Ep 276- Prospero.ai CEO George Kailas

Hey, it's the Bintek newscast. My name is John and with me, as always, in perfectly good health is Steve. Welcome to the podcast. You don't have to jinx it, John. I'm good for now, but now I feel like I'm gonna get something weird. Nothing will ever happen to you, Steve. Don't worry. Good. Yeah, yeah. And the reason I say that is because the job numbers are all about health care. It seems like the medical health industry is just taking over the economy. I'm not sure that's a good thing when you see other wealthy countries. I've seen the graph that essentially plots how premiums from Obamacare have almost doubled since 2022 or so, but I'm curious what is driving this growth in healthcare spend? High executive bonuses. Oh, great. Yeah. And the aging of the country, we're getting a higher and higher percentage of the population is over 60. And along with, I guess, robots to help around the house and the medical industry is too bad. We don't know anything about investing to take advantage of that. Do you, Steve? I certainly don't. Give him my track record. No, I know. Zero. Well, fortunately, we do have someone on this week that knows a lot more about investing than we do. We're lucky to have George K list, the CEO and co-founder of Prospero.ai. Thank you for having me out. Excited to do this. Yeah. Any thoughts on the healthcare industry taking over the economy or any ways to profit from that? Yeah. I mean, I think it's interesting. And if it's adding jobs, I don't really care. I think we need to add jobs, however we can. Right now, it's definitely one of the things that I'm more concerned about structurally. There's a pretty scary graph where you see when Chatsy PT was introduced in job openings in the S&P and there's a split where traditionally job openings in the S&P are aligned and the same upwards or downwards motion. So, yeah, anywhere that's adding jobs, I'm very happy about right now. But my frequent advice to retail investors is really get engaged in things you can odds make well. And what I don't like about a healthcare sector right now is how much it moves based on something that's largely unresolved in terms of these premiums, things around Medicaid. So, my advice would be unless, what I always say, are value investors, unless you work in that field and you feel that you have some good angle on figuring out something like what's going on with United Health and the extent that maybe the storm has passed in terms of how much it could impact revenues, how much that's baked in already. A company like that has been really beat up, and there's potentially a lot of value there, but it's value I wouldn't really touch because I don't know how to evaluate it because I don't know how much more downside there is if there's negative news coming out of the government. So, yeah, my advice is stay away unless you work in that field or you feel you have some way to understand how to value that potential discount upside risk reward properly, then by all means go for it. I just don't make a habit of odds making what the government's going to do. That is so such good advice. And actually the only way I've invested is when I feel like I followed some area very closely or at least feel like I know the dynamics behind it to know that this will happen or eventually happen and invest based on that. So, it's funny you bring that up right away. But I'm curious, you know, so it kind of take the contrarian view on this. It's unlikely that you know many industries. So, how do you then overcome that gap between understanding an industry from the inside and actually being able to invest in companies? Oh, I mean, I think there's things that I understand very well. I think I understand tech very well. I think I understand consumer cyclical very well. I think I understand communications. Those are probably like the three that a lot of our good picks and we're beating the S&P significantly in the last four years since we've been doing the picks leading into that and understanding those sectors well. I mean, I think that's a big, I think that's a thing that a lot of people miss when they're trying to say like, oh, I need to understand the whole market. But like, even me, you know, I worked at my first set from when I was 17. I taught myself a county to get that job. No value investing very well. Doesn't mean I think I can research every sector, you know, sufficiently versus understand like, I'd say three very well. And then I'm opportunistic in a wider range of sectors. So, you know, you mentioned the term, their term twice already. And I think that for folks who don't know what exactly is value investing. Yeah. So, so value investing is looking at, you know, either DPS, a lot of value investors like the free cash flow that the company is producing, you know, which is a little harder to manipulate for some of like accounting practices and say, you know, if you're taking a depreciation expense, you know, that's something that could hit EPS. But it's going to hit differently in terms of the cash that's being produced. So a lot of value investors like free cash flow or cash flow better for that reason. But basically value investing is saying, you know, throw out all the noise, all like this short term price, you know, what's happening on even a day to day week to week month to month basis with price and say that eventually these ratios of like earnings per share or, you know, free cash per share. Those are what's going to determine your long term gain and you might have to wait a year or two to get that full value. But it's going to work out and, you know, that's what Warren Buffett especially has made, you know, famous. But one thing that I always caution people and I think the market is moving away from value. But I, you know, I always