I remember this well in RPA days someone come out of a course and they were being paid $50,000, $60,000 because nobody knew the tech. We're in the same space with agentic AI. There's a few of us globally who actually understand this and have been in business 20, 30 years, who have done scripting and automation and digital transformation, done the RPA, iea, the cognitive, the hyper and every other phase that you got. And then there's a very much narrower market of people like me, who was a chief AI officer for 15 years in businesses. So you will get hype. You will get you know, ai overreach, where people are claiming their technology is AI and agentic and generative and it's clearly not and they've never worked in the space. When I'm reading some of the things you're saying, it's absolutely nuts. But the operation efficiency piece is a given. It's the cost of implementing this and the knowledge gap is the only thing that might cost a little bit more money. But if I have digital workers across my entire business and I have lots of agentic ai workers in mind then all of a sudden that one person, one billion dollar business isn't it isn't a mystical something that's going to happen in the future. This is happening now in a way that RPA could never deliver intelligent automation. So operational efficiencies is a given.

Kieran Gilmurray: 17:55

Now we go back to emotional intelligence, the second part of your question, because everybody goes, yeah, but technology is not intelligent. Well, so said the people that got beaten at jeopardy. So said gary kasparov when he got beaten at chess. So said a friend out in in china who got beaten at go. You know what is intelligence? Because I'll tell you this, let's be honest, the, the average person isn't, isn't as intelligent as we might think, because when you and I think intelligence, we think computer programming, a genius artist, someone who can write the most wonderful book, someone who has emotional empathy, and, and, and, and, and, and.

Kieran Gilmurray: 18:31

It reminds me of a scene from the Will Smith movie I Robot, where Will Smith is going yes, but can a robot create the Sistine Chapel? Can a robot create you an opera? Can a robot write, like back, to which the robot went yeah, can you. Let's be honest, the average human can hardly tie their shoelace at times. So, look, you know, intelligence is, is a variety of things. If we're talking about logical, repeatable things, that's, computers eat that stuff up all day and we should be using them for that, to free us up to do the things we're good at. You know the creativity, the critical judgment, but what is creativity for many? Creativity for many has learned examples of past behavior, to join dots together to create something new. There's, there's not so many creative reds in the world or yellows as you might have them in a psychometric profile, but let's look at emotional intelligence.

Kieran Gilmurray: 19:26

I had this figure from a professional psychotherapist. They said like only 36 percent of people are emotionally intelligent. And for you and I, who've been on the receiving end of many a poor customer call and it doesn't matter whether that's in person or online it's that figure certainly stacks up a lot. And that's not to critique people, because I'm sort of sitting there going. It's that figure certainly stacks up a lot, and that's not to critique people, because I'm sort of sitting there going.

Kieran Gilmurray: 19:45

It's incumbent upon businesses to hire the right people and then train them in the right way. And very often and I come across this all the time people hire people, put in technology and then go, yeah, we don't have any money to train them, and you sort of go, you've just defeated the entire purpose of customer service. But now all of a sudden you go okay, well, what is emotional intelligence? And that's's the ability to have empathy with others. Well, there's a superb technology now that's been trained since the early days of large language model ChatGPT2, that has a bank of what you might call emotional intelligence statements, a bank of what you might call emotional intelligence statements that it is rewriting emails, for example, that have been sent into contact centers or service centers that your normal agent, who is poorly trained, would send something out. That would just result in an upset customer.

Kieran Gilmurray: 20:38

Or the bit that I see all the time is, you know, email tennis back forward, back forward, back forward, or phone tennis, and this technology actually re-translates the phrases that people are saying and turns it into what looks like an emotionally intelligent response. That was the early days and flat email. Now in real time, based again on data, computing power, processing speed and cloud technology, as the agentic agents are having a conversation with people. Because you can do this in written text, you can do this on pure audio. You can do this using an avatar, visual text. Now, as they're having a conversation on audio.

