EP 05: AI Takeover Credit Shakeup and the Foundever Fallout

[MUSIC PLAYING] Welcome back to Credit Lens, Europe and Beyond. I'm Chris Hafendon. And I'm Phoebe Epinting. This is where we break down the biggest credit stories across Europe. From high-sakes, restructuring, and the dynamic debt market to unexpected twists, and turns, and policy, trade, and even culture. And today we're diving into credit and the fourth resolution, artificial intelligence. What's new? What's working? And what is the intersection between credit and AI? Well, also confront the ethics and discuss AI as a potential force for good, with a healthy dose of dumerizim's for good measure, as well as the oft overlooked environmental impact of AI. We will start, however, with how AI has disrupted some of the credits that we cover in the business processing outsourcing sector, such as found Ever and Transcom, and how it could impact the SaaS sector with the advent of AI agents that enable users to vibe co-bed, collaborate, and deploy applications directly from their browser. Then we'll wrap up this episode with a few light-ahead lines about some of the interesting and eyebrow-reason ways people are using AI in everyone's favorite afternoon tea segment. And firstly, for compliance reasons, the show reflects our own views, and it's from informational purposes only. It does not represent the views of Octus or its affiliates, and it's not investment advice. Right, Chris, with that out of the way, let's get into it. It feels like every other headline these days is all about AI. Does it affect any companies in Octus' core coverage? The short answer is yes. AI is shaking up the business processing, outsourcing, also known as BPO sector. Shaking up is an understatement. It's an earth-shattering shift. For BPO's Generative AI, isn't just a shiny new toy, it's a fundamental game changer. We're seeing AI-powered chatbots, virtual assistance, and robotic process automation handle in repetitive tasks from customer inquiries to invoice processing. In theory, this means humans are free to focus on the more complex high-value interactions. The message is clear. For BPO's, AI integration isn't a nice to have anymore. It's a matter of survival. Survival? Is it really that existential? Totally. OK, so-- But before we dive deeper into the impact on the BPO sector, can we quickly clarify the types of AI we're talking about? It feels like there's lots of terms and that kind of nissings out there, and they get thrown around quite a bit. Absolutely. I had so much fun, like, deep diving into different types of AI. And that's a great point for clarity. So when we talk about the big disruptor in the space, we're focusing on Generative AI. Think of it as the ultimate content creator. It's built on massive language models that learn from vast data sets to identify patterns. Then, based on a prompt, it can generate entirely new content, such as text, images, videos, and even music. Examples everyone knows about will be OpenAI's ChatGbT. Then you've got Anthropics, Claude, and Google's Gemini, my personal favorite. OK, so Generative AI creates content with fire human prompts. But we also hear this phrase of autonomous agents, or as a Genic AI. They're the ultimate independent contractors. So they perform tasks independently, make decisions, and use various tools without needing sort of constant human oversight. And I suppose in these areas, you're talking about robotics, self-driving cars, or highly complex automation where continuous human input will be in practical. Yeah, so you're thinking more like Waymo, have you heard of that taxi in America that you can go around without driving? But you also have AI agents, which are also like super talented assistants. Think of your virtual assistant on your phone, but in a more sophisticated scale. They still rely on direct human input to function effectively and perform tasks. They're supposed to be excellent at what they do, but they don't operate fully independently in the way autonomous agents or a Genetic AI would. You've got companies like Replic in the space, Manus in the field. And I think they could potentially disrupt their software as a service business model, because assisted air functionalities enable users to code. So all you've got to do is speak into what you want into Replic. And then the AI translates it into some kind of code, and then you've got your output. And this allows them to deploy applications directly from the browser. It's subtle, but crucial difference in their level of autonomy. Now, we can pick out an AI type from a lineup. What kind of impact are we talking about in terms of job displacements and operational changes? OK. Obviously, it's quite a big question. So let's try and bring it back a little bit to the sort of narrower BPO sector. Sure. So the estimates out there are 25% to 30% of BPO operators could be replaced by AI within five years. But some experts think it could be as high as 40% to 50%. And given the fact that this is a very, very big sector, which was growing fast, this has led to quite a lot of re-rating of some of the BPO operators and the sell-off in-depth prices of companies that we cover such as Fandever and Transcom. And then this is despite huge growth forecast in the sector. It's expected to exceed 525 billion in total as a dressable market by 2030, according to the