Mailbag, incl: What moats are defensible in an AI world? March 29, 2026

26:47And it's frustrating to not have a firm view on it. But all it tells me is like, well, if you don't have any view on it, don't take a position because otherwise you're just investing on hope. So I don't have any zero shares. I'm not looking at buying any zero shares. Not because I think people who do it are silly. I personally, my very, very limited intellectual capacity just can't see far enough around that corner, you know. Because if, and you might strongly disagree, but if there is a stage where I just like save all of my receipts and whatever in a Google Drive and then point Gemini at it, you know, and say do the tax and it just gets done, it's like there is no world in which zero or certainly doesn't exist in a world where it's going to be able to charge the subscription fees that it charges.

27:30And so it's that kind of – I talked before on Friday and a lot about asymmetrical kind of outcomes, right? This seems like, and I'm picking on zero, but I think a lot of companies in this bucket where it's sort of like, well, heads, actually the market's overreacted. It'll more or less muddle ahead. And if that's the case, it's probably a, you know, a fair value, maybe even on a little bit south of that, some reasonable value. And maybe I can get 10, 15 % compound over the next 10, 20 years, which is great, which is great. Or I could lose everything. You know, it could be like Appen. and it's just like, whoa, tech darling to zero.

28:10Well, not zero, but close enough, right? And that to me isn't a bet that I will make without that strong conviction. If you are thinking, no, I am absolutely convinced that this will be around, then that's a different calculus. So it's a wishy-washy kind of answer, but I guess all I'm saying is don't make the mistake of going prices are down, therefore it's cheap. Cheaper, but maybe not cheap. Not cheap, yeah, I agree. What can I add to that? Um, very little, actually. I am less uncomfortable with Xero than you are, Ram. I think it's a pretty good choice. We've talked about that before. I'm not even strong in my opinion, by the way.

28:49No, it definitely is being disrupted. I just, I can't handicap it. Yeah, exactly.

28:56I think what's, yeah, look, I'm going to echo a little bit what you said, only just to say, Paul, ignore the past prices, which is impossible to do but yet we should do it right because it really doesn't matter where the price has been if each of those companies had doubled the last 12 months would you email said hey should i buy these ones because they're up it feels expensive right um and they've halved so they seem cheap now and i'm not coming out you paul in slightest what i mean is human nature is human nature and and by the way there are two different types of people there are there are the people who say the shares are up they get they're expensive another says shares are up they're going to keep going up and there's who those who say shares are down they're getting smashed and they say shares are down they're getting cheap and so it depends entirely on your frankly on your personality as a as an investor as an individual is how you see falling prices some people say i'm not going to catch a falling knife paul you're saying these are down they look cheap i like the businesses um so my only counsel is the impossible one which ignore where the price has been and back to ram's point about dcf you're just simply asking yourself if i buy the shares today what's the future likely to look like now it doesn't matter whether they've halved to a double between now and then.

30:00All that matters from here to the future, because this is the only price you get. You don't get to buy it yesterday. You don't get to buy it last week. You don't get to buy it last year. Maybe you did, but either way, you can't add to them now and you can't. And well, you can sell them at today's price too. You don't get to sell it at last week's price or last year's price. You have to say at this point, Templar website is doing this with this sort of business, this sort of margin, this sort of future available, this sort of price. Is that attractive or not? And so as much as it's impossible to do, you've got to desperately try to ignore whatever happened to the price in the past.

30:27The market doesn't care. The company doesn't care. Its customers don't care. Its marketers don't care. Its distributors don't care. All I care about is this company solving my needs and it's solving my needs better than it was last year and more people turning up saying, oh, this is cool. I want some of this. If that's the case, then you've got to grow in business and then you've got to ask yourself, how quickly is that happening? And does today's price seem attractive relative to what that future might look like? Of those, I mean, you mentioned some great companies here. Honestly, I'm nowhere near as strong on ProMedicus quality as you are around, not because it's not great, just because I think it's small and early.

30:59So I'm not prepared to thrust greatness on it yet. I may well in time, which is not, again, I'm not saying it's not great, just that I, you know, in some version of the future that the margins moderate and our competitor turns up and we go, remember when it looked like it was the only thing that was there? I'm not saying it was going to be, by the way. Same with zero. I could be wrong about zero for the same reasons. But I don't, but it's made every post a winner, right? You can't criticise ProBeticus. You really can't. It doesn't mean it's going to always be as good as it is now, but man, what do you want other than what it's doing?

31:25It's doing everything right. It's making every post a winner. It's giving itself and its shareholders the very best chance of winning. CSL, super great pedigree. Current growth, questionable, frankly. Innovation seems to have fallen over. They're not releasing many breakthrough drugs or treatments or diagnostic tools. What happens next for CSL? I don't know. Xero, I think the market's overreacted to but I could be entirely wrong. ARB I own so I'm biased. fallen a lot. I think I've talked about this before. I don't want to push my own company, but I will say quickly, on zero, AI is the risk. On CSL, innovation is the risk.

32:03On ProMedicus, valuation was the risk, at least in terms of the market's perspective. I say that kind of by, I haven't asked every individual investor, but you can kind of work it out in reverse. ARB, the local sales were rough because there weren't as many new cars sold that needed new bull bars. And if you believe that's permanent, then it was expensive. If you believe it's cyclical, which I think it probably is, then it might be an opportunity. That certainly was my view, but I might be wrong. Templar Webster sales up 20 % and the market just went, not enough. And again, to Ram's point, that may well be true because it's selling for a very large multiple earnings.