try to use it as a tool in my toolkit is, you know, one of the things that value investors, again, kind of to the theme of what can you do and whatnot. A lot of the times value investors especially Warren Buffett is helping to run and guide these companies. So that's why I think especially as a retail investor that's just kind of like a passerby, especially if you're buying in after you see that Buffett has taken a big stake and, you know, that stock has already gone up, it might be a little more challenging especially in this day and age of algorithm and trading for you to get, you know, quote unquote value from value. But one of the stories I always really like to kind of get people between, you know, the balance of kind of like stringently looking at value versus like the evolving nature of value in the market is, you know, Jeff Bezos will kind of say one of his biggest, we're sorry, Buffett will kind of say one of his biggest misses came out of a meeting with Jeff Bezos where it was earlier on in Amazon and Jeff Bezos basically offered him a discount to the current price and Buffett's response to that at that time when I learned that you're investing, there was a rule they said you don't invest in technology because you don't know how much of the future growth is already built into the price. So there was a meeting Buffett was offered a discount to the current share price of Amazon and, you know, Buffett was basically like I can't pay these growth rates like it's not even that, you know, you may or may not hit them and, you know, of course Bezos beat those growth rates, but he said it's not within my methodology to be able to, you know, forecast this pay these rates that's just not what I do. So there's always this, I'd say this, this kind of dual action that I'm looking for in my value investing background. One is it's a nice way to interpret things to make sure my losses are minimized and I'm missing more, my missing less things. But there's this other side where you also have to say like be prepared if you have conviction and this kind of goes back to what you were saying in tech. Like if I have conviction on a product a way, you know, whether it be like Palantir or Tesla, these things that are way overvalued with earnings because how people perceive the potential of those products. And I've done very well with Buff over the years timed right. That is one thing that I will, you know, kind of toss value at the window and say I don't think value really applies here. But the big message that I always tell the people is, you know, learn all of these analysis frameworks tools in your toolkit and be able to understand when you want to apply them, why you want to apply them and maybe when you want to underweight them, Chip. Is it more difficult to do value investing now that essentially there's rice and earnings per share evaluations seem to be largely divorced from economic reality, especially with more tech heavy stocks? I think it's definitely more, I think it's definitely more challenging to do pure value investing. But, you know, I think there are always value opportunities. I think a really, really good example is, you know, one of our best long-term case studies is met. And not only did we kind of one of our, you know, but one of our best most reliable long-term signals, not only, you know, the funny story is it was actually, that 100 out of 100 in that score for months before I finally said, "Hey, Meta's a good stop to add," because I thought that I knew Mark Zuckerberg better than my software designed to be objective. But after I saw it stay there for a couple months, I was like, "Okay, maybe I am missing something and add it," and I was very early to that. Part of what reinforced that story is I looked at Meta was a fraction of the PE, of even closer comps like Microsoft or Google or certainly bigger ones like Amazon Tesla, the rest of them accept. And that is how I reinforced it. I did look at our software that really focuses a lot on the options markets, I would say, long-term options markets. And I also said, "Well, the value story is here too." And like that's, you know, when you talk about tools in a toolkit, I think that's the really important thing, because even though I saw one of our most important signals there, love Meta, love it for a bit before I jumped on board, I still, because my relationship and trusting Mark Zuckerberg at that time, I still kind of need a tiebreaker and value as the tiebreaker. We're speaking now in the middle of February in about a week and a quarter, so there's a massive sell-off in tech. And my thesis is that this sort of, this seems like it's kind of an overreaction by the market and I'm actually quite bullish on the tech sector going forward. And I'm curious, does that ultimately present more opportunities for you and the users of your app or do you think this is a bit too uncertain to make any, to call any shots? So I'm not sure, so I agree with you that there's some good room to the upside in a lot of these stocks that are big, earning stocks, where some people are a little afraid of AI. And I would say just looking at these companies, they're earnings relative to where they've been and some of the growth that I believe has gotten hammered a little too hard. But I'm not, we're more short than long right now because our index level QQQ and SPUI and an option sentiment designed to look at supply and demand in the short-term options markets. Those are very bearish on both the indices. We're topping out of it. We actually just saw the Dow get to the all-time high and kind of swing past SPUI and QQ. And that's, that is also a kind of topping signal outside of it. And then I think we have some structural issues. That is, that is, that's explaining I think why we're seeing a lot of hedging in the