Credit Crunch: Where AI Fits With Cognitive Credit’s Rob Slater

Welcome to FICFocus, where Bloomberg Intelligence fixed income credit currency and commodity strategists and analysts discuss their short and long-term views on debt markets and issues. Now here's the Bloomberg Intelligence FIC research team. Good tidings, dear listeners, and welcome to the latest edition of Coretac Crunch. Part of the FICFocus podcast series where we focus on all things credit. I'm your host, Noel Hebert, Chief US Credit Strategist of Bloomberg Intelligence. That's the research arm of Bloomberg LP today's guest. Rob Slater, Rob is CEO and founder of Cognitive Credit. Cognitive Credit is an analytics firm. They're focused on streamlining credit analysis, including everything from data extraction and model creation. With an eye on expansion, we're going to talk to Rob about the founder's journey, the challenges in succeeding where many others have failed. And what's next? As someone who's long been oriented, that being me, to trying to broaden the intersection of technology solutions and credit, this is going to be a great one. I think everybody's going to find a very useful, Rob. Welcome to Coretac Crunch. Thanks to all really happy to be here and looking forward to the conversation. So as I kind of did in the intro there, I mean, there are a ton of areas that I think are really interesting here and obviously your company's here at an interesting time in terms of the intersection of technology and credit. But before we get into all of that good stuff, kind of curious just in terms of your background a little bit in terms of what life looked like before cognitive credit and then sort of what led you to one day go, you know what, I'm going to put this shingle up and we're going to try and do something. Yeah, happy to. Now I told you before we did this, I'm not one who loves talking about himself. But what I would say is if I'm in your listener's shoes and I listen to other podcasts and stuff, it's always very interesting for me to hear, what was that person's journey? How did they end up where they were to do what they did? So I happy to give you a little bit of background on my career. I started in corporate credit in New York at the turn of the century right around the time of kind of the telecom crisis, right? So early 2000s. So at that point, we were a couple years after the dot com bubble had burst, but there were some follow on troubles in the economy. And so it was a little bit of a quiet period on Wall Street before things then really got busy in the middle part of that decade. My first role was working on what we call at the time, the frequent issue is desk at city group. So this was coming out of the Solomon Brothers period that they were the number one team on the street issuing bonds for the most active finance oriented companies. So G capital for motor credit businesses whose business was finance. And we were issuing bonds all day every day. And that's a job where we did a lot of work on rel valve, curve analysis, deal flow, but where every basis point really mattered. So it was a great introduction to Wall Street, to credit, to financial markets. So I did that for a couple of years and then moved to a different part of city group, the leverage finance business, really because I wanted to move a little bit away from market analysis and learn more about corporate analysis. How do you analyze companies? And so after a year of working in New York, I moved to Hong Kong and I worked in our leverage loan portfolio team, really doing corporate credit analysis on our loan exposure in Asia. And I did that for a couple of years. And that took us to the start of the global financial crisis and implosion of subprime and all the bank balance, she trouble and whatnot. And I had an opportunity to move to Europe. And I joined our bankruptcy and restructuring on workout teams and was part of the group. If you remember at the time, a lot of the big banks set up bad bank arms to work through all of their bad assets. So kind of the special credit recovery team. And did that for a couple of years. And perhaps of everything I've done in my career, that was the best job. It was a period of tremendous learning, a lot of responsibility because there was just so much work to go around. And we got to do a lot of things all across Europe. So for me, culturally and just learning, meeting different people, learning about all these different European countries, it was great. After that work was done for a couple of years, I moved to our distressed credit trading business in Europe. And that coincided with the European sovereign debt crisis. So you can see now, of course, this is the theme of the past 25 years in markets and working in finance, but we're kind of bouncing from one crisis to another. And I was very involved in representing our business in Europe, working through both sovereign restructuring and also bank recapitalizations in southern Europe. So I did that for a couple of years. And then my colleague and I got hired by a very prominent family, really behind the growth of capital markets in general in the US over the past couple of decades, to join their family office. And we ran a pool of capital to help them diversify some of their assets internationally. And we really ran that like a hedge fund and did that for a number of years with a pretty broad mandate. So across asset class, where can we find the most attractive risk reward opportunities outside of the US? And that was really where I spent a lot of time trying to think about the intersection of technology and credit, really kind of like this necessity is the mother of invention. We were trying to do a lot with a pretty small team and wanted to find how we could do that as well as possible. At the time, and this kind of gets me to cognitive credit. So I hope I'm not going on too much here. But at the time one year, I was at the Goldman Sachs Quant Conference and the head of equities for GS at the time gave a presentation. And I'm not going to give the exact numbers because I won't