32:36So if it can't continue to grow up fast enough, then yeah, it wasn't worth that price. I think it probably is worth today's price, but anyone's guess. And Catapult, I don't have a strong view on. If I haven't looked at it enough, I like what it does. I don't have any, I'm not close enough to the company car sales one of the best businesses on the asx without doubt um network effect just to the moon uh it was stupid expensive it's getting less expensive now um i i would like to own car sales at the right price it's one of those companies that probably goes on forever um is there a chance ai beats it kinder but ai's got to find a database of cars for sale or a database of buyers who want to buy a car we're going to do that probably at car sales so while you chat gpt you might, and you might click through to a car sales ad maybe.

33:19And Matt, that can't hurt the business. It can. Until someone invests in an open source database of your car details and buyer details supported by advertising, and guess what? You've just created car sales. So I'm not saying it can't be disrupted, but I think the chances of it being disrupted are less than most of the others think. And you're right, Matt. These shares are down almost half. Zero, more than half. ProMedica is down by two-thirds. So I get why it looks like, ooh, there might be something here. I will take a half a step back, actually. I said don't look at past prices. I don't, with one exception.

33:49I do absolutely use it sometimes to look for stuff that might be attractive in terms of how do I make my investigation easy. If I find stuff that's down, I've got to go, okay, is it down for a good reason or not? If it's not, I look further. So I do use it. I just want to be really honest and clear. I absolutely do use it as a way of looking for things that might be interesting, but by no means at all in terms of my valuation work. It's only, hey, car sales down a lot. I might maybe have a look at it just in case it's worth buying is what I will do. But I won't buy because it's down. It's just an indication of maybe there's an opportunity there.

34:20I should have a look. Yeah, another thing, another way I've said this before, but I find it so central to the way I think about value that I'll repeat it. Hopefully it's helpful. I think too often we feel as though we have to do a fancy valuation and we have to have a valuation. Yeah, that's true. And I just think that, I mean, you do. I mean, you do, really. You've got to draw a line in the sand. Because at the end of the day, do I buy or do I not? Do I continue to hold or do I not? So you need that line in the sand. But I really spend a lot of time just doing scenario analysis. Like, all right, let's just see what happens if this happens, right?

35:00And so just when you were talking then, because it's really easy to do. It's not, I'll walk you through it, right? So I just picked up zero. They do about$12.33 in per share sales, right? So we'll just work on a per share basis. I don't know. Let's say they grow their sales at 20 % per year over the next five years. Incredible top line growth for a company of that size. Like, okay, cool. $30.70 in per share sales in the year 2031. Let's say at that point too that they've really got their costs under control. They've really scaled effectively and they're now generating a 20 % net margin. Net margin.

35:34Very few companies. Some companies are 50%, but they're rare as hen's teeth, right? 20 % is a very, very healthy, good margin and comparable to a lot of software business. Okay, they're doing that, right? So they're now at$6.13 on an earnings per share. And let's say that they're attracting at that point in time a PE multiple of 20, which feels a bit quaint in the modern era, but that's not a growth multiple, right? The average multiple is 16. It could be 30. It could be 50. And the math changes if it does. But I'm just, again, I'm scenario testing here. And you can test any kind of combination that you want.

36:07Well, if I do that and I multiply and divide and everything, all of that together, I get a share price in five years time of$122. Now, obviously, I want to get a return. So if I buy it at$122 and that's what it's still worth in five years time, it's not bad. So I discount that by 10 % per year for five years. I get, as it turned out, worked out really nicely, the exact current share price. So in other words, what am I saying here? Don't go, oh, Andrew said that they're going to grow at 20 % and they're going to get this money. No, no, no, no, no. I'm saying if, if all of that happens and I buy it today, I will get a 10 % return.

36:44And you go, is that good enough? How confident am I on that? Now, there'll be someone out there listening going, 20 %? These guys are going at 30%, you know? 20 % margin? No, no, no, no. They're going to get 40 % margin. Oh, the market will happily pay 50 times PE for that. Do it. And then try a bearish and then try a base case and then try a whole variety. You just get this cone sort of going out from the present, this range of different outcomes. And you don't know what history will, you know, at that point in time it'll be history, what it will deliver. But if there's only like 10 % of scenarios that work out in your favour, then it's sort of like, ah, the odds aren't good.

37:27Unless I've somehow got an incredible amount of conviction that these favorable scenarios are the ones that are going to eventuate. And it's so useful as an exercise because you just, you get a very, very good feel. And that's really all it is, a vibe almost. So like what needs to go right? It's why I always push back when people say, but it's a great company and it's going to continue to grow. It's like, cool, let's have much. Oh, they easily grow at 20%. Okay, what kind of margin? Oh, they would easily do it. Well, take your own advice and plug it into a spreadsheet. All it does is tell you that this is an okay investment.

38:06And that's fine. I'm not saying 10 % per year is a great return. But it's not what you wouldn't expect from an ETF with a fraction of the risk. So there's also the opportunity cost here. So it's like if I'm going to go specific on an actual individual company and the best I can do on a probability adjusted basis is get the market returns. Like, why bother? I'd rather just buy the ETF and never have to read an annual report ever again and basically get the same return anyway. I need something that's much better than that to justify the risk into doing it. So just experiment and just play around with things.

38:47And it will give you no certainty as to the future. but it will show you what the future needs to look like if you're going to get a good return. What's the combination of those things? And this isn't advanced stuff. It's just like sales, margins, and multiples is all you're asked to do, right? How much can they sell? How much can they keep? How much will the market pay for that? And what kind of return do you want off the back of that? And it very elegantly falls into place, at least in a way that I know if it is going to work out some combination of those variables has to be affected, right?