venture capital firm, Anderson Horowitz. That's a massive market, Chris. But it sounds like maybe the companies who can adapt in the quickest ways will benefit the most. We've seen this situation play out with Klaners, isn't we? Oh, old friend Klaner. Regular list as a remember, we referenced the Klaner in episode five. And it's a perfect illustration of the current volatility within the sector. So in 2023, Klaner outsourced 500 customer service jobs, mostly to Fandever. But then at the start of 2024, they made huge headlines claiming they could cut costs materially by bringing a large chunk of their call center work back in-house using this generative AI solution. But they walked back on that, right? I think I read somewhere that CEO Sebastian-- I don't know how to pronounce his name-- I'm going to try. Semia Towsky later admitted that his pursuit of cost cut and had gone too far and that they were reinvesting in human support. It highlights a crucial point, I think, where while AI is powerful, it's not perfect. And the nuances of customer service still often require human empathy and judgment. Despite this, though, other Blutch-Up companies like India, Nadia Bank have all launched their own in-house virtual agents, which is causing some significant concern across the credit markets for service providers. So the perceived threat is real, even if the execution is still finding its footing. What's the current state of play for BPO's dead Chris? Are they panicking? Are they embracing it? They're embracing it, but they're embracing it cautiously. So many think that generative AI is an opportunity to reinvent themselves and offer enhanced services. Anything with AI linked to it, hopefully can have a higher sticker price. Interesting, the slowdown in BPO earnings we've reserved in 2024 was primarily driven by clients deferring investments due to the tough economic climate. So it wasn't that focused on AI, but I suspect if you're thinking about renewing a contract in a year or two's time, you might be considering whether you want to go with an AI solution rather than just renewing your contract. So I don't think we've seen the full force of the AI and its impact on the sector yet. But and the real hit is probably to come further down the line. Well, that's a crucial distinction though. Let's talk about some specific players at Octas were particularly interested in FoundEver and Transcom, among others. How are they navigating the space, do you think? OK, let's take them one by one and start with FoundEver, which is a US call center operator used to be called Cytel. They have a pretty large cap stack of just under 2 and 1/2 billion. And their 2028 term loan fees are currently trading below $0.60. So you know, it's deeply distressed. And numbers haven't been great. They just reported a week Q1 25, revenues down 7%. You could dial down almost 13%. And that was mostly driven by customer churn and lower new client wins. So they're not able to replace the customers they're losing. And they're putting their hopes on cost cutting as a way of trying to recoup some of that lost grounding in Q2. But there is clearly going to be a some formal debt solution that's going to have to happen with FoundEver. And we reported ourselves at Octas earlier in the year that concerned lenders had organized with 80% of them signing a corporation agreement. And there was the expectation that the company was going to come to market with an enemy proposal in April. But it hasn't appeared and we're still waiting for it. Yikes. And Transcom, the Swedish player, could you paint as a picture of what's going on there? Yeah, they're smaller. They've the main sort of debt that we cover is the 2026 SSNs. It's 380 million euros, which would you in December 26, that got a little bit of one way still. But they are trading at quite stressed levels. They're trading just below 80. You're doing 24%, 25%. They are starting to see a little bit of a turnaround It's a very, very soft first half of '24, the improvement started. started to feed through and we actually did see modest revenue growth and EBIDAR growth 1% and 3.2% in the first quarter. And I think the key metric for them is the AI-enabled solutions penetration of the market, which is actually quite high. And they have been talking up their possibility that they can refinance these December 26 months, they're talking about saying they'll come to market 18, 12 to 18 months before maturity. So that's pretty close to now where they should be starting to talk to bondholders about refinance solutions. However, the leverage is still quite high, given where the debt is currently trading, the expectations are that you wouldn't be able to refinance. So we think a demand and extend is more likely than a straight refinancing. So as a credit investor, if you had to take a position in either of the two companies found ever or trans-gombons, what's the play here? It sounds like a very tough choice. It is. I think investors agree that genitive AI won't make this sector obsolete, but it will certainly impact the near-term profitability because AI will drive down costs and revenue growth rates. And if you look at founder, ever, the loans are trading at levels that reflect the risk to the sector. And while trans-gomb has got that sort of near-term maturity, it does have some operational resilience and a more proactive