Kieran Gilmurray: 21:14

Whatever, it's real-time translation of language and emails and text and imagery and it's playing back emotionally intelligent responses that you or I would find very difficult to discern whether it was a human or not, and you would complement the quality of the responses because you and I have been in contact center customer service for decades and you go. Actually that looks particularly good and that feels like someone who's been in industry for the last five or 10 years. They're really well trained. So now the technology can mimic what we're doing and again, that shouldn't frighten anyone because, again, the best agents are not born well. Some of the best ones are born emotionally intelligent. But the best agents have learned learned through multiple interactions with people how to respond and not to respond, what to say and not what to say, and so on. But we can train that. We can train that in a computer through reinforcement learning, and it's not as complex as we think. Nor are we as humans, apart from our emotions, maybe.

Sirte Pihlaja: 22:18

We would certainly like to be called emotionally intelligent and feel that it's something very, very difficult to gain. But yeah, you're so right.

Kieran Gilmurray: 22:29

And again, it shouldn't frighten people because I'm going back to if I can provide amazing customer service at low cost, then you know we're going to get more customers. You know there's a bigger pie to be had, and it shouldn't frighten people. Well, what am I now going to do? Well, you're going to do a thousand and one other things, because you suddenly need to create all of the personalization, the individual channels and the intelligence and the operational efficiency to run frictionless. Frictionless businesses that delight and excite customers, because the more you do that, the bigger your pie will be. Do that, the bigger your pie will be.

Kieran Gilmurray: 23:06

And all of a sudden, instead of sitting there all day long, spending 80% of your day dealing with, you know, queries or complaints, or the same call type over and over and over again, now you can spend your day doing the thing that you signed up to the day you joined the organization, the bit that got you excited, the bit that now allows you to actually use your creativity, to use your inquisitiveness to suddenly build an even better business that further excites and delights, or open up new markets or does a whole host of different things. So I go back to what I said at the very beginning. Amazing people with amazing technology usually can do amazing things, so let's have more amazing.

Sirte Pihlaja: 23:45

If we go a little bit back to the adoption of agentic AI, what would you say are the biggest hurdles to mainstream adoption of agentic AI and I mean both from the technical and cultural side of viewpoint within organizations, and what are the kind of challenges or risks that companies should prepare for as they adopt it?

Kieran Gilmurray: 24:07

Yeah. So the key challenge at the moment is there's a very narrow number of people in the world like I who truly understand a mixture of business plus technology, plus agentic AI, plus automation, plus have the emotional intelligence not just to throw technology at the wall and hope it sticks. Change because this technology, you know, if last year and the year before was large language models and people were getting excited, the smart money has gone into agentic ai at the moment because we desperately need, you know, more operationally efficient businesses that can work 24 7, never mind the fact we need it economically because we always live in in interesting and challenging times, but productivity wise, we need it because of that change in in demographics. So it's the, the lack of skilled people currently, but that will change because the money goes, what money goes and people will follow for the where it actually follows. So that bit's the easy bit. The harder part is, and it's always been the same and it will always be the same. Let's be honest.

Kieran Gilmurray: 25:10

Uh, digital transformation's dirty little secret, or technology's dirty little secret was well, people don't want to change very often because change is difficult. You know, and you've sort of got that third. The third you'll never change. The third, you will run at anything. It's nice and shiny and bright, like a kitten chasing a piece of silver foil across the floor and a huge group in the middle who were sort of were going.

Kieran Gilmurray: 25:31

What does it mean for me? Will I change, will I not? Is it good or bad? So you know, as people, we resist change when we shouldn't. Every time I've changed, I've grown. Things have been better than the past and, as someone once said, you know, building a business is hard, but redundancy is harder. You know learning is difficult, but not learning results in a more difficult time. Growing a business difficult, shrinking a business hard. You know everything's hard. Pick your heart and for me, every time it's a case of you know, I choose comfort or growth and I want to grow. So comfort can go out the window because it doesn't matter, I have a lot of fun. So people are the biggest resistance. And it comes back to let's be honest, we don't help people very often. And let me explain that comment as well. And why are we constantly surprised by new technology? Why haven't we career path people so that now we're genetic AI, generative AI? It?

Kieran Gilmurray: 26:25

shouldn't be a surprise, you know, but it is. We're constantly surprised. Therefore, we don't have the right people in the right place or the right training. It is we're constantly surprised. Therefore, we don't have the right people in the right place or the right training. So I'm a huge fan of a business strategy, a people strategy and a technology strategy working in coordination. We should always be thinking 18 months ahead of what we currently need. So if everybody career paths and trained to get to where they need to be but of course we don't.