remember them. And I don't want to misrepresent anything. But the crux of his presentation was so this would have been like 2015 plus or minus about 10 years ago. The crux of his presentation was 10 years ago, this is what the GS equities business looked like from a headcount staffing perspective. And this is what it looks like today. And it was a unbelievably dramatic decrease. It wasn't 25% less people. It wasn't 50% less people. It was something like 80 to 90% less people. I mean, it was dramatic. And then he got into all of the innovation, all of the automation, how that business had been transformed through technology. And myself sitting in the audience, first and foremost, wearing my credit investor hat, it was just very drawing for me because we experienced none of that in the past decade, right? And the thought experiment was if you had a time machine and you went back 10 years before and you told the equity team what was about to happen with their business, it would have been dramatically transformative. And if you told the credit guys, nothing is going to happen. Nothing is going to change. It's going to work exactly the same way. For me, that was a real point of inflection, just thinking about what is the future of credit look like. And around the same time to wrap up this story, one of my friends working in Silicon Valley sent me one of Jeff Hinton's papers on convolutional neural networks and the implication of kind of what cutting edge looked like for image recognition at the time. And putting this all together, I got really motivated in a zero interest rate environment in Europe to take a little bit of time out of the market and think through, was there something bigger that I might want to be spending time on? And so I did exactly that and spent a year really kind of going down the rabbit hole of machine learning, thinking, what are the implications for an ever more digital credit market? Was there tooling that could be built that would not just benefit an individual firm, but the market as a whole and decided to team up with some engineers and start building software. And that was really the genesis of cognitive credit. Fascinating journey. Fascinating decision, I suppose, the zero interest rates probably facilitated somewhat there, as you mentioned. But, you know, it strikes me because it's interesting, right? Because it's not, I feel like, you know, over my almost 30 years in the market, right? Maybe not that first decade, but certainly in the last two decades, there's definitely been aiming, people aiming to sort of find some sort of technological solution, whether it's on the pricing side, whether it's on the research side, whether it's on the data side. Clearly, you know, in your work, I'm assuming you kind of looked out and said, wow, these other people have tried this and they've failed. I wanted to take a certain amount of confidence to kind of go, I'm going to try it anyway. But I guess was there something about the marketplace that as you sort of saw it, you know, at that time in 2016, 2017, where you're like, you know what, I think I can do this because of where the market is today versus maybe 2012 or 2008. Yeah, so I think there's a couple of things. And again, we always want to speak very modestly about what we do because it's a really challenging problem. It's a gigantic project. So it's of course nothing that a single firm is ever going to solve. We just like to think we can make a meaningful contribution. To answer your question, again, around the time I was learning more about this, this was kind of the start of what has now in hindsight been a decade-long creation of this machine learning AI innovation that even a couple of years before the underlying technology didn't exist and then all of a sudden something existed which was pretty transformative. So that at least as a partial contributor or enabler to what we wanted to do is why this became possible when we got focused on doing it, whereas before it was not so possible. A related point which for us is really important for why cognitive credit exists in the first places. In almost every case, technology as designed or built for capital markets has really been built for equity markets. And there's always been a disconnect between what credit investors do, what they need, and what an equity tool can give them. And it created just too big of a gap. Again, when we got really focused on this project between how things worked and how they coulder should work. I don't mean to suggest that a decade ago technology could perfectly give a credit investor what they wanted. But for our whole career, you and me, the experience has been a credit person tries to use the equity tool, gets frustrated with it, puts it down, and does everything from scratch. And so we've all experienced it because we lived it ourselves, right? So the point is there are shades of gray along that continuum where surely the credit industry could be doing a better job with certain tools, certain technology, as opposed to using nothing because they're so entrenched, they've been so put off by what existed previously. And that's one of the issues we're trying to address. The other point is, and again, I know this having worked at a big bank, but I now know this better having talked to many, many financial firms in the market. It's really hard to build really high quality. Let's call it, use this term software even though that's a bit generic, but what I mean is a tool to help you do your job better. If the project isn't totally owned by the user, and the main disconnect that the market typically has is there is an engineering team, and they have a project they're working on, but they're not directly with the end users. And if you're on a trading floor or you're working at a hedge fund, et cetera, et cetera, and someone gives you something, and it's not really, really, really good, you don't have the time to work with them to really refine it and fix it. You know, the analogy we gave people is, because of course, financial firms have lots of resources, lots of capital, lots