approach to integrating AI into its platform. So in a way, you could say trans-gombons in a slightly better place. So theoretically, if you're forced to choose between the two, your side in with trans-gomb? That really depends on your risk appetite. And I'm a little bit risk-averse, which is a bit weird for a former distress trader. I suppose, found ever, offers the best opportunity for capital appreciation, but there is much much higher execution risk as factored in by the price. Trans-gomb is much further along with that AI integration. And arguably, if it does show some more quarters of improvement and the risk appetite for high yield remains, which at the moment you can get lots of deals away in a risk-on market, then it could potentially get a refi done. However, I think a lot of investors will maybe use this as an opportunity to get out of these credits. They've taken a lot of pain on the way. And many might want to see how the implementation of AI into this sector truly plays out before committing. So where would you say the smart money is right now? Probably on the sidelines or looking at maybe some of the larger, more diversified players. There are investment-grade peers out there, such as concentric and teleperformance. And given the fact that they're more financially resilient and much larger, then it's probably good to see how they perform and make a decision from there. And for those who want to know more about the sector, our webinar replay from last November is a great starting point. Well, you've heard it, folks, for credit, the smart money's on the sidelines. Chris, I feel like all this talk about AI. It would be remiss of me to forget to mention that Octis has its own large language model or GPT called Credit AI. This launched in November 2023 and is trained by our own in-house credit and legal analysts. So it essentially provides analyst-grade responses to complex credit questions. It learns from over 300,000 intel that we have, Octis prioritising credibility and constantly improving through feedback. If you're a subscriber, give it a shot. But before we run away with the world of possibilities opened up by AI and Credit AI, let's bring it back to it down to earth for a minute and talk about the ethics around AI. Chris, with a recent surge in investment in media coverage, it would be so easy to assume that AI is a modern invention. But you'd be wrong. It has actually been a journey almost 70 years in the making from the crucial code-breaking efforts in the Second World War to the very first patent recognition networks popping up in the 1950s. AI has deep roots. It was the American computer scientist John McCarthy who co-authored a document that first coined the term "artificial intelligence" in 1956. Did you know that? No. I'm going to throw one back at you. The concept of singularity, which is the point where technological growth becomes uncontrollable and irreversible, resulting in unfathomable changes to human civilization, was first introduced by John von Numen in 1958. It's this idea that venomins in science, particularly in informatics, nanotechnology and biology, will lead to the invention of AI artificial general intelligence. Now some argue that we've already arrived at this level and singularity and we're now hurtling towards superintelligence singularity. Which if you think about it, is both exhilarating and a little bit unsettling. Imagine having an employee with a master's degree who can speak multiple languages, who's whip-it-smart but not wise, as it's still learning, can work 24/7, never ask for a holiday or sick leave and has absolutely no family drama. That's essentially what AI offers. Which human being, what and all, can compete with that? And if the experts are right, that autonomous agents are only two to five years away with full-scale adoption, Chris, it makes me wonder, how will humans even begin to compete with that? And what do you think will happen in the base case? That's a great question. I don't. I think in my personal life and personal career, I think I've been running from technology for the last 30 years. As a trader and as a journalist, I've been trying to get into things which are more complex and moving down the credit curve so I can avoid being replaced by a machine. Smart thinking. But I suppose it's also a question of whether the sector is hyping up their capability as well. I mean, we saw Sam Altman in his blog post recently saying, I think he titled it, "Gental Singularity," where he said, "We're past the event horizon that take off and started. Humanity is close to building digital superintelligence." So, and then he says, "We've recently built systems that are smarter than people in many ways and we're able to significantly amplify the output of people using them." And this next bit really struck me, in some big sense, chat GPT is already more powerful than any of the human that's ever lived. And we have hundreds of millions of people relying on it every day and for increasingly complex tasks. So, that's a profound statement. And if we are relying on it so heavily, do we need to be thinking a little bit more about the impacts of whether we understand, not just the ethics, but whether we understand the implications of this as well? Chris, I don't know. It's actually a little scary for me and it makes you pause, doesn't it? It seems like most