Kieran Gilmurray: 26:46

And then I look at the press, and it's not a criticism of the press per se or governments or whatever else, but, as a journalist once said to me, if it leads it, if it bleeds, it leads and therefore, rather than this being, you know, technology's amazing. It's a bit like right, the robots are going to kill us. The world is over, your job is being destroyed. You know what's technology? Doomsayers as opposed to technology. Instead of doomers, bloomers. You know people who see technology as something that can be amazing, something that can add benefit.

Kieran Gilmurray: 27:16

Now, I couldn't do my job and I work internationally if I didn't have a genetic labor. You know, ai, coming out my ears, it's only 100 worth a month. So this isn't big price. But by god, when I look at 10, 15 years ago, the data science team at a million dollars, now I can do the same thing, quite literally, because I have up to almost the same degree for 16 a month. This is this is nothing, no money at all.

Kieran Gilmurray: 27:41

But we need to change. We need the, the government and the press and everything else to be more positive. But and there's a huge, huge but, or maybe a huge and to your question on risk, we can't just suddenly go AI is awful or AI is brilliant and not understand it, because AI comes with risk, as does any technology. And that's not to diminish AI's risk.

Kieran Gilmurray: 28:07

Because we mentioned deep fakes earlier on being used for brilliant customer service in multilingual languages to get me what I want in a frictionless experience to drive an operationally efficient and profitable business brilliant. But I can use deep fakes to commit fraud. I can record a bit of video that you have had, or record your screen of something you've published on YouTube and then I can go out and defraud your clients and that's happened. You know, $25 million. Or there's a chap in the UK who spends a lot of his time genuinely trying to help people make sense of financial, get them more financially literate, and the guy's brilliant, but he is the most deep, fake, deep-imaged person that you've ever came across in your life. And now it can sound like him and look like him, and you and I understand tech and it's not to criticize your.

Kieran Gilmurray: 28:53

Your your average person, but most people don't, and therefore it looks and sounds in every way possible like him and the company in hong kong was ripped off for 25 million for this, where they invited people into a fake c c-suite meeting in zoom, and you know. So we need to teach people like what can be used for good can also be used for not so good. So and here's how to detect it. On that one, by the way, in case people are listening, I always sort of said, you know, for sums of over a certain amount of money or a percentage of importance to your business, have a code word. The same thing with families. I encourage people to read articles. On my online linkedin or website, karengalmurraycom.

Kieran Gilmurray: 29:29

I write about sharenting, and sharenting is where people have deep, faked their children's voices and all of a sudden there's a panicked sounding phone call that sounds like your child saying I've lost my phone, I'm deeply in trouble. I'm on my friend's phone, I need you to transfer money to this account and they get ripped off. You know, no-transcript joyous bit that shouldn't put you off, because I'm reading articles all the time going oh my goodness, gen ai can get it wrong. You know, 10 times out of a thousand. And I'm going oh my god, two years ago it was 200 or 400 times out of a thousand. Next year it's going to be once. Have you ever met a human being that got things right 990 times out of 1000? If you do, I'll hire them tomorrow morning. You know. So again, it can still make mistakes, but guess what? So can we as humans? You know and, and, and.

Kieran Gilmurray: 30:45

So there's about 10 or 11 different risks that when I'm delivering programs to C-suite or executive teams or businesses that I call out, once you know them, you're more cautious of them. Just like you learned the risks of driving a car, just like you learned the risks of any technology. Just like you learned the risks of having a conversation but forgetting to mention to a vulnerable client about a credit policy the risk of them taking out something. So risks apparent in everything.

Kieran Gilmurray: 31:15

But we just need to make sure that we invest in people, invest in managers, invest in calling the good and the risks out so that we account for all these things. But very often we take shortcuts as businesses and we don't cover off all the things that we need to. And guess what? We suffer for years, the things that we need to, and guess what we suffer for years afterwards, and somehow we celebrate when we fix them. We don't celebrate and we forget what we should have done in the first place. But not a criticism. Everything's an opportunity to learn and grow where I'm coming from, so let's learn and grow.

Sirte Pihlaja: 31:45

One thing that maybe you didn't mention there if we think about the role of data, what role the data plays in enabling agentic AI in CX? How can businesses prepare for their data infrastructure for this shift to agentic AI?