options market. I do think the uncertainty around the shutdown that caught up with people, I think there's been some political instability. I think there's some geopolitical instability, both in the tensions with Europe rising a little bit, some talk of those tariffs. And I think, you know, there's uncertainty around the Fed and interest rates. And I think we kind of saw a reaction to that today. So that's, you know, I'm always looking at different angles and I will agree to you. You just look at the companies and the setups and how they trade it in even the last, you know, six months, year or whatever. There is potential there. But I think there's, I think there's structural problems with the market and the way the institutions are playing the options markets that scare me way too much to look for those opportunities. So what we do in that kind of situation is I still have some high growth longs. You know, a few of our favorite ASTS, L-E-U, SNDK. So we'll hold those, but we'll actually hold a greater number of shorts on the other side to kind of balance those concepts that we're talking. So how did this lead into creating Prospero? So yeah, I mean, that's not an easy question. I think to do. Yeah. So, I mean, I started, I started in finance when I was fairly young. I worked a lot on the buy side. And, you know, the big thing for me was even working at small funds. I felt disadvantaged versus larger funds on, you know, information. There was a situation where one of my jobs, my boss was the largest shareholder in a few companies, and he made a nice 10x return as a result of that lawsuit. And me buying shares alongside him, even within the same company with different knowledge pieces. You know, I lost, you know, 92% of what I invested. Well, he made a very nice 10x return. And that's just how it can go with these different, you know, knowledge services. So, you know, when I got into AI 15 years ago, started building some of my own algorithms. So, you know, really, I would say use value investing principle, use good logic systems, and match them with what I call intelligent scale to surface underlying market equations. And that was really the goal when we started Prospero. It was actually a much more modest goal than these excellent, you know, trading investing signals that people can use consistently to find alpha. You know, what we actually set out to do was just solve this very simple problem of basically what I went back to when I lost all that, you know, lost all those, you know, funds myself saying, oh, like I think there's things signatures, especially in the options markets that can alert people to these things in very different ways than even, you know, a value investing or other good, you know, quality of approaches or even, you know, a lot of quantitative approaches that are focused on technicals. That kind of thing is not going to come through in the same way it comes through in the options markets, especially the short-term options market. You know, we got to work and really what we wanted to do was basically keep people out of the way of institutions by servicing a lot of these, you know, different equations. We have 10 different signals. You know, our most successful highest value ones are probably looking at the options markets, you know, things like dark pool, you know, percentage trade of the dark pool. We control for churn a little bit with that, but we have 10 total signals that we think actually do well in painting a complete picture of what's happening with a stock without over complicating that situation. And, you know, obviously what we invented was significantly better than that and it's turned into being able to get ahead of the market because not only are we keeping people out of the way of institutions, we are picking up, you know, where they're betting, like, one of my favorite case studies is twice in 2022, 2023. We saw, we added Alibaba and then dropped it and added it again before major news events. So we didn't know that these news events were coming. One of them was a, was a, a new financing for Ants. Another one was, at Financial, another one was a restructuring of Alibaba, but twice within three months we saw the same signatures of short and long-term options markets. And so we knew about a news event before it happened and that's where, you know, a lot of this, you know, significant returns that we've seen because sometimes we see things in the signals that are about to happen before they happen. Actually, that's not infrequent. There was another really fun example where we, we have a momentum score that combines some of our signals and we saw a record low momentum score the day before Wolf speed announced a, like, had a bankruptcy announcement. We didn't know, but it made up 17% of our investing portfolio and the short side, like, overall as a short and 11% of our trading portfolio because we just saw these, these signals flashing red. So we like to think it's our good deed for trying to help people that we invented these powerful, you know, trading market signals again for the market without really intending to do that. So, again, it's like that focus on paying attention to a specific company, then, you know, looking at the, trying to read the whole market, that's the difference. So what kind of tools does Prospero have to help you with that focus? So we have, so we have these 10 different signals, right? I can just kind of like go through them and I'll do a breeze by as of them and you can ask me more questions if you like. Market similarity, that's like a simplified beta where we're just basically simplifying a concept of like, okay, if you get close to 100 market similarity, you know, you're probably taking on too much risk because you're basically getting the market and then all the single stock risk. And