of we're with all to invest if they want to, but the analogy we would always give people is you can't juggle grenades all day long at your hedge fund, put the pen down at 8 p.m. at night, and then say, okay, for the next five hours, we're going to work on our tech innovation project. It's just not realistic. You're exhausted, you're stopping not because you're done with your to-do list, but because the to-do list is so big, you got to go sleep and come back the next day. So, it's really hard to develop really high quality software tooling if the end users don't fully own it as a project. And that kind of collaboration is just not very common in financial firms within the market. And so, what we do at our firm is take experts in credit and in technology and pull them and say, this is our only focus, and we're living, sleeping, eating, breathing it all day long every day. And it's been eight years of doing that, and we still have years worth of work, but to the extent that we get a great reception from the market, it's because we have kind of filled in the gap between that disconnect. So, you bring up so stuff, and I definitely want to pull on that a little bit. So, maybe a good way to do it is to think about, I'm kind of curious, when, you know, what was the first sort of problem that you were trying to solve for as a firm? And sort of, you mentioned, sort of bringing the right people into the room, and obviously, you need, you know, your developers on the one hand, but you also need a little bit of industry expertise as well to make sure to your point that they're sort of building the right solution. So, I guess, you know, day, whatever, right, you've got funding, you're starting the business, like, who are the first people sort of in the room, and how do you go? Like, this is the first thing we think we can tackle. Yeah, so the original focus of the business, which is still kind of our core IP, it's just expanded a lot in terms of what we do and how we do it, was this concept, could we reliably at scale machine-read financial disclosure trapped in PDF documents to build a credit model that would look and feel and meet the requirements of a human credit analyst? The intent or the spirit of that project was that there's a whole industry, right? Well, of course, you know, one of the world's biggest asset classes of highly paid, really intelligent investment professionals who are spending-- Thank you. I appreciate that. Who are spending a lot of their day typing numbers into Excel spreadsheets, right? And again, going back to this analogy of the gap between how it worked versus how it could work was just too great, right? It's not to say that it needed to wholesale change, but there's some graduation where people could spend more time on higher value added work, less time on necessary, but lower value work. That's what the technical project was, so to speak. And I don't want to get too technical on this call, but basically a PDF document is itself computer code. We might not think about it that way. There are very precise internals of a PDF document, what it is, how it works, trying to extract and structure data from an unstructured PDF has been a long standing technical challenge. If something existed at the time, we kind of did an exhaustive research project as we were getting set up because we didn't want to go spend time building something that already existed. If something fit for purpose existed at a time, we would have teamed up with them or licensed it or whatever, but we couldn't ever get anything good enough. All the tests we did for all the off-the-shelf PDF tools out there, they typically were good 60-80% of the time. And the issue with that is what are you going to do with all the cases where it's not good enough? And so we almost out of necessity started developing our own approach for doing this and ended up putting in place a very detailed data extraction process that is a combination of rules-based and machine learning-based elements so that we can automatically identify, capture, extract, enrich, and validate all of that financial disclosure that's relevant to a credit analyst and then construct detailed structured data sets that we can flexibly manipulate in lots of different ways. And some of the ways we present that data to clients is in functional Excel spreadsheets so they look and feel like something an analyst would have built. Other ways we deliver this is in kind of machine readable formats so people can algorithmically process the data. So there's different ways we distribute our data depending on the use cases, but that was the core problem, right? I have a stack of PDFs, surely we could get a computer to crunch all these numbers so that I as the human analyst can then spend my time applying my human brain to analyzing the data and deciding what do I want to do with it, right? A lot of, you'll know we'll come back to it later, but this maybe latest AI term, this term agent, for me it's a little bit unfortunate. I know why everyone's using this term, but the way I think about agent or agency is you are behind and have the control of what you're doing. I always really like this term agent to describe the role of the credit analyst, you know paired with modern tooling. We can get a computer to crunch all the numbers, distill it down into what the most important data to be analyzing is, but ultimately it's the human credit analyst, the human investor with agency to decide what are they trying to achieve, what do they care about, what is their analytical conclusion? So unfortunately for us, this term AI agent is a little bit conflating with almost our vision, for the future of credit. We'll come back to this later. So let's pause the point about the agent version versus agency. Exactly. So just to answer your question, that was the kind of day one vision. And we started the company in Europe for a couple of reasons. A lot of FinTech starts in Europe, starts in the US and then goes to Europe. We did this the other way. There's a couple of reasons for it. But I felt because financial