Kieran Gilmurray: 32:06

Forget agentic AI for a moment, and I will come back to it. Forget intelligent automation and RPA. Good people need good data to deliver exceptional customer experiences, and that's everything from back in the day to remembering your customer when it was all done in person, making good notes that were readily shared across your business to allow you to engage with someone in a personalized manner. That felt and should have been intelligent when we, in any type of technology crm, erp, finance system, procurement system, rpa, intelligent automation, generative ai, language model creation or agentic ai everything is fundamentally important on the quality of the data that underpins it, be it a human dealing with it or a machine dealing with it. So we should not, must not, should never have forgotten that Again. It's always surprising for me that companies have made a mess of their data for decades, and I'm not talking about all of a sudden now they need an AI strategy or they need a data analytics strategy. You should always have had one. You should always have looked at. 30 years ago we were putting together, you know, descriptive and diagnostic analytics, long before we were talking about predictive or prescriptive analytics. We should have done that, but we've taken shortcuts for years. We've made a mess of data. Client records weren't accurate on paper, nor were they accurate in the first databases, nor, nor, nor, because it all took a little bit of of time to do or a lack of training or a lack of software. Now we're in the digital age where you know the exhaust fumes, or every fume or every piece of fuel is data related. So if we want to deliver, you know, better services, then we need better data, and there's a compromise there between the amount of information we collect for purpose and the amount of information we could collect on customers. But we need good data to run every service. You know to pay someone correctly, to process an order, to feed the agentic systems with the insights that they actually need to run. You know predictive, prescriptive. You know diagnostic and descriptive analytics. Data is the gold dust. Now again, be careful with getting you know. So het up about it, because everybody goes right, I need a data strategy and all of a sudden I'll make my data perfect and three and four years later, having mapped their processes and got every piece of data hopefully perfect, they suddenly realize the world has moved on. So we're in a bit of a phase now where, yes, we need a data strategy, yes, we need to get our data right because it fuels everything and will for the next 20 years. So take the pain now for the next 20 years worth the benefit, otherwise you're just going to struggle.

Kieran Gilmurray: 34:42

But data doesn't have to be perfect for everything. For a cancer scan, absolutely. For understanding the average number of customers who might renew every month of an average year, not necessarily. So don't over index on that and that alone. Go back to what I said is have a great data strategy, great people strategy, great tech strategy, great business strategy and work out where you put your money and your time. But get your people trained. Give them the tools that they need, be it data or language models, or agentics or crm or erp, career path them. Have a strategy, communicate like hell with everyone and get all of those components right, because they're all lego bricks that need built to allow you to build. You know, the hospital today, the prison tomorrow and the successful business the day after. There's no, there's no shortcuts, um. So take your pain, take your medicine.

Sirte Pihlaja: 35:31

That's for sure. Love the metaphor of doing that as Lego bricks for obvious reasons with my background, If we shift gears a little bit here and talk about another one of those Lego bricks. Generative engine optimization. You recently wrote a blog about this concept. Can you explain what GEO is and how does it differ from traditional SEO?

Kieran Gilmurray: 35:55

Yeah, look, this is the crazy bit. I rarely go near search engines, you know, whereas I would have spent a huge amount of time researching and it made me think, my goodness, if I'm doing this, surely, as more and more people use AI tooling on a daily basis and go into large language model interface like ChatGPT, gemini, perplexity, whatever it is, then, oh, I'm not actually in front of a search engine. If I'm not front of a search engine, then that business model doesn't work, because they're collecting my data to make their algorithm smarter to present adverts in front of me. And now I'm not and I'm not the only one starting to notice this. I deal with some companies. So, for example, one company in Ireland I work with is a large gambling company. Their business model is built on the preposition that people go online and search and they present ads.

Kieran Gilmurray: 36:46

Well, as more and more people get like me and open up every day, you know, when I'm up in the morning, I've got chat, gbt, perplexity and a whole host of other tools. It's my first opening as opposed to google, to find things and do deep research or reasoning or creating, and remember your google search or your or your bing search or bard search. They can only search, they can't create or do. Now, all of a sudden, I've I've different and more unique demands these days. Then if I'm not using it, a whole lot of people won't, and about 400 million people are using plus one of the tools never mind all the tools in the daily basis.