we have a dashboard of like important items you should be paying attention to. Yeah, 10 signals, five short term, five long term educational resources around them. And then we have two different newsletters and investing letter, a trading letter that just shows how to use them in different strategies. You know, we write them, you know, we teach some lessons, some macro lessons along with that being like, here's the macro situation we're looking at this week. Here's how we adjust our portfolio based on that, but it's always just using those 10 signals. I understand that you've basically released four versions of the product I think three or four. And I'm curious, you've, how do you talk about things like new? new features for the product and how do you handle things like quality control and how has the product evolved again in your multiple year journey? So yeah, one of the things I would say creates, that creates such a positive experience for our customers is the quality control. I was the only person that used, you know, we're six years old, but about the first three years, I was the only one that used the app, the signals, and I actually made myself not read much new, so financial news is none of it. And the thinking there was that until I could do really well with it, then no one else would be able to do well. So that's our quality control there, and then typically, you know, we will put, you know, we just released a trade, we just released an alerts product. And again, that was about a two year testing period where it's not good enough that we just see some good back testing and see some good results. We've got to look at it live and make sure that it does well before, you know, we put it in front of anyone else, and that was one of the most frequently requested, as people talked about, requested signals. I mean, requested products, so we have a rule we generally build what people ask us to build. You know, obviously we're doing some, you know, effort versus value calculations for the roadmap on R&D2, but generally, especially if a lot of people are asking for it, that's what we prioritize. But, you know, we also like to think of ourselves as, you know, making sure that the product we're building is really the product that people want. And like my favorite story about that is, you know, one of our most requested features is building price targets and people love price targets. And there's a couple of reasons I don't love price targets, but we still put it through this test. The reason I don't like price targets is I think it makes people lazy and they don't actually learn how to make the exit decisions, which are the most important decisions, and they could be sitting there with a 20% gain, and they're just waiting for it to hit that 30% price target. And that's like a laziness that I don't like to encourage people to use, because what I encourage people to say is you really have to know when to exit. A lot of people can pick stocks that go up, especially in an overall bullish market. The market goes up a lot more than it goes down. What's going to make you better is figuring out when you want to get out of that. But that being, so I always had a problem with them, but that being said, I was like, okay, maybe we'll build these price targets. I certainly know how to build price targets for people. That's not an issue. But before I do, I want to know what people would expect to be a win rate on a buy recommendation. And when we surveyed people, you know, 60% like most of them said 60% or more. And more people said 70% than 50%. And so like for people that don't know this, there's a saying on Wall Street, you're right 51% of the time, you can start a hedge fund. So people wanted this product that couldn't exist, right? So actually that was one of the reasons that we, that actually piece of data had us pivot a little bit and had me, you know, start this newsletter. And the newsletter does get close to 60% win rates, but it gets close to 60% win rates on a population that is curating that's high conviction. So I basically said, let me back into what they're asking for, but I can't give them the specific product that they're asking for because what they want can't be built. Is the time frame that a lot of people want just too short? You have to kind of like, it's part of the education that this will take longer than you might think. Everyone wants a quick turn around, but you might be looking at an event. You don't know when it'll happen. And if we could be six months, it could be two years. And it'll still be a win. So your win rate for a six month thing is going to be much shorter than your win rate for longer term. Yeah, absolutely. And that's why we use and use the newsletter as two, right? Because we're saying this is our example portfolio. You can see it work. We're not guaranteeing those results for you. Like, yeah, if you follow our portfolio, we do expect you to see that. But, you know, it's not necessarily that anyone uses the Prospero signals. We can't guarantee that you'll use the web, right? And that's an important distinction. But one of the things like some of my best power users, and there's some great, like if anybody's wondering how to like use it, one of the best ways is going to our crowd fund. There's a lot of reviews and everybody from, you know, law, you know, people that are new, that are becoming investors to experience traders, they can all see back. So one of the things, and it speaks to your point, I do a lot of one on one education session with people. And by the way, if you want one of those, you can just email me at George at Prospero and AI. And in some times, those sessions, I will tell people that, you know, they should just invest in indices because they don't have, you know, the time where I think the desire