tech giants are now fully invested in this AI race. Do you see something called a race condition? The idea that if one company doesn't do it another will, so then everyone just does it. But then you also have this terminal race condition where I think that's far more alarming, which is there's runaway incentives for AI systems to just accelerate their own capacities and definitely potentially leaving humanity far behind. Yeah, and that's the thing. Do you just let them accelerate away or do they actually find some sort of way of containing or putting guard rails around them? One of the things that to my mind is interesting is we're going to play out the same way we had with the social media platforms and the impacts of social media. Do we allow the companies themselves to self-regulate? Or do we actually have to provide a higher level regulation and controls around it? We've seen prominent figures in the AI world express a lot of concerns. Some through their actions and some through their words. Ilya Sitskever, who was one of the co-founders of OpenAI, recently left OpenAI to start something called Safe Superintelligence Inc. And that is a safety-focused AI startup. It makes you wonder why he left to start that. Yeah, and especially the other interesting thing about OpenAI was it played out very much in public about this whole idea, about the investment in OpenAI, the potential issues with the Sam Orton going to Microsoft and then coming back about whether it would still be run by this charitable foundation for the wider good or whether it's actually going to be run for profit. And that's one of the other problems is that there's so much money that has to be raised for these AI companies and for their models and for the technology and for the infrastructure, which we'll probably talk about a bit later on. So in a way, they have to find ways of fundraising as well. So there's a bit of a split between that dynamic of doing good and actually trying to make money. But can they not just be both like true? You can do good and still make money there. Yeah, and we've seen that with companies this whole idea of becaupts as well. Yeah. It's social, socially responsible companies. So yes, you can do both. I suppose the most interesting book that I've read on this is from Mustafa Suleiman, who was the co-founder of DeepMind. The book was called The Coming Wave. And he was talking about ensuring that AI systems aligned with human values and interests and finding ways of stopping bad actors from gaining access. So he talks about containment. How can you limit the dangers of AI while still harnessing the benefits? And he argues that stability and trust are needed first before you then start to move into this area of superintelligence and this acceleration beyond our control. And I think it's also not just existential risk, Chris. There's a very real immediate concern about widening inequality and the gender-used disparity. I read in a time article last week that women are potentially being left behind in AI. So I think this is probably a call for more women to get engaged so that they don't become dinosaurs and wiped out. Is this also a tech industry thing as well? You know, it's very sort of male dominatio. Male dominatio, yeah. And also a very interesting statistics from the international labor organization, which says 25% of global employment falls within occupations potentially exposed to generative AI. With that number rising to 34% in high income countries. So that's like a course for concern. Yeah, that's a significant portion. And I think the gender gap within that exposure is stark as well. It is. The ILO stats show that exposure among women continues to be significantly higher. In high income countries, jobs at the highest risk of automation makes up 9.6% of female employment compared with just 3.5% of such jobs among men. So it might not just be that AI will necessarily take our jobs, but rather people using AI will result in fewer jobs overall. It's subtle, but quite an important distinction. Indeed. And I suppose that leads us to a critical question of how do we prepare the next generation for this new world? What happens to entry-level jobs? What do you do with entire generation of people who've gone to school? Got them degrees, but find no place in the workforce that's been largely automated? Or are we just being doom arrests that some will call it doom risk and being a bit premature? I don't think so. I mean, sometimes I think if I had a kid, how would I advise that kid in the job world? Like, I don't know what-- I can't say be a doctor, be a lawyer, because potentially AI could disrupt all that. Or maybe they could still be that. But they must have some kind of AI skills to navigate the new reality in which they'll live in. But I guess the question is, are we as a species outsource and are very core competences of thought and agency? As a business owner-- not that I am, but if I was, I guess I'll be thinking, well, automated basic function sounds incredibly exciting for efficiency. But as a human being, the idea of a gentle singularity where machines truly eclipse human intellect, that's a bit terrifying, isn't it? Have you heard of the term not your model, not your mind? So this is the idea that if you didn't build this model, then whose biases are you actually buying into? So if Altman believes a