39:24And in the way that leads to that, because there is no way, like, there is no way that I'm going to get a good return if the top line growth is anything less than that kind of thing, you know, unless somehow they're on insane, unheard of margins for that kind of business, or the market just loses its collective mind and decides to give it a PE of 100, even though it's never really sustainably done that for any company over any length of time. It's just, you know, you want, Buffett talks a lot about, you know, you don't want to jump over six foot bars so you're looking for a one foot bar to step over.

39:58And they're rare, but you'll come across some of these. Some of my best investments are that it's just sort of like, I can't read the future any better than anyone else, but it's like, if this company just does that, right? If they can, now all bets are off if they collapse in a massive financial fraud tomorrow. But, you know, more or less, if they just muddle their head at this kind of rate and they just attract an average multiple and this and that. And it's like, ah, things look really good, right? And it's a nice way to sort of at least clarify your thinking and not tell you what will happen but what needs to happen.

40:37Oh, shut up. No, nicely put. I don't think I can add more to that other than just the reverse DCF idea, which is you mentioned different scenarios. The other way is you work backwards and say what does the current price imply. So there's a million ways of doing it. The tool is the same. It is your friend. I will say just quickly, then throw it away, or at least conceptually throw it away. Don't stick to the absolutes that you get from this stuff, as Ram's kind of pointed out. Change your mind. It's so easy to go, hey, I got this tool and I got an answer, therefore that's the price. Remember that the future is inherently impossible to predict and all estimates of value are completely rough and guesswork.

41:10So a lot of people, Ram and I have been in the same situation. You discover the tool. So as you should do this, you go, oh, man, I know exactly what Woolies is worth now. It's$32.85. And you've already said, might use different scenarios, but just be careful with that as you pick it up and start using it. Again, can I just also tell you, you will find that changing your mind is really, really important. And I'll throw myself under the bus because there's too many options to choose from in terms of making a fool of myself. But I'll go with a relatively recent one. I was quite fond of a company called Ava Risk Group.

41:43You know, they do a bunch of stuff. And I think my, I'm just looking now, my average price is probably somewhere around 20 cents or something. Anyway, I bit the bullet at about 10 cents and sold out. And I did that because I just revisited my assumptions. And it's like they might have, you could argue otherwise, they might have held water at one point in time. But when the world changes and the facts change, it just didn't, right? And so I sold it. I lost 50 % of my investment in that. And it's actually, I'm really proud of that trade. because they're now five cents. In other words, you can be down 99 % and still lose 100 % of your remaining capital, right?

42:23Like it feels as though, well, how much further can I go? We can go to zero. It can drop from wherever value it is, it can drop 100 % from here, right? And I always find it's the people who, no one in this world has an investment career where they just never make any mistakes. The difference between those that survive and go on to prosper is just that when they make a mistake, they recognize it and move on. It's painful at the time, but in retrospect, you'll look back and go, I made the right decision, right? Yes, there will be other times you sell at 10 and then go, God, it's 50 cents, what was I doing?

42:55But you've got to be not rigid in your thinking. And as facts change, move around with all of that stuff. It's amazing how much schadenfreude there is in this industry, particularly if you've got some kind of public profile, you said this, but you said that. And it's like, yeah, I changed my mind. Like, I don't know what, I don't know what, this is not the dunk that you think it is. Other than you thought that I was an all knowing, omnipotent soothsayer. Like, well, thank you that you thought that, but I wasn't and I'm not. And so all I can do at any point in time is make the best judgment with the information that's available.

43:34And if that information changes, my conviction changes, then I change. And I don't lose a second sleep on any of that kind of stuff. And it's just that money that I sold, I put into something else that's gone incredibly well, right? And it's like, yes, in hindsight, I should never have touched it and I should have put it all into the good thing right at the beginning. But this is what I just see so many people wrong foot themselves. They start out with the best intentions. They have a couple of fails and they're not really fails. They just they actually have a couple of eventualities that are inevitable.

44:03And they go, oh, it's too hard. This is risky. Rather than, oh, this was an absolute inevitability that this was going to happen will continue to happen. That's not the problem. The problem is, is my reaction to this and the way I prepare for this kind of stuff and the way that I maintain a conviction when warranted and a flexibility when not is that's the superpower of investing. Not that you never make a mistake. And it's a very, very hard thing to say to someone that, listen, put a significant part of your money into something is entirely probabilistic in nature. And you probably like not 10 % of the time, like 40, 50, maybe even 60 % of the time make a mistake.

44:47And it will still work out in your favor. Like that's the other thing people don't see when things go well for you. It's like, oh, are you lucky so-and-so? I was like, yeah, well, let me direct your attention over here. And let me look, let me highlight this for you. It's just, it's counterintuitive, I guess, which is why I sort of flogged the horse a little bit on it. Because that is the path that every single listener is on here. Even if you are incredibly successful, that's the kind of stuff that's going to happen. And then I'll just shut up just by saying, if ever you get to a point where you can't even begin to put numbers at it, well, that automatically tells you what to do, which is not that.

45:25Because you're literally just flipping a coin. Probably worse than flipping a coin, right? You're rolling a 20-sided dice where only the number four will save you. You know? Correct, correct. That's it. Yeah, it's very difficult. It's highly, highly overrun. Oh, yeah. Motley Fool Money. For more, subscribe to the free newsletter at fool.com.au forward slash listener. Third question for Paul. What are your thoughts on the current anti-AI sentiment? And that the very reason for the AI boom is also the demise of potentially so many. Do any businesses stick out to you as having strong economic moats protecting against AI?