to educate themselves, that it will take to get better or get good. And that's really what we promised people that if they use the system, they will improve. That's like our promise. And I've never had someone say that they didn't improve when they keep telling this to Steve all the time. I just don't listen. I just don't listen. Yeah. And you have to make realistic promises like that, right? And it's kind of funny. Some of my best power users, I've ended the call. And I've said, look, I hope you find value in Prospero. I hope you decide to make a bigger time commitment. But the way you're talking about this, I really just have to overwhelmingly recommend that you invest in an index. You know, just put your money and bring our total return, maybe, you know, 12 dividend and call it a day until you're ready to put more time in. And then you know, I've had more than a few people, you know, email me a week later. And I was like, you know what? I was thinking about it. I do want to put the time in. So sometimes that's the best thing you can tell someone that they have unrealistic expectation of how this is going to go for them. You have 100 million data points and 10,000 models. Why have so many models and so many data points when a, when maybe a singular model would cover this? Or am I looking at it from the wrong perspective? Well, singular models are always terrible at it. It's really funny. And like I will do the diligence on them. Because I've had probably been in AI for 15 years. I've had probably at least five times over the years. Someone insists that, you know, the auto encoder can do well. And you know, you put an auto encoder into, you know, to an RNN and it'll work well. Like we do have a specific type of analysis that uses that, but we structure it a lot. We structure both the inputs and the outputs a bit before we feed into that. Because you just feed in all the data we have. They almost always perform terribly. They almost always over fit severely. And it's very easy to see why they don't work because they will, you know, basically they'll train at, you know, a 70% accuracy. And then they'll test at like, you know, 55 and then the moment you take them out of sample, they'll be lower. Or you know, sometimes they'll be a little better than that. But you'll run them for a little bit like we say. We run a lot of this stuff live and then they'll just catastrophically miss some turn in the market and have no idea because, you know, they don't really, these models work well. And you know, the way our system's built and it's been, you know, my first company did AI service for eight years. We had some really cool evolving neural network technology. We had our own reinforcement learning system that we kind of invented before that was a thing. But a lot of those models, a lot of the experiences had the same problem and, you know, help, you know, Prospero works now and it's a very good system of both testing ideas and then figuring out and integrate them. And essentially we have these foundational models. You know, finally, I went back to coding the foundational models myself actually when I was in San Francisco during the pandemic to stay up in sanity. I would intro to Python through deep learning because I think that there was too much in that work between my knowledge and the formats. So we built those and basically we have a lot of linear models and scales at the core. So something like net options that the most powerful metric. There will be information on skew, you know, the extent calls above where this orchestrating are trading above puts below in the short term options markets will look at, you know, sheer dollars, you know, open interest of, you know, contract calls above or puts below. And then we'll look at, you know, some volume volume changes, distributions. And so we have probably 20 different of these linear formulas in those categories. And you know that we've integrated because they've tested well. It didn't start as that many, but they basically can graduate into that system based on what I'm telling you where they have to test, train and predict their own accuracy to a good enough confidence that we can take them, you know, take them into the system. And then we let AI gather and integrate more information and test, you know, variations of those linear formulas. as well as play around with their weights. And an interesting example of how it works really well for us in that framework is originally the formula isn't that any volume at all. It was just pricing in open interest calculations. And over time, they've gotten kind of more and more awaiting attention because they've proven it works. But on a given day, you might have like two to five percent of the total net option sentiment as volume. But what we've actually let it kind of graduate into is near options expiration dates, we've seen that go as high and test well enough to go as high as 15%. So basically, that's how we balance those things in the system and expand the tech. Anything can get access to a small weight that it has to earn a bigger and bigger weight based on not only how well it does, but how well it interacts with the other linear formulas and experiments. I want to go back to something that you mentioned. You mentioned that you basically coded with your softwite by using a free tool. You learn Python on your own while being here in NSF, right? Yes, yeah, I took classes. Wow. And you found that that was enough of a foundation for you to be able to code this massively, you know, as you say, 10,000 AI models by just using a free course? Yeah. It's been nine months probably spending like, you know, five, six hours a day. Wow. Not if you're not if you're Steve, but if you're a George K list, yes. It takes me 10 years. Yeah. Yeah. Interesting. You also like