certain world view, then he models that in chat GPT. And then that's kind of the model that's reflected to you, not anything that's actually authentic. It's very kind of bent towards someone else's bias. I think I read something recently about some of the models and they were actually saying that they were tailoring their answers more to what you wanted to hear, and very much reflecting your own individual biases rather than being objective. And I think that's something they're trying to counter in sort of the future versions of some of these models. Yeah, but it's not all dystopian, is it Chris? No. I suppose there's the economic impact versus the human impact. And there is potential force for good. You're now seeing AI tools emerging, which such as synthesis tutor, which is used to teach master kids. And AI helps you personalize the learning experience. And you can adapt for people's learning style and competence as well. And synthesis tutor is cheaper. It's about $29 a month compared to $20 to $50 an hour for a human mass teacher, so the cost is significantly cheaper. So potentially that could be a force for good. And by making everything so much more efficient and easily scalable at a fraction of the cost, I think never before in history has there been an easier time to set up a business, scale, and generate wealth. So you can go from idea to implementation in a matter of minutes in a few keyboard strokes. Or you need some vibe coding from the comfort of your home where you describe into some AI agent chat box and replete. And then it handles all the implementation details. But there are real and serious concerns like tech fair. There's potential for technological warfare or the rise of sophisticated scams, deep fake voice and impersonations and impersonating people's banks. And these tools are getting incredibly convincing. And now you've got a Google V03. Have you seen a video from-- it's freaky. Oh god, it's an advanced AI video generation tool. It can create realistic videos with synchronized audio, including sound effects, ambient noise, and even dialogue. I think it's a death of Hollywood. Not only that, there's potential for misinformation. Imagine someone making an AI avatar of you, Chris. Let's call it AI Chris. And using that to scam your friends and family. My god. And AI Chris is generally frightening prospect. But I think we've got those capabilities now, particularly. Yeah. So I had a former colleague I bumped into recently. He told me about a cybersecurity conference he went to where he was called on his phone by a bot who was masquerading as someone from his internal company security. Wow. And they used his LinkedIn profile and all his public postings online. It's a paint a very, very convincing picture of him and the links to him and that this was genuine. And the idea was that they were going to try and gain access to his security codes. And I thought that's one thing we have to keep an eye on now is trying to work out how much of our own personal personas are online and how easy would it be for somebody to replicate? But how did he catch it, your friend, because I've been scammed twice. No, this has actually been done with his concern to the conference. So effectively, they were just showing how easy it would be to someone to use his online presence to try and scam him. Very interesting. And I suppose these large language models are already forming the foundation for even more advanced AI agents. I think Google DeepMind has been one of the companies at the forefront of this. And also DeepSeek, obviously now very, very famous from what happened earlier this year. But I'm not sure if they have full access to AI agents yet. So these agents aren't just generating text. They can also build software process payments and even create entire podcast without human intervention. I mean, which brings us back to Clarner CEO's recent comments and that Bloomberg article. After pushing for extreme cost cut in centering AI in customer service, he's now saying he went too far and is reinvesting in human support. It seems the big sticking point is customer satisfaction. Are we truly going to outsource so much of our data and our delicate customer interactions to AI? The future, as Clarner seems to be discovering, might be about combining AI's benefits, productivity, quality, and profitability with the irreplaceable advantages of human empathy, nuance, decision-making, and improvisation. OK, so augmented intelligence, maybe. Yeah, yeah, yeah. I think a mound of mentions we're giving them, I think, carnage, just bonds at this podcast. Yeah, here. Pay in full. Pay in slices. The question remains, how do we, as individuals and businesses, participate in this evolving value chain so we aren't left behind? And on a more philosophical note, what about robot rights in the future? Is that something else we need to consider? It's not just philosophical, it's ethical. What about the darker side? The potential for lethal autonomous weapons, acting as judge, jury, and executioner on the battlefield? These are critical conversations, and we need to be having now. Well, this is not new, but remember 2001? Oh, what happened? The computer to save the spaceship ends up killing the occupants that it's meant to serve. Our space honestly. Yes. I've seen that movie about five times in each time I fall asleep. I know it's