disclosure internationally is less standardized and in certain cases, messier. I'll use that term to be diplomatic. It's sometimes messier. We felt that if we could validate this approach in that financial reporting context, it would allow us to scale that much more effectively into other markets. And so we started our service in the European high yield bond market. And the original year one vision was that if we could effectively build models for the 100 most liquid issuers in the European high yield index, that would be a real value to the market. Okay. Today, we cover over 3000 companies. So our remit in just a couple of years has been by a factor of 30. Okay. So that is just a reminder of how the target keeps moving. Yeah, it's interesting because at least for me personally, a lot of times, I kind of focus the other way around. Let's solve the easy problem first and then build up from there. But obviously, I'm in a very different seat than what you're in. And I haven't found a software company to solve all these big problems. But I have a couple of curiosities. Let's sort of piggyback on that because not least of which because you had mentioned some of those early PDF readers, maybe they've got 60, 80% accuracy. So I guess, you know, fast forwarding a little bit from that sort of here's the vision. And now you've got a product. And now you're trying to go to the banks or to the hedge funds or to ever else and say, hey, listen, I've got a product to sell you. I guess what's that education campaign like in terms of not only sort of getting them to demo you, but also convincing them that, you know, you're a better brand of ice cream after maybe they had tried some of these earlier tools that the efficacy was maybe not so great. So the answer is tough and ongoing. Commercializing a product is always challenging. I think what we have experienced is bringing our product to our market has very much followed a similar pattern as as a lot of software sales go that there's a pretty standard concept in Silicon Valley called the tech adoption curve. And no surprise, it's relatively shaped like the bell curve, but basically there are different consumers in the market that are earlier or later adopters. And what's been really interesting for us is there's absolutely no demographic correlation to early or late adopters, at least for cognitive credit. We have had old people be relatively old, you know what I mean, experienced investment professionals is maybe a better way to put it in the early days, be power users of cognitive credit, as have certain people read out of school. But vice versa, you know, their compliments, not necessarily that's not necessarily always the case. Just go through without me rattling them all off. Any possible slicing and dicing of buyer profiles in our market, it's been pretty, it's been pretty heterogeneous. And so the point of that is what really defines that is your psychological makeup as a consumer, right? Some people are more interested in experimenting and adopting and thinking about how to change how you work, other people less so. One of my favorite anecdotes when I talk to people about this in year two or three, we had a sales call with someone and they were like a 20 year experience credit analyst, so been around for a long time. So this person said, I totally understand what you guys are doing. And it looks pretty good. Call me when 90% of the market is using your tool and then I'll be happy to buy it. And so he is a late adopter, right? And when you're commercializing a product like this and you're really working hard to sell it, that can be a frustrating conversation if you're not familiar with that framework. And once you are familiar with that framework, you can just brush it off and say, okay, well, now we're going to call him back in a couple of years, but let's go find other people who are more forward looking and lean into it. So that is kind of the buy the book paradigm for how we've experienced selling the product. And then of course, what happens is the product keeps getting better and you keep refining it based on client feedback. And there's more word of mouth and you have a great user at one client and he leaves and he moves to another firm and then he brings this tool into his firm. And as long as we keep doing all the other things that are required at a startup, which I think we'll talk about in a few minutes, you know, it can build on itself. And so that's really been the experience. But it's not an easy problem in terms of defining what is the minimum threshold for this to provide enough value to an investment firm for them to subscribe because people have pretty broad requirements. Different firms have different priorities. If you talk to a global asset manager and they're active in different credit markets, sometimes it's challenging for them to consider working with a data service unless you can support more than more than one of their needs. And so when we only were active in European high yield bonds, very commonly people would tell us, well, I also need European investment grade bonds or I also need U.S. high yield bonds. And all these different combinations of coverage requirements is one of the reasons we've kept growing. Of course, our ultimate ambition is that if you are a global credit investor and you're active in lots of different markets, we want to be able to deliver you a very high quality consistent data set across all those different markets. But for many firms, that was like the prerequisite for us to be able to even engage with them. So that's why this is a long-term project. So I do want to definitely get into the project and some of the technology pieces. But I have a couple more for you just done sort of the founding and the firm itself. And I guess, you know, when I think back to some of these origin stories and I think back to the Bloomberg origin story, right? I mean, you know, for Bloomberg, at least in his book, you know, that relationship with Merrill Lynch early on as sort of both an endopter and an investor sort of helped sort of stabilize