this wasn't production code. Let's be clear. Like I handed it off to my CTO. Like the ideas were there and he made them run. Like it's just like a theory. Yeah, it's just like like a PV. You also have what you call a 10 signal framework. And sort of time back to my other question, I'm curious. How do you resist the temptation? First of all, can you explain that? And secondly, how do you not add more, more frameworks to this or more, more signals to your framework? It seems like once you're on that path, my inclination would be to just, you know, keep adding stuff until it's fairly comprehensive. But I'm curious to see how you think about it. So yeah, we're always looking at new products. Like I mentioned that like that auto encoder to RN and product is called market profit. And we're getting closer to releasing that. And that will actually be in its own sandbox because you know, the 10 signal framework, we were trying to solve a very specific problem that we don't want to deviate from, which is what number of signals do we need to we feel explain stock analysis in a complete way where nobody needs any other information. Obviously, that's not the optimal, like we talk about using other frameworks. But can someone come and up to the app and just look at those 10 signals and understand everything that they need to know to make a quick decision. If it's going to be quick anyway. And we were trying to, we were trying to fit between two distinct polls that I think, you know, you look at Yahoo Finance. And most people don't really know how to use that. There's a ton of information, but it's kind of information overlaward overload. And again, that the utility of that skews more to experience investors. Then, you know, there's these very specialty products that have things like dark pool prints or options sweep orders. And, you know, I think that's, we're even a price target. And I feel it's an oversimplification of the problem. So we're very intentionally trying to sit somewhere in between those two things. And, you know, the way we look at improvement, as I mentioned, is on the back end. Right? We look at, you know, good examples, like we have profitability in growth. And it metrics zero to 100. And you compare those to say, the Yahoo Finance, you can look at a lot of different things that surround the concepts of profitability in growth. Right? But a lot of people have to tell something has like $1.59 EPS. They have no idea what to do with that information. But if you tell people that something has like a 70 profitability rating, that's a lot easier for them to understand, integrate, compare, and then as we've evolved that, you know, before it was like it started as really just a forward EPS on a scale. But we've added a lot more economic simulations to the back end of that, you know, some sector-specific analysis views that, you know, are technology in terms of projection of those sectors or market cap levels in terms of, you know, what we kind of see for those kind of pairings of sector and market cap. And all of that is behind the same number. What we say is we're trying to keep things simple, yet make them more information rich on the back end. So a lot of our development, people don't know how much additional work we put in, because it's all, they're dealing with the same simple, simplified 10 signals. But on the back end, we're constantly improving them and also improving them through that framework that I mentioned, where all of our ideas that we code get a chance to earn parts of the formula, parts of the signal. But they, by no means all actually end up earning that. And then, you know, obviously, we're always looking at new products. A trade alerts product is a really good example of something that, you know, it's a different way for people to consume our ideas, the signals, like one of the things that does, as opposed to just our pure signals, is that is looking for confluence with generated AI models. And when it produces an alert, it's taking like a grouping of our signals that we think are the most promising, especially the moving averages. And then seeing what generated AI has to say about like, a list of our best, you know, kind of momentum score. Tickers, and then just telling us which one to think the best and the marriage of those two or becomes an alert. - Awesome. And that's essentially the way in which you're currently looking to integrate Genai into your product, essentially finding a way to make those alerts more useful for your end users. - Yes, yeah. And we heard a lot of what people said in terms of like, hey, like Prospero is great, especially by learning it, but can you make it more simple for us? And that was the response to that, where we said, okay, we're just gonna send you one alert a day. And, you know, we're gonna add Genai to our, we are looking at other ways to add generated AI, but that's the primary way we're integrating into the experience right now, yes. - Awesome. I understand that you had, you know, you mentioned in a newsletter called, I think Don Trim the Hedges, that you essentially kind of predicted a downturn. And I'm curious, you know, looking under the hood, what systems did you, or what signals, rather did you see that sort of foretold an incoming downturn in the market? And how far in advance did you actually see that before you put up the newsletter? - So that one we love, because that's the one we kind of rewarding for a long time. In advance, in every letter, we look at our S, P, Y, and Q, Q, Q, and adoption sentiment. And, you know, there's a bit of a different flavor to the relationship always. Like this year was kind of interesting. We had a lot of hedging, so close to, you know, under 10 net-option sentiment in S, P, Y, while Q, Q, Q was in the 40s or above. And some people that, you know, that