Kieran Gilmurray: 37:20

The model doesn't stack up. So I'm sitting as a vendor going oh my god, in this digital age where most people are using ad campaigns to get people to their, to their websites or to see their offering, it won't exist anymore, or in much reduced numbers or declining numbers. Therefore, you need to do something called generative search engine optimization, which is how do I actually construct and update generative engines and their spiders as they come across my website, to come and to get to me? So in the past, where traditionally SEO might have been, you know, keywords and backlinks and on-page optimization. Now we're talking about you know, entity-based, structured data presented in a contextualized, relevant way that I hope encourages you, instead of clicking on a SERP, but to click on or receive, you know, synthesized, conversational answers that suddenly highlight that you'll want to come to me as a particular company.

Kieran Gilmurray: 38:13

You know, instead of organic traffic or click-through rates. I want to be into AI mentions or brand presence in generative summaries. I want to be seen as an authoritative source so all of a sudden, authority signals become, you know, a lot more important. I want to be seen as a verifiable resource, so I can't write, you know what I describe as fluffy, happy, clappy content that suddenly puts itself up on a search engine because it looks semi-relevant. Where I've click baited you.

Kieran Gilmurray: 38:40

Now it's about, you know, referenceable, cited, trusted, verifiable content that truly suggests that you're an expert. So you know that truly suggests that you're an expert. So you know I want AI that pulls content into direct answers, you know, without additional clicks, and I need to present and structure that content in a manner that the generative AI engine can find. So we're talking about, you know, structured content. You know adopting entity-based SEO. You know having, you know, writing in a conversational natural language process, ways that it can do it, leveraging schema markup or knowledge graphs that allow me to have my information in interconnected databases that generative AI can consult to answer queries. But that key piece of showcasing authority and credibility, so that cited resources link to reputable references like mine and include statistics and expert quotes and everything else.

Kieran Gilmurray: 39:45

All that needs to change if you actually want to appear in a generative result. So let's just say companies won't be bored for the next couple of years. But I put this out here again to alert companies and SEO teams and marketing teams and sales teams to quickly jump on the back of this and again the same thing. You know like. It's not about putting it out and encouraging people to panic, but I've written an article recently on generative engine optimization a new frontier in content visibility. You'll go to the website it's there. Start diving in. Get ahead of this before this really becomes a tsunami that washes over a lot of what you're doing.

Sirte Pihlaja: 40:27

Yeah, that's certainly my interest because we're talking about GAIO. Yeah, and that certainly piqued my interest because we're talking about GAIO so generative AI optimization for machine customers because we are more and more seeing that on the customer side, people are taking their own digital avatars or their digital assistants to help them in doing their AI-based business decisions, buying decisions and so on, both on the B2C and the B2B spaces. So it's essentially we're talking about B2A, which is business to agentic or agents kind of business, and that's a huge, huge thing that is going to sweep over the businesses, over the businesses like a fifth of their existing businesses is going to be, you know, based on whether or not these machine customers keep coming to buy from them. And that's definitely something where you need to optimize your websites and the content, both from the kind of point of view that they're accessible for these AI agents or machine customers, and also from the content point of view, which is also important for GEO as well.

Kieran Gilmurray: 41:44

Yeah, look, you know, and this is the beautiful thing you and I are calling this out to people now, before they're surprised. So there's still time, and that's that's not to say to folks. Just relax and take it easy. You know, don't worry about it, this will happen someday. That's to start to say look, this is the starting pistol at the start of the race. You know you're in the race, but do do something about this.

Kieran Gilmurray: 42:09

Before you know, generative search becomes the more dominant search. Uh, adjust your business models. Both are highly relevant for the next period of time and I would suggest probably both will be relevant over a period of time and you'll start to see. You know folks who are going out on the generative in other words, the generational content stage, all of a sudden adopting, you know, deeper research models. So I'm starting to see, you know search engines, as we call them classically, building in a genetic labor and generative AI into them.

Kieran Gilmurray: 42:42

So if I'm now still using search and I tend not to myself, you know I'm looking at. You know I've got a family of two. I'm thinking going to, you know, norway this year, whatever else, by the end of that entire search sequence I can just say to the agent. Take off, you go. You know. I've searched, you've got my exhaust fumes. Suggest four holidays, get ready, get the flights booked. You know, do whatever and allow me to press the button and then just go ahead and do it. You know. But if I'm, you know, dependent upon people clicking on an ad or making the next phone call or something else, I'm going to be sorely disappointed in the not too distant future.