really bad. I need to finish it. And I suppose for any of us that express concerns about these rapid advancements, are we just lulled it? Protesting about new machinery, fearing all the job losses, or is it an legitimate call for caution? And that we should be thinking about responsible AI development and some regulation? I think as far from Luddism, Chris, it's about critical questions, ensuring that as we build these incredibly powerful tools, we're also building in safeguards and thinking about the long-term human impact in the high, base, and low cases. So tell me, if you could create a super intelligence, what would you code it to do? For me, I'll code it to do my house chores. All the crap I don't want to do at home, basically. Oh, for me, it's going to be admin, 100%. On that note, we'll segue to the next one. Generative AI and robotics demand immense computational power and hardware, which means a staggering amount of electricity and energy. While some argue efficiency improves post-training, the initial demand is huge. So it's a power hungry beast, but where do these beasts live? In data centers, Chris, these facilities are incredibly energy-intense, requiring constant power and cooling. And we're seeing companies like Google's emissions increase because of this missing net zero goals. An NPR report even put it in a deeper perspective. One chat GPT query uses as much electricity as lighting a bulb for 20 minutes. So if you scale that up to millions of users, that's a huge amount of power. Not only is it a lot, it's an insane amount of power. And it's not just energy, it's water. More data centers mean more water to cool them. Bluembag highlighted how an AI chat book to summarise a document strains water resources. To borrow from one of our ed ops associates Gabby, the world is actually in fire and we're running out of water to put it out. Great quote. I know, I love it. And then there's projects like Stargate which is a 500 billion collaboration involved in Open AI and Oracle. This would only exacerbate this demand. And what's even more grim is that the Guardian also reported that the AI boom means that environment agencies have no idea how much water England will be short of in future decades. As data centres do not have to report how much they're using to call their service, the whole thing feels dystopian. So as well as my garden, data centres need more water. But what about the actual hardware? That's another big one. Producing specialised AI chips like GPUs is incredibly resource intensive. It requires extracting rare earth metals like lithium, cobble, nickel, often through environmentally destructive mining. And this manufacturing itself is energy intensive. Then you have the networking components, the tensor and central processing units, as well as the memory and the storage, adding robotics and it's. Oh, the output element as well. So you have constant upgrades leading to? Waste. So while AI offers incredible advancements, you can't ignore the tangible environmental costs of energy, water and hardware waste. Absolutely. It's the hidden cost of the digital age. We've unpacked a lot of things in this episode and I don't know about Eucharist, but I think existential environmental concerns aside. The direction of travel and where the AI game is at now, we have no choice but to participate in it or risk getting left behind. The race condition, rearing its head again. I look at my parents' generation and I almost envy their analog existence. OK, I think on that note we should move on to our lighter afternoon tea segment. So this episode's edition of afternoon tea has a bit of an AI infusion. I remember when chat GPT first launched, I used it to write a love poem. Of course, it denied loving me for a few too many times for comfort. Can I read the poem to Eucharist? Yeah, sure, great. This is for chat GPT. Not for you. So I wrote, write me a love poem and it said, my love for you is like a rose, deep and true, it forever grows. With each passing day my heart sings for the love that our future brings. Aww. You smile, ignite to flame within. My heart races, my soul to spin. I am yours forever bound and love sweetest, most precious sound. With you, my world is bright and new, my heart is for my love is true. I promise to hold you close to love and cherish you the most. In this life and the next, my love, I'll be yours, my angel from above, forever yours, my heart to keep in love's eternal endless sleep. Oh my god. That's intense, right? It feels like that's been written by a 14-year-old teenager. I don't know if I can write that up 14. It's a bit deep. It's a bit weird. For that was chat GPD, I think, not long after it launched. I suppose one of the fun things or the difficult things about AI is it actually has to use a huge volume of work and public-inverteible information and data on the internet to train its models to come out with masterpieces like your love poem. So experts are warning that the current rate of data consumption, there's going to be shorties of training data. So how could it train? Yes. And I suppose the problem with that also is a lot of the information that it's trying to train on is copyrighted. Yeah, it's on the internet, but there's copyright and there's royalties attached to it. So that's leading to a lot of lawsuits. So took us to one of the most interesting lawsuits? This