or sort of give stability around that sort of growth phase. Is this sort of all you or do you have partnerships or relationships that, you know, you have those early adopters who are also sort of engaged and say, hey, listen, we're going to be with you for this journey. Then I guess second early or the second question on top of the back of that. You know, you mentioned that sort of the early team is kind of like this pirates and romantics thing. First, I'd be curious which one you see yourself as. But then secondarily, you know, why is that the case? Why do you need that sort of combo platter of personality types? So a couple comments. So first of all, I probably should have said a minute ago. So I'm going to take the opportunity now just to give a huge thank you to our early adopters, right? We would not be here without them. We love our early clients because these are the ones who, quite frankly, took the risk to bet on us. So we will endlessly forever be grateful for all of their support. But then the second point is it's a great prompt for me to just say thank you to the entire team, right? Of course, this is a product of lots of people working very hard for a long period of time, not just myself. And we have had really amazing people along the ride. One of the things I'm so proud of is how great our retention has been across the team. We've worked really hard to find people who understand our mission, understand our motivations for pursuing this and have been really committed. So you use this term pirates and romantics. And so this comes from Jason Lemkin, who is really kind of a SaaS guru out of Silicon Valley and puts out all sorts of amazing content talking about how you build software companies and lessons learned throughout that journey. One of his rules, he often states is that an early company needs to consist of pirates and romantics. And if you can't find these pirates and romantics, don't hire anyone else because big company people at a really small company that has nothing, it's hard to make that work because it's really challenging to go from nothing to something, right? And if you haven't had nothing and taken it to something, you're not even really sure what you're signing up for. And so this term pirates and romantics, it's such a funny term because you kind of don't know what that means, but you also know exactly what that means. You have to be a little bit strange to join a four person company that has nothing. You have to be a different sort of person. You have to be in for the adventure, right? So yeah, yeah, that's right. So to answer your question, we had a great group. You know, cognitive credit day one was myself and into kind of co founding engineers. Richard Wheelden and Guy Hindell and they're really the technical brains behind the whole operation, certainly not myself. And then a couple months thereafter, we hired two other individuals and and then kind of kept growing the company from there. But yeah, that first year was five, six, seven people. And they were they were super committed and you know that that group is still still together. Everyone's still like cognitive credit eight years later. Maybe I'll pause there and hand back to you, but that's what that term means, right? It's really about finding people who are up for the adventure and super committed to try to build something, you know, believing in the mission, but also up for the challenge. So you mentioned that everybody, you know, sort of that core group is still sort of intact and obviously bought into the mission and enjoying it. I guess, you know, thinking about sort of the growth, trajectory of companies, right? Sometimes the people that are good for the early phases aren't good for the later phases. How have you sort of grown through that part of the cycle? Is it everybody that walks through the door is just that they end up being a lifer or is it do you see turnover just because the needs of the business change over time? Yeah, so for me, one of the worst learnings and you know, all these are learnings and that's good. That's how you grow and become a better person. But what I have found a lot of these things, you can read about them. But there's such a difference between reading about this and actually experiencing it. And a perfect example of this is what you're raising, which is that people who are good at a 10 person company and who can really effectively take you from nothing to something are not necessarily. Now, this is not a global rule. Like I said, that first core group, everyone is still together. But we've had other people who, you know, hire six through 12 or whatever, not not all of them are still with us. They were great. We just had a great group. But what you learn is that one of the things that motivates or interests those people is the lack of structure. Being able to do lots of different things unconstrained as a company grows, it's only natural that you have to put in place some more structure. You have to ask people to specialize. And so it's really tough because you work so hard to get that initial core group together just so you have resources to get on with what you're trying to do. And then you achieve success and the company gets bigger and then all of a sudden those people who have given you everything, they're, they're a little stuck. They're not quite sure where they fit in anymore or they're frustrated that they can't do four different things anymore because you really need them to specialize or they're not sure why all of a sudden a new person was brought in to lead a division and how they fit into the mix. And so that's a really tough transition. Now, sometimes when I reflect on these things, I try to give myself the benefit of the doubt. That's not a cognitive credit challenge. That's a challenge for almost literally every company that's ever existed. But it's really tough as you go through that transition because you are so committed and so dedicated to that initial group. Sometimes it's tough to see people move on, but then that's kind of the balance of having to do what's