46:02What do you feel may be the strongest economic moat in a rapidly advancing AI world? Those are three very good questions. Let's stick with the middle and the last one, mate. What business does stick out having strong moats? And what is the strongest moat to have in an AI world? I've said it before, and I'm increasingly convinced of this. I think it's, and you touched on it just before, I think it's access to proprietary data. You can have the best AI model in the world, but it needs something to work on. I'll shout out, I own quite a few Catapult shares. I have for a long, long time. First bought a 50 cents.

46:38to make myself look smart, but then I have suffered probably up to my fifth or sixth 50 % drawdown on that company. And again, I don't say that to brag, but to highlight that's what success looks like. It's losing half your money multiple times in a row and still being grateful for the experience, which is a little bit hard to wrap your head around. Anyway, I'm not trying to sell anyone on it other than to say it's a nice one because they own the data. You know, it's sort of like I could, I mean, the componentry in the hardware is not hard. You can order it online from China and put it together.

47:15It's not sophisticated stuff. Right. You know, and frankly, and there's a lot of IP that's gone into the algorithm stuff. It's sort of like, yeah, very impressive stuff, but I don't see any barrier there. So me and Scott, we employ our AI super genius and we come up with our own algorithms. We find a contract manufacturer. We put some gear together. you know, we call it trebuchet and we go out there and we are much more superior to Catapult. So for those, anyone who's wondering, the trebuchet is much better. And we put our trebuchet product out there and it's just like, we're going to really struggle because of a network effects, because there's just more and more teams that are using it.

47:55And also because our heuristics, our algos aren't going to work very well because we just don't have a very large data set. So companies like that, who each year that goes by, they get more and data, more data means more, more insight that the model uncovers, the better the insights, the better the advantages, you know, which means more people will use it, more people use it. Guess what? You've got more data, which means that you get better output. And just these, these self-reinforcing feedback loops are super, super, super, super powerful. And I think I've said all of this now and it'll collapse in a heap tomorrow.

48:30But I think that to me is different as opposed to, I can't think of an example now, but something that's like very cool. Well, Canva, you mentioned, I won't mention Xero, but Canva, right, is a potential thing. Like they do really cool graphics. You can make posters and slides and all this stuff really cool. I'm not convinced the moat is as strong there when I can just say to my model, do this, and it just does it. what advantage do you have by having a million uses or 50 million uses or whatever it is that makes sense i you know it does i think um what are my thoughts i think that's right i think i mentioned before there are some businesses where proprietary data isn't enough to combat um uh open source data uh i've given the example of there's a mob in the US providing loan credit score type data and where the combination of all that is better for all the banks that use it rather than actually use zone data because it simply has less data than the rest of the pool.

49:38So I think my only asterisk I'll put on yours is where proprietary data is more relevant than open source or otherwise available data, you're 100 % right, I would just be mindful of in that case. It just makes sense, right? The bank's like, well, hang on, I know this much about the market. They know much more than that because they've got all this data from all the banks, I'm better off using their data than mine, even though mine's proprietary and there's open source, I can do a better job of running my own business using that open data or the publicly available data rather than my own. So I'm not saying you're wrong, I just that would be the only asterisk I added to that one.

50:08I'm going to take a different angle to yours, mate, which is, you've kind of said, how do you win AI? I'm going to say, how do you win in an AI world, which doesn't mean you have to necessarily win AI to win. And I would go back, Paul, not surprisingly, my favourite competitive advantage is brand, always has been, probably always will be. And that is, again, not going to work everywhere because the AI advantage might simply overwhelm brand. And so this is one of those all else being equal questions you're asking, right? Which is if everything else is equal, what would you do? A smart business will use AI for their own purposes.

50:41And it's like saying who's going to win in the internet world, right? Everybody wins in the internet world. My strongest belief, and I could be speaking of being wrong, my strongest belief is AI will end up like the internet where the value actually accretes to the consumer or the user. rather than the companies using it and it becomes a tool to be more efficient, more effective, more tailored, more all of the things the internet does for us already that mail order slash physical shopping and physical interactions doesn't do, right? Ram and I could have a town hall meeting every Sunday or get a corner in Hyde Park in Sydney and shout at each other across, you know, on some stumps or some upturned milk crates or we can use this.

51:16And so I suspect I accrues to the end consumer at the value of it. I absolutely believe if you don't use AI well, you will be bypassed by your customer. So AI adoption is a really important thing for me and the companies that I'm looking at to make sure they're at least keeping up. Not in a press release kind of way because everyone's using AI because it's the buzz phrase. Like blockchain was the buzz phrase years ago. So, you know, for me, if you're competing, if you're trying to win AI, then to Ram's point, the proprietary data matters a lot. if you're trying to win in a world that has AI as part of the toolkit you've got, including everything else that you have, like marketing and logistics and, you know, programming and R &D, whatever else you're doing, to my mind, if AI becomes an enabler, then I actually wouldn't focus more on AI than you need to unless your company is in an AI space.

52:10If you're a software, like the SaaSpocalypse we've just talked about, yeah, that's a massive concern. You've got to be across that stuff. If I'm Woolies, am I going to get beaten by AI? No. It might have been someone else using AI. Yeah, you bet. So I've got to do at least as well as the other guy. But if it's an enabler for everybody and it doesn't actually confirm meaningful competitive advantages to anybody, then to my mind, the answer is the things that actually matter now will matter then, other than those specifics. And the example is simple. Blockbuster and Netflix, Blockbuster absolutely didn't get with the program on the internet and got smashed.