was put to me by one of our legal analysts today and he was saying that there's a very high profile case going through the English high court at the moment between Gettie and a UK company called Stability AI. And it's accusing them of copyright and trademark infringement. So Gettie images, we all know, it has lots of images that are available under license. And apparently Gettie is saying that Stability AI has scraped millions of its own images without consent to train its stable diffusion model. And that actually is a direct competitor for Gettie's own AI generator model. And I think they're really sort of guilty as charged because some of the images that were shown in court still actually had the Gettie images to watermark on them. So not quite as original as everyone thinks? Yeah, exactly. And we are still talking about some large companies as well. You know, Reddit is suing Anthropic, which owns Claude for accusing it of illegally using data from its own site. And also Disney and Universal, so big content producers are also suing companies for copyright infringement. So they're both at the moment suing an AI term called mid-journey over this. Yeah. And in a much lighter way of using AI, there's a company called Hey Jen. So if you want to create like AI Chris or AI avatar of yourself, there's a company called Hey Jen and you can generate various types of videos, including anything. So I think they scan you send a picture or you scan your image and then you can do voice overs. It records a few voices of yourself. And then you can have it answer do Zoom calls so you can have your AI avatar doing Zoom presentations if you're busy. It's like turning yourself, right? You can be. Or you have double booked in a meeting like 4pm this afternoon. AI Chris can take the second. Yeah, AI Chris can just nod and say mm-hmm, mm-hmm, mm-hmm. Yeah. And then summarize everything. And then if you're too tired to create a presentation, you can use something called Gamma AI. I feel like this is an ad, but it's not. And you can create presentations easily in two seconds. You can also use icons, ad maker or even write and edit a book using type.ai. But how can you tell whether it's really AI or whether there's something else going on under the hood? A great example of that recently was the insolvency of builder AI based in the UK, which was funded by Microsoft, it was one of the largest sort of backers. And it had an AI app building service called Natasha. And apparently Natasha was able to pump out programs in record time. But what we didn't know was that this was 700 employees based in India rather than AI. And apparently this went on for eight years before anyone spotted this happening. Oh, wow. Yeah. I hope they were paying them well. They were very good to be generating things in minutes. Well, exactly. But I mean, there are situations where AI can be used for forces of evil as well. I think you were talking to me about the use of AI in Iran. Yeah, I think in Iran, before now, maybe they were using AI to track women not wearing hijabs. So AI face recognition software. And we can jump straight to my favourite, which was the 135 year old tortoise who became a dad at Zoomai Ami for the first time. Wow. And I liked it because I think that they tried to make him with other females, but it was only sweet pea who managed to solidify their love. And the sweet pea much younger than the. I think sweet pea, they said, is between 85 and 100. But do we need to adjust this? Because you know, you just up for dog ears. Do you have to adjust in the tortoises? No, this is human ears because the tortoise, an old Galapagos tortoise called. Something, which I can't remember. I think he was brought over to the states in 1885 or 1890. So, in human years, he's actually over a century old. Wow. And finally, the sports section. There's a first underwater rugby club, which is emerged or submerged in Putney, UK. What other sports fee do you think will be better off done underwater? No other sport. No other sport. No other sport. I just don't understand it. Why would you? But what I don't get rid of B is that one of the big things is when you score a try afterwards, you have to kick a conversion. So how would you do that underwater? I don't know. Get creative. A bit like Quidditch. Anyway, if you made it this far, you're definitely our kind of listener. We appreciate you sticking through with us about the talk of AI agents or autonomous agents. And even the question of whether we're all just lovedites for having concerns. Be sure to follow, rate and subscribe whether you get your podcasts, whether that Spotify Apple or maybe even via a generative AI-powered holographic projection. We're probably there. And if you liked what you heard, share it with a friend, a colleague or perhaps even your virtual assistant. You never know who needs a little insight into the AI apocalypse with a side of sober financial analysis. And thanks to the team behind this episode, Tanya Hubbard and Charlie Hall for producing the show and keeping everything running smoothly. Fowers and Daniel Whirl for handling the sound recording and all things audio. And a big thanks to Gemini for doing the first draft for the content for this episode. We'll see you next time. Until then, stay curious, keep an eye on those bond yields, and maybe go easy on the data sent to water usage.