right for the company, do what's right for your investors and continue on the journey. So that's a tough one and it's a fast continual evolution. We we have about 55 people that are company right now. So still in the grand scheme of things that that's quite small for most companies, but you know going from one person to 55 itself over a couple of years. So that's a transition. So let's maybe change gears a little bit and and sort of get into sort of the technology and I guess if we're going to talk technology, probably I think we're almost required nowadays to start with AI. And so maybe we do that in sort of a credit friendly way, which is just talking about the evolution of credit itself, right, whether that's the introduction of electronic trade or some of these other dynamics. And then sort of where AI fits into it and maybe specifically how you see a eyes for complimenting what you're doing with cognitive credit. Yeah, so that's great. So I have been wary of using this term AI in previous years and the reason is a lot of businesses that are not that tech oriented or aren't really doing things in the field of AI like to use that term as a marketing term. We actually are a genuine machine learning company, right, we have invested years in training models and refining them and generating data and in certain cases innovating with the latest greatest technology. And so I've never loved this term AI and I think I have moved on from that because you know that this is now in the zeitgeist and it's counterproductive to hold back. I think if we talk about AI generically in society right now and then reflect on credit markets, the genie is out of the bottle with AI, right, this is not going back full stop, all right, we'll see where things are a couple of years from now in terms of how it's been applied, where it's been beneficial, where it's not been. But you have to be doing stuff with AI today, you have to treat it as a tool that in some ways will benefit you personally will benefit your business. I think the first point for your listeners kind of industry agnostic, if you're not deploying something that looks and feels like AI in your role today, you are behind the curve just full stop, you have to force yourself to do something. Now that's not easy for lots of reasons, but but you have to be doing something right when someone responds to that comment and says, OK, well, I'll give you the benefit of the doubt, I'm open minded to doing something, but I have no idea where to start. What we would say, and this is not just my personal advice, this is going to wear a lot of people are setting out, there's two areas within the enterprise where it makes sense to consider AI investment or experimentation. That's low risk areas, so if there are things in your business that need to get done, but they're pretty low risk. This is a great place to start because it has a very attractive risk reward payout. If you do an AI experiment and it doesn't really work, you're not really any worse off, but if it does work or you have positive results, that could be very creative. And then the other area is kind of the other extreme, which is really high pain points in your business, so if there's things you're not doing today, but you really wish you were or there's things that are super expensive or create tons of friction or there's a personal issue or whatever. A pain point in your business, that is a great place to start because again, if it doesn't go well, you're probably not worse off, but if it surprises you to the upside or you find a way to make something work, it could be really beneficial. And so these are conversations we are having all the time, and I think when we talk about AI generically, if you are being told from a top-down perspective in your firm, go figure this out, those are nice places to start. So I want to maybe kind of extend upon that a little bit, so I mean, I think when we talk, you kind of expressed also this dynamic of the model for your firm being sort of both coverage and then insights, sort of building out the coverage. And those are your primary two axes or axes which you're trying to, trying to grow. And I guess I'm kind of curious in terms of just sort of A, the sequencing in terms of how you think about which insights you're going to try and deliver once you've got the coverage piece. And then I guess how do you sort of account for reliability or maybe lack thereof for things like LLMs, large language models on the quantitative side where, you know, on the end user side, you really need precision, you really need transparency. Yeah, so let me take those in reverse order. In respect of your second question, I think this is really important for how we see our business in the market. To the extent no one's given this mandate, I don't want to sound too self aggrandizing, but we really view ourselves as the advocate of the credit market as it pertains to technology. What would be. be a real fail is if in a couple of years, the credit market is situated like it was 10, 20, 30 years ago, where there's been a lot of innovation for other asset classes. The equity guys are doing all sorts of really cool stuff, but no one's really invested in tech for credit. You've had this whole gen A.I. wave, but it's not really fit for purpose for, you know, the credit world. And we're still in this position where credit people are doing all this work manually. So we think about that 24/7. This is one of our reasons to exist as a company quite frankly. I think LLMs, as of right now looking into 2026, can do a lot of valuable stuff in terms of tech summarization, document synthesis, distilling certain points, right, can save you time, and depending on how valuable text output is to your investment process, can do work that humans would otherwise be doing. What we don't see them fit for purpose for at all is providing the quantitative support that a credit analyst needs for precisely the reason you're describing precision and transparency. And so as we incorporate LLMs into our product, it's much more about rounding out the