52:40So I'm not saying it doesn't matter or you can't do it or you shouldn't do it. If Blockbuster had done video or had done streaming with Netflix did, maybe we're watching Blockbuster rather than Netflix right now, right? Oh, we should be. Right? Exactly. Absolutely, we should be watching Blockbuster. They didn't get beaten by the internet. They got beaten by not, and you've used the example a million times with the Chinese, is Chinese or Japanese? And gunpowder, Japanese. Japanese, yeah. Chinese. Who's going to win the gunpowder world? Well, nobody, because it's equal playing field unless you choose to eschew gunpowder altogether, in which case you're screwed.

53:11So, again, I'm saying Ram's right about the use of AI specifically. There are competitive advantages you can accrete from them. I don't want to dismiss that at all. But separate to that, do Woolies or Coles win the AI war against each other? Probably not. It's probably just faster, cheaper, less stock, faster stock turns, more targeted advertising, all the things that AI can let them do. Who's going to win? The one that does the best job of serving the customers. Not the one that uses AI best in the sense that meaningfully best. Maybe there's a margin point in it or something because I have got this data and do that with it.

53:42The other guy catches up really fast. I don't think there are most, again, the internet matters. So Amazon doesn't exist without it. Google is entirely internet business. I mean, Amazon could have been a physical bookstore, I guess, in some version of Jeff Bezos' head. Google can't exist without it unless they're talking about search cards at a library, so it's not going to happen, right? So there are companies that will absolutely win because they're AI. For the rest of them, I suspect it's the usual stuff. And so I am not desperate looking for AI competitive advantage. I'm looking to make sure my companies aren't going to get beaten by AI, and that's what we just had the conversation about.

54:12like Catapult and Prometicus and Templar-Webster and Xero. Yeah, that's my take. So the strongest moat in a rapidly advancing AI world is still brand as long as you're not in an industry where the adoption of AI can literally and seismically change the market you're in. Yeah. Just in general in terms, because this is the latest thing, I think one observation I've had with all of the things, whatever the latest thing is, whether it's IoT or edge computing or whatever, you know, they come along every so often, is that it's hard because there is an element of truth in all of it. I think the world is right to be excited slash terrified of AI.

54:59Oh, yeah. It is. It's a thing, right? Like it's more than like a little demo project that kind of does some interesting stuff. Now, there's very much real. That was harder to see five years ago, by the way. But now it's sort of we're at the point to deny that this is a very important layer in business and life going forward. I think you have to be you have to be a little bit mad. At the same time, you always get people who just like it's not unreasonable. It's sort of like it does everything and it's this and it's got in a box, essentially, you know, And there is a lot of distance between a really capable agentic model and super intelligent, AGI super intelligence.

55:47They're of a different caliber. And one changes everything and might even be existential. And the other one's like, oh, it's a cool little tool that we can do things better than we ever did before. That's a big gap. So you've got to, I think part of it is not be the, I see two types of investors, broadly speaking. Ah, humbug, I used it and I made a mistake. It's all rubbish. It's like, well, you're an idiot. No. You're an idiot. It's like, you need to do a bit more research. And then I get the other side of it, which is, it's everything. You know, we'll all be like floating on, you know, floaties in our pools and we solve scarcity.

56:25And it's like, wow. You know, the truth is in between those kinds of two layers. And I think what this is, the more recent thought I've had on AI is that what has been, I said this the last week or the week before, it's like, I really encourage people to play around with it just because that will teach you more than anything blog posts that you're going to read. And having really played way down the rabbit hole, really played with it a lot recently. I, in the early days of AI, I just thought it was going to be, there is a world in which it's Anthropic or it's OpenAI or it's Google. It's one of these ones that just, they are this monolith.

57:03And now it's like, I actually think the world of AI is going to be far more modular. And you're seeing this now in how I can build an agent or an agentic system, or I can incorporate AI into my workflow or add it into my website. I'm pointing to the same model that you're pointing to. But mine just works a lot better because what people don't realize is that there's a lot of traditional coding that acts as a shell around the AI that really distinguishes it from absolutely useless, never gets it right, so just works seamlessly and like magic. Without going too far down this path, just stop me when I start.

57:42Not making any sense, but the thing you've got to remember with what I have learned with AI is that they call them inference engines and they are stateless in the computer term in the sense that they're these mega geniuses that have zero memory. So anytime you send a prompt to an AI, it basically looks at probability weightings between words and then it spits it out, but it has no recollection of what you just did. Now, it doesn't feel like that when you're using ChatGPTX. No, it remembers what I said. What's going on behind the scenes there is that there's some really clever traditional programming that is essentially a prompt builder.

58:21So you're sending your prompt, but behind the scenes is actually a, there's a bunch of, it's called context that you add to it, right? So user's name is Scott. Here is the last 50 messages in our conversation. Here is what I have in my long-term memory. Here is the tools that I have available to me. And when it works really well, I know because I'm on version three of my own age. You're getting to a point where I got to start again. because I've learned enough to know that I was going in an unviable direction. I mean, and really, I guess what I'm saying is here is that there is, I think there is a world where absolutely there's the anthropics and open AIs of the world.

59:03I actually think a lot of the value will accrue to the people that take those incredible inference engines and apply them to their particular problem in a very novel and unique way. OpenAI is not going to be working on an accounting package. They'll offer the model and then someone will put that into their system. If anyone's been following the whole OpenClaw or Maltbook or these kinds of things here, why that's been such an explosion, which I'm sorry to go down this part because it's a whole other rabbit warren to go down, but people have discovered that they can just get an Anthropik subscription and then they can build the shell themselves in a way that's highly personal to them, that runs on their computer, that has a tool set that's applicable to the things that they want to do in the way that they want to do it.