Kieran Gilmurray: 43:16

They're just the world has changed. Time to you know again. As I said, choose comfort or growth. Why not choose growth? It's more fun.

Sirte Pihlaja: 43:27

I'm wondering how we could make your machine customer choose Finland over Norway.

Kieran Gilmurray: 43:34

I was in Norway two weeks ago, which is why, but if you invite me to Finland, I'm sure yeah.

Sirte Pihlaja: 43:37

I do. You're very much welcome to visit here. After all, we are the happiest country on earth for the eighth year counting.

Kieran Gilmurray: 43:46

Well, my wife will be delighted. It is the place that she keeps telling me she wants to go for forever. So we did get to Norway, which is one of her things that you know. I was a very good husband that particular week where I made one of the dreams come true. Now, if we get over to Finland, I could be the husband good books for years afterwards.

Sirte Pihlaja: 44:05

I know I know Good for you. Hey, we're approaching the end and the closing of this podcast, but I wanted to still ask you about your long-term vision for how agentic AI and GEO and all of these things that we have been discussing, how they will shape the digital experiences and, why not, physical as well.

Kieran Gilmurray: 44:27

Yeah, look, so we'll end up with more technology in our lives, not less. The bit that I want to see happening, though, is that technology works for us. I was standing on stage in Dublin in front of 450 human resource officers recently, and one of them almost panicking panically, is that such a word? Panicked, a little bit, sounded worried in their voice, saying but what if the AI tells us to do this? And what if the AI tells us to do that? Panicked, a little bit sounded worried in their voice, saying but what if the AI tells us to do this? What if the AI tells us to do that? What if the AI makes us do something? And I went AI can make us do nothing In amongst using this amazing technology to allow us to again, as I said, grow amazing businesses so that we expand the economic pie for more people and ourselves, in addition to, you know, hopefully using AI and other technologies to improve society.

Kieran Gilmurray: 45:19

You know, as someone said to me recently, isn't it amazing that the best minds in the world, the greatest PhDs or computer coders or AI geniuses that we've ever met, instead of fixing humanity's greatest medical problems or working out how we click on ads more often? I had to laugh when I heard it and think at the same time well, that's pretty true. I'd love us to see us grow businesses and grow society to be an amazing place, but we will have a huge number of AIs and agentic AIs in our stable and where someone said there's an app for that, there'll be an agent for that in the non-too-distant future. But let's make sure the agents are working for us under our control, that we don't lose our human agency, our ability to question and to query and to challenge, and let's make sure these agents or other technologies free us up to spend more time to enjoy with our families, our friends, to do things that truly excite and delight us and humanity in general. We can have everything, and why shouldn't we aim for that?

Sirte Pihlaja: 46:28

That's beautiful.

Kieran Gilmurray: 46:29

Let's have everything because we can Thank you, kieran, that's really, let's have everything because we can.

Sirte Pihlaja: 46:32

Thank you, kieran. That's really beautiful advice. How can our listeners contact you if they would like to further discuss these topics with you?

Kieran Gilmurray: 46:39

Yeah, I spend a ridiculous amount of time on LinkedIn because, again, I've written hundreds of articles to give them a way to try and help other people. It's a sort of mission of mine. So get me on LinkedIn. Kieran Gilmurray there's only three of me, I think, two of which are me for some reason but get my website, kierangilmurray. com, where I put all of my content and then, if you want, look the books that I've written, and the money does go to charity. It doesn't come to me at all.

Kieran Gilmurray: 47:05

Those are all on Amazon. They're written specifically to help people, in a non-technical way, understand all of these things and learn about what it means and where it's coming. There's huge numbers of practical examples in those, and the money from the profits from the book whatever our friends Amazon leave left over go to a really good cause. So not only will you, hopefully to goodness, be benefiting yourself, but you'll be benefiting some super charities as well. But if I can help anybody personally, give me a shout. I've got a digital twin on my LinkedIn, digital twin on my website. I'm constantly keeping that up to date. So I'm on hundreds of articles, three books so far and a fourth one on the way, and a digital twin and myself, then hopefully the goodness you'll be able to get answers to help you.