qualitative side of the end-to-end research product, not the quantitative side because our process that we have already for generating all the structure data is really good. And at this point, I'm not really seeing how LLMs would augment it. I think going back to this point about you have to do something today that doesn't mean you have to use a tool that's not fit for purpose or you have to accept something that's not good enough. You can see all of this money that's getting put into LLM wrappers, something that's kind of going to sit on top of your big providers. If it's not custom tailored to the work that a credit investor does, if it doesn't really understand the domain, it's not going to give you what you want. And so you still need expertise from a given provider that aligns with your interests, your requirements. Yeah, I think that was sort of the primary question. And I guess, you know, and you kind of also hit on sort of maybe the types of problems that you're trying to solve for. I guess it strikes me as interesting to hear a technology company in LLM type of algorithmic oriented companies say the qualitative is the problem we're trying to solve for. It's just a little disconcerting in my head, but let's just to be clear. I mean, a lot of our product is already the quantitative and they do that very well. And so my point is, if the question is the latest greatest iteration, right, pretty off the shelf, Genai technology, how does that fit into what we're focused on? It allows us to round out or do more on the qualitative side. That's what I meant. Yeah. So I do want to maybe change gears and kind of move to the data side because data obviously is the lifeblood of all these types of things. And when we were talking in the past, you'd mention that sort of proprietary data is the asset and the model, the AI model is really just the commodity. I'm kind of curious to have you just sort of expand upon that a little bit for our listeners so that they, I think it sort of sets the context for the data discussions. Yeah, no, that's great. I mean, I think what's happening is very exciting and everyone needs to view it as an opportunity. I mean, this is revolutionary, the fact that you can get for very little cost such powerful technology that can do a lot of really cool things. You're going to have multiple providers in the next couple of years, give you access to this Genai technology that to the human user, it will be very hard to differentiate quality amongst these different providers. So well, in the whole, the capabilities will keep expanding, it becomes a little bit of a commodity because you can get it at a very low cost from multiple providers and it's hard to differentiate. Well, A's Genai model is meaningfully different than B's or vice versa. So then what really matters. And again, like I'm speaking from the perspective of the enterprise, a business trying to decide what am I going to do with this? What really matters is the data, your proprietary data that you're able to leverage from the LLMs. The LLMs, you know, no disrespect for these giant companies doing cutting edge innovation to generate the LLMs. But my point is what really creates the value for the individual firm using this technology is their own data. And so that's what I really mean by this point that it's the proprietary data that's the asset because that's what allows you to generate differentiated insights or in the context of your own firm and its requirements, deploy the LLM to give you what you want. You can't pass and won't name any names. But if you think about one such provider of this LLM technology, in many cases when you work with it, it's not quite giving you what you want because it was trained on all this generic information, but not necessarily your specific information. So that's the point about data being the asset and the advantage that every individual firm has to get more value out of this innovation or compete is being able to deploy your own proprietary data vis-à-vis this technology. Interesting. And I guess, you know, kind of along those lines in terms of, you know, credit data strategy as sort of a framework, I guess, you know, A, how do you think about it, but two or B, you know, sort of, it seems like we're in sort of this rapid stage of evolution. You talked about sort of some of the conversations that you've been having. How do you see this, you know, where are we in sort of the gestational phase, number one, and then I guess, you know, how quickly are you thinking about it, sort of evolving? So this topic of credit data strategy is pretty near and dear to our hearts. So we put out a white paper last year and any of your listeners who are interested can go to our website, cognitivecredit.com and download the white paper, but it's, you know, I don't know, 20 to 30 pages. What is that? Well, for years, we've been talking to investment firms about their approach to technology and talking through, you know, for their business, what's the implication of, you know, ongoing evolution in the market? And these were brainstorming sessions and we were all kind of talking it through real time, trying to figure this out. Well, they got to the point last year where we said we've literally had hundreds of these conversations, not dozens, hundreds, and there's a lot of overlap. Some things are unique to the individual firm, but in many cases, 80% of these conversations are very similar. So we said, let's aggregate all of these lessons learned from these conversations and get it down on paper. So we have a somewhat concise, consistent message to the market. How should you be thinking about these topics? And we anchor it all around this topic of, or this concept of a credit data strategy. And so what that means is, if you are a credit investment business, it could be the research team, it could be, you know, slightly more encompassing, we recommend, at a minimum, you should have a statement describing what is your approach for incorporating structured data into your investment business and why. Now, let's say you don't want to