59:54And I've tried it. I've tried OpenClaw and I just, I can't get it to work as well as what you see other people doing it on the internet. There's the people out there, it's like, how did you get it to do that? It's literally the same software. I'm pointing at the same model. And it makes a lot of sense when you think about the internet, right? Because we've all had the internet. Jeff Bezos built Amazon, right? And then, you know, and Blockbuster laughed at it. So the technology is there and it's available for everyone, but it is not going to be, to my mind, it's going to be, okay, great. We've got a new, humanity has a new tool.

1:00:30Who's going to use it the most effectively? And I don't think that's, that's what's going to emerge out of that. And I suspect that some of these big, large language model developers will be very, don't get me wrong, extremely successful, but I think they will be more the commodity side of things. It might be different if there was only just one monolith that was just Anthropics is miles ahead of everyone else and that's what everyone's going to use. But there's a pretty, you know, between sort of even XAI, you know, Grok and all of the, you know, and the deep seeks of the world and that, they're all in the same field as each other.

1:01:02So I don't know they have the pricing power that people suspect. but there will be people who just take those things and build on top of them and integrate them into things and that will be the secret sauce, you know, and that will be the things, even though I can ostensibly do that, it's like how did you get that to do it in this particular way such that it had that magical experience for the user where it's like I have no idea what's going on under the hood but it works seamlessly. And that, again, is the lesson of the internet, right? Like it is the same protocols that we're using today as we were using 20, 30 years ago But the experience is radically, radically different because people just took that existing tool set and that tech layer and they applied it in different ways.

1:01:46So this is something you want to very much to the earlier point, have some firm views, but very loosely held views on and adapt and change. But it's not nothing, I guess, is where I'm going with all of that. And it's not as obvious as some people make it. and I think reality will end up being far more interesting and far more nuanced than a lot of the talking heads would make you believe. Sorry, mate, that was a long, long spiel. No, it's a really good point. Yeah, I don't have anything to add. I think we're pretty good. Paul's got a fourth question we're going to ask so I can make this a Paul episode.

1:02:23He's only going to get to one question really, isn't it? Yeah, one email, Paul the Paul podcast. This is a very specific one, mate, but it's just an interesting way to talk about an individual company. What are your thoughts on Credit Corp? At the time of writing, a 5.6 % fully frank dividend yield, 8 % grossed up, a PE of nine lows, I should mean share price lows, approaching the COVID bust. What am I missing that my trusty Gemini research agent, speaking of AI, cannot tell me? Are investors worried management are simply seeking to buy it off more than they can chew with expansion plans? And with a strong Australian dollar, sorry, sort of slowing ANZ market, any lack of growth results would likely further penalise the business.

1:03:03I don't follow it closely enough. No, I have a first go. I followed debt recovery for a while. Collection House was the only recommendation of ours at ShareAdvisor. You got some of that. Back the wrong pony there, didn't you? Credit Corp pile. Mate, Credit Corp, if you look at the long term, again, not wish it was about share price necessarily, but man, Credit Corp had a time of it share price-wise. It's been all over the place. It's the one company, I owned this a million years ago, and it's one company I've used an example people say what mistakes did you make and this is one I've told this before I'm not for a while and I don't know if I've ever said that it was the company I think I might have basically I bought it on someone else's tip and then the price went down so I sold it because I didn't know what I was doing and of course it went back up again so I made I made a buy without doing my research then selling without doing my research because I realised I'd bought without doing my research and got burnt both ways so this was the classic lesson of understand what you're buying as you say mate know what you own and why you own it so that was my story on Credit Corp having followed it for a while i'm not going to give you an absolute answer paul um but i will say by the way just quickly this this company uh 10 years ago was ten dollars and 12 cents it went to 35 dollars that was 2016 35 dollars in 2020 uh it was down at 13 dollars by march 2020 COVID bust, back to$34 in 2022, now to$9.82 in 2026.

1:04:27I've actually got nowhere in a year. There's probably some dividends there, but share price-wise, dead flat despite a massive range of ups and downs all over the joint. The challenge here, Paul, I'm not going to give you a view on Credit Corp as an investment, but I will explain my thinking and how I would think about it, and you can see if it makes any sense for you. The company basically buys other people's uncollected debts. So your Telstra, I don't know if Telstra's a provider for Credit Corp, but use an example, and they've tried to get Ram for six months to pay his phone bill and he just has completely gone off the grid, hasn't answered the phone calls, refused to pay it back.

1:05:00And he owes a hundred bucks. And Telstra's like, well, I'm never going to get this money back. And Credit Corp say, ah, Telstra, we'd like a chance to collect that, please. And Telstra goes, well, I've tried for six months and I haven't got any money back, so it's dead to me. Andrew's bills, I've written it off, I'm never going to collect it. And Credit Corp goes, we give you two bucks for that bill? And Telstra says, well, it's worth a hundred to me, but I know I'm going to get nothing back. So yeah, I'll take you$2 because it's money for jam for me. You're taking the risk off my hands. I'm getting some money for it.