had seen other, you know, that had seen S, P, Y net-option sentiment, especially be so predictive in the path. And they're like, well, why is it zero? I was like, well, people are obviously hedging in the S and P, while they're being more aggressive in tech. So we can always kind of see that relationship when they're both low. Like there was a great example of, before there was a big dip in the market 10.9. And we actually sent out a big warning on 10.8, because we actually saw Q, Q, and adoption sentiment go from 30 the day before to zero the next day. And that was a huge move. So it's really just those two signals. But we've gotten out of in front of a lot of market like flips, which just those two signals. And that's really like where, again, we're balancing simplicity and complexity. All we're ever doing is looking at like our favorite and least favorite stocks. And then saying like, okay, to what extent are we going to go more long or short based on these index level values and trends? - Just to be the future, that's only one to know. - We can't always, but we can sometimes. - Yeah, yeah, to some degree, yeah. Yeah. So this is very interesting for the investment side. And for our FinTech startups and founders out there, there's a lot of tools and a lot of ways to invest. When you were starting out, how did you get those, I like to ask, 'cause I think it's a mystery to people who don't start companies. How do you get those first and first customers? And you're not like an enterprise thing. You need a lot of people to use the product. How do you get that word out to get that scale that you need to keep growing? Especially at the very beginning. - Yeah, I think you always, I mean, I think you either, well, you need a lot of these things, but I would break it down like this. I think you need a very good plan. first, and then you need kind of like a go to market around that plan, right? We had a plan of, you know, a lot of quality control, a lot of value and patience and, you know, most of the users that have come into our system, it's less true after the crowd fund, but I would say before the crowd fund, we actually had, I would say, 60% of our users come through three influencers. And when we were in V3 of the app, you know, I reached out to a lot of influencers and I said, "Hey, you know, been following you on, we did X Twitter, been following you. I think our tool will help you." But also, like I love experience in finance, like if you don't like the tool yet, but I can add value in terms of how you analyze things and we can figure out how to get that in the product more. You know, let's start that conversation. And, you know, obviously, with anything, didn't have like a particularly high hit rate, but developed relationships with these, you know, three influencers, especially. And they became power users. And it was very simple. They did a very simple journey of, as I asked them to, integrating Prospero into their process, showing where their ideas came from. And, you know, that thing kind of sold itself. I think, you know, one of the things that people always ask us about our business model, and I think a lot of people just fundamentally don't understand why we would do it this way, but to me, it's the only way that makes sense. A lot of people say, like, you know, why aren't you an asset manager? All right. And the simple response is, you know, wealth front betterment paid about $1,000. A user and customer acquisition cost. And I understood those analyses. You know, people in terms of like our influencer strategy, you know, we paid less than a dollar per user. And that's gone up a little bit as we've grown, but still very low compared to those kind of numbers. And it's because when we become an asset manager later, we think building the loyalty, letting people see the signals, kind of how the sauce that he's made, you know, building that trust and then saying, hey, we can convert this user base to asset manager customers who we think will also be more loyal because they understand and have a history with us and our signals. Like, we actually think that's the correct way to build an asset management business. But that was always the plan. You know, we had like a five, ten year plan from when we started Prospero. And you know, I will say it always takes a little longer than you think. But, you know, I think having that plan to be willing to execute not being, you know, we had a pivot on the product. You always have to be willing to pivot on the product, build things that you didn't anticipate. I think that's table stakes. But I think having a go to market plan around that that you can stick to and really budget around. I think that's where a lot of people go wrong. They're spending too much in their go to market and they don't really understand, you know, why or how that fits in with their plan. I think that's something that we did very well. Yeah, I just have some common sense. That's a good thing. And also the secret is to have three good influencers. That's a secret. If you're a FinTech founder or in a startup out there, you've just heard it here. So I won't ask you if the about the AI bubble. I'll spare you that. Sure, you hear this all the time. In lieu of always trying to play the market, doing your homework and focusing and having realistic expectations and using Prospero.ai. Is that a fair assessment? You've listened well. Yeah. Yeah, yeah. So some really good advice for our listeners and FinTech founders. We wish you all the luck. Even though I know this is not a luck thing, we still wish it for you anyway and keep up the good work. Thanks for having me on. It's been great. Yeah, thank you. That's George K list, the CEO of Prospero.ai. Please hit subscribe to keep up with the latest in FinTech news and thank you for listening.