1:05:27Thanks very much. And Credit Corp says, great. Well, I collected that for two bucks. Andrew owes me a hundred. Now, Andrew might pay me back or he might not, but John and Jane and Fred and Sally also owe me a hundred bucks each. And I bought all their debt ledgers for$2 each. If only one of them pays me back even half of what they owe me, I'm going to make money on all of the debts I bought, even though I'm not going to collect most of them. And that's exactly what they do. They pay an absolute pittance for a whole collection of due debt. And they see how much they can collect. As long as they collect in aggregate, more than they paid the utility or the telco or whatever for the debt, they're ahead.

1:06:01That's how they make money. And that's the entire debt collection business, right? And that's what they do. And you've got a couple of things going on. The first is they have to keep buying debt at attractive prices. What's an attractive price? Well, I use an example of$2 that you might pay for Ram's$100 debt. If it goes to$5 per$100 debt, is it still going to be profitable for them? Well, now you're kind of not sure because if you collect only one every 10 and you only get half of what you actually owe, you're going to break even and you've paid people to do that. So you're losing money. So now you're saying, well, I don't want to pay five bucks.

1:06:31And they say, well, the other guy's going to pay five bucks. So what do you do? You either pay up and try and collect more or you don't do it and you lose your sales and revenue because you've got nothing to collect. Now, mathematically, it should be an easy decision, but public companies aren't always that clever and the market loves certainty, right? loves regular income. Or you buy it for two bucks thinking, well, of course I can collect it. Then the economy goes to crap. And so the money you thought you'd collect, even Andrew can't find the 50 bucks to pay back half his debt. And so no one pays anything back.

1:06:58And so you lose money even though you paid$2 for$100. You collect absolutely nothing or almost nothing. And so really what your, the credit corps entire business is betting you can pay a price, incur some costs and collect more revenue than those two things combined. And that's kind of all it is. really straightforward. The problem you don't know is how much they're going to pay now, how much they're going to have to pay in future, how their collections are going to change in future. We just don't know any of these things. And so that is the bet you're making. And that's why even though the business looks cheap, and I have no particular view on it value-wise, it looks cheap, but you've got to believe that the future will look like the past or the recent past to know it's that cheap.

1:07:39Because again, PAs look backwards, right? So by the way, higher interest rates right now, are more people going to... Now, on one hand, more people might default, so Telstra's got more debt to sell. And Credit Corp might make a fortune because they buy lots of debt cheaper and they collect lots of it and they make more money. Or they buy the debt and their collections team can't get blood out of a stone and so they're losing even more money. And that's... I don't mean to say that necessarily equal risks. I don't know what the risks are. I haven't handicapped it. I haven't looked at Credit Corp for a while.

1:08:05But that's why it's really, really, really hard. You don't know what debt they're buying. You don't know how much they're going to pay for it. You don't know how likely they're going to collect it cost is going to be incurred in that collection. And so it's a big, you mentioned before, Ram, sometimes you don't know. For me, I'm like, I don't know. I have no view. I know it's cyclical. I've just told you about the share price all over the joint over the last 10 years. So, you know, over that 10 years, they've, earnings have been as high as$1.34 a share, as low as 25 cents a share. Two years ago, 42 cents a share.

1:08:33Last year,$1.37 a share. Now, nine times earnings looks cheap if they deliver, they earn$1.37 a share this year. If they're back to 42 cents a share, it's all of a sudden 27 times earnings. That looks really expensive. Somewhere in between, maybe. But Paul, that's why it's a really, really difficult business to try and value on any ongoing basis. It's a business, well, you're right. But at the same time, it should, the lumpiness you can't do anything about because there's either a lot of ledgers that are available for sale at good rates and, you know, or there isn't. And they've always got to be building that loan book because that's visibility on future earnings.

1:09:10You just continue to run down the assets that you've got while there's nothing else left to collect on. So you've got to always be buying. And sometimes it's just better offers in the market. But I'll speak on their behalf, not to defend them, but their argument would be, we've been doing this for a long time. 100%. We've got a real... Now, how do I work out the odds of Scott paying his bill back? No idea. How do I work out the odds of a million Scots? Ah, that's very... Now, this is where you kind of like actuaries can like be very, very accurate with mortality rates. So it's sort of like predicting what one person will do is impossible.

1:09:47Predicting what 10 million people will do is actually really easy. And in theory, they should actually do extremely well because what they have as an incumbent and one who's been doing it for a while is very good systems and processes and experience. the real art here and it is an art is to be able to look at a debt ledger and go i know that we can at least collect this much no matter what happens and as long as that is more than what we can collect for it we're going to make money and we're going to make a good return on that and as an investor what you really have to do is is ensure that you've got a very you've got a you've got a business that's very good at collecting debts, but more importantly, very good at pricing debt.

1:10:35Now, Collection House and others in this space who have gotten into trouble, how did they get into trouble? They bought a bunch of ledgers at prices that were too high and they weren't able to collect on them. And it's the kind of business that if you're not careful, you can go bust very quickly because all this money goes out the door and then nothing comes back in, which means you're either having to raise capital or get some debt in to make you whole again and recover. And it's very risky. But when done right, it should be a very, very stable business. I mean, arguably, speaking of AI, maybe that should make things a little bit easier.

1:11:12I don't know. We've been able to look at very large data sets and talk about proprietary data. We've been collecting debts in this country and overseas for a long time. We know what people look like. We know how they act. We So, yeah, it's the maths is interesting and it's worth pursuing. But where I got to on it was I know what you should do, but you do get to a point where it's like it's actually there's a degree of art in all of this. Who's the team that's behind all of this that's actually making the purchase decisions and are they being sensible? Because don't forget, you can do a lot of things where actually we can buy a bunch of distressed debt because we hit a recession tomorrow and everything looks super cheap.

1:11:55So we buy a huge amount. We've got this massive pipeline of future revenues that are going to come in the door, except we didn't account for the fact that our collections are also going to drop or something. And it just, it can look really good. And then all of a sudden, things just change really radically on you. So it's just one of those, it's not a bad business by any stretch of the imagination. It's just a business that is very hard to measure the quality of, Because it is really down to the judgment of the individuals involved in the systems that they've got. That's just hard to sort of address from the outside.

1:12:32One thing just to throw in here as well on the ethical investing side of things, because it's kind of that kind of company that people get very thingy about. I don't want to tell anyone how they should think of their own ethics personally. But I actually think it's not as unethical as it sounds. It's like your debt collector. they're not sending Tony around with a hammer to kneecap you right like they're not doing that kind of stuff and the thing that you've got to remember is that when people don't have access people who are desperate don't have access to credit they often do these things and they get into very very serious trouble so it's kind of one of those things it's like it's as ethical as it can be and it's it's there's an unsavory quality to it but it's far more savory than than the alternative it's like cash converters right payday lending is is a very horrible business in a lot of way shape or form and you know people will talk about credit uh predatory sort of um interest rates and the rest of it i mean i don't have strong views on this other than just to inject a little bit of nuance into it it's sort of like what people fail to realize is that if you're not charging that high rate of interest you're not doing anything because you you build into your model like it's not like the banks could charge 20 and they choose not to right Right?

1:13:47You have to do it because you know that a lot of those debts aren't going to be repaid. And so after you account for the losses and that, it's like you're still sort of – the profitability is still, you know, reasonably marginal. It's not to excuse it and it's much more complicated than that. But really just to inject that into it before you do make – and reach whatever conclusions you want. And it is more complicated than I'm making out. But it isn't as black and white. Sometimes I do bristle a little bit because it's really presented as this black and white kind of thing where it's like, well, actually it's a little bit more complicated than that.

1:14:21Fair. I think that's fair. They also have Wallet Wizard as a brand where they collect$141 in costs over nine fortnights on a$1 ,500 loan, which is an annual interest rate of some stupid number. But yes, you're right. You've got capital over multiple times throughout the year, so on an annualised basis, it's incredible. But why aren't they in the biggest glass tower in the middle of the city? because a whole bunch of people don't pay them back, right? And so, you know, it's sort of like you don't have to like it and I don't like it. I think, frankly, that you just don't offer credit in those kinds of ways, I would argue, and in a much broader sense as well as a separate conversation.

1:15:00But for them to exist as a service, that's just kind of the shape of it. It has to be. And what you tend to observe, and you actually see in a lot of other countries where they don't have these kinds of things is what you do is you have the mafia is doing it. People will always take credit that is available to them, you know, and if it looks too good to be true, it's because they're, you know, they're doing an IMF on you. They're giving you a loan knowing that you're not going to pay it back so they can extract all your resources. Oh, totally. And look, it's a societal problem. It's easy to blame those providing the credit.

1:15:39If people who need the credit are not getting the credit, then you have to resort to that. someone's probably going to make the line or people are going to go hungry or whatever and it's messy all the way around but it's too easy just to blame the credit provider in that circumstance. It's complicated. It's all I'm saying. It's icky and it's horrible and I've got to be careful because it feels like I'm trying to defend. I'm not. I'm just saying it's not as black and white as it seems. No, you're right. No, you're right. I'm agreeing with you in that sense. I find it really distasteful but also if people feel like they need that or do actually need it then it's hard to be super critical unfortunately for a better voice.

1:16:11Yeah, we've got to be careful with that too. with gambling and that kind of stuff as well, which I think a lot of that should be handled a lot better. But, I mean, it's just like prohibition. Alcohol's bad. Yeah, it is. Let's ban it. How did that work out? I mean, and how does it always work out, right? You know, human need and desire is a thing, and it's just like that's one of those things that's just sort of you can say about markets and people is like wherever there is a demand, it will be satisfied. It will be satisfied, right? How long have drugs been illegal for? You think there's no drugs out there?

1:16:47It's just sort of like prostitution was the other great one as well. And the best thing that ever happened for the industry was for it to be decriminalised. And again, it's like, oh, so you're a no. No, I think it's horribly exploitive. I think it's bad in every other possible way, shape or form. I just don't think certain blanket bans actually just, A, the problem doesn't go away and it makes the damage far worse. So it's kind of like, do we want a perfect solution or do we want a better solution? And I'm going to shut up now because I'm going to sound like a real piece of work here that I'm in support of all of these.

1:17:20So I'm not, I'm not. I just, I think we are too myopic and childish in how we think we can address these kinds of issues. And on that note, for the tomatoes at Andrew Page, I'm the good guy. He's the one supporting kneecapping and extortionate. No, I'm kidding. Mate, that was fun. Paul, you're welcome. Everyone else, there are plenty more questions in the hopper, but hit us up if you want to have your question answered at info at fool.com.au. We will try and get to your question in due course. We've got a few, but we can always do with more. Always some good stuff to talk about. A great opportunity for us to have more of a chat.

1:18:02I will say, Ram, just quickly, Paul finishes with PS. Although you may think no one is listening, I'm always on the pod in the darker depths of the episode as we approach the one and a half to two hour mark. Thank you and Fool on. There you go. You're welcome. Nice one. Fool on. Cheers. The Motley Fool and people appearing in this program may have positions in the companies mentioned. General advice only. Please speak to your financial professional to understand how it may pertain to your situation. Subscribe to the free newsletter at fool.com.au forward slash listener. The Motley Fool operates under Financial Services License 400691.