do that, right? The credit data strategy could be, we do not want to incorporate structured credit data into our business because of these reasons. Well, at least you have an argument. But in most cases, people won't take that position. They'll say, yes, we do want to do it. And you'd want to do it because there are certain performance benefits. There are certain cost and productivity benefits, but then you articulate specifically what are the things we're trying to achieve in terms of how we operate today versus how we want to operate going forward. And that's really the starting point because what, you know, going back a couple of years, what we would experience is we would talk about, you know, we define structured data, explain what it is, explain, you know, one of the analogies know what we give people is. If you had a Windows drive with a thousand Excel models that were built by individual analysts over different time periods and nothing consistently formatted labeled all sorts of different calculations. If you have that, which most credit firms do, or you had for those thousand companies, very consistently machine readable data, where the same concepts are represented, but it can be algorithmically processed. The latter is worth many multiples of the former in terms of what you can do with it in 2025. Okay. So that's just a definition of structured data and what the motivation is behind it. So then people will say, okay, well, I can go along with that argument, but where should we start. Okay. So then where where you start is define what is your data strategy? What are you trying to achieve? And you can target all sorts of different things, but at least you are clear on where you are starting. So then what we do in the white paper is map out through our experience and doing this many times. What we think are the prerequisites or the necessary conditions for a successful implementation of a credit data strategy to go through a transformation, right? If you're not doing anything with structured data today, you really want to what's required step by step to get to the other side of that and have a productive beneficial experience. And then what we do is talk about examples of best in class because within the white paper, we have a bit of a grading rubric. I think you've gone through that checklist, but it's just a number of questions to help you think about, well, where does my business stack up right now in this topic? And we give you a score, you fill it out yourself. And no surprise, most people don't score well, but that's not meant to insult anyone. That's just to give them a realistic sense of where they are versus best in class. So the one I most like to give just because everyone can get their head around it really easy. We have a continental asset management client in Europe. They have a global high yield, they use the global high yield benchmark as their index. It has around 1,300 issuers in it. Now, don't want to get too specific on this forum about the nature of their business, but let's assume they have a mid-single digit number of research analysts. So any of your listeners who do credit would understand that if you have a mid-single digit number of credit analysts who are tracking 1,300 companies, that's a challenging remit. So what these guys did with us a couple of years ago is build, leveraging our data, but their engineering resources, they built a market screening tool where through a cycle based on feedback or sub-allocations from different LPs, et cetera, they can toggle different search parameters, currency concentrations, rating buckets, you know, fundamental credit profile concentrations, et cetera. And this market screener goes through every possible debt instrument in the market for their index. And every morning when the analyst team comes in, they're given a list to say, "These are the situations we want you to focus on today." And so what that has done is turbocharged the analytical output or capacity of that team. But when the PM talks to me about this, he describes the biggest benefit of this system is they're never second-guessing themselves. They're never saying, "What are we missing?" Because they know, they're actively searching the entire market. Now, in 2024, at least in terms of who we were in touch with, who we talked to most in the market, that was a pretty unique setup. But that is not rocket science. It's not something that every firm can't have with a little bit of effort. And it's the sort of thing that, because it is totally achievable now, and it's so obvious why you would want that, if you don't have that in a couple of years and everyone else does, your firm is going to be at a real competitive disadvantage. For me, one of my biggest surprises is we've had all these conversations with the asset management community is how low of a percentage active research is on the index. And again, don't want to criticize anyone. But it's not so common given how much public credit markets have grown in the past decade that a brand name asset manager, their high yield team might only be actively covering 40, 50, 60 percent of the index. If that's the case, you're not even giving yourself a shot. If we're going to be honest, and so embedding tools like this that are built around data are total game changers. And the point of really having this conversation about credit data strategy now with the market is just so people are aware of what's possible today. Make it not sound so threatening or like it has to be that big of an investment to get in touch with really game changing capabilities. But also just think for it a couple of years, what does the world look like if I haven't made those investments today, but my competitors have. So maybe let's do that. Let's look forward a little bit. I want to also stay mindful of time. But thinking about sort of where you see the future going here, both in terms of how maybe the industry sort of real lines around technology as well as how you see cognitive credit sort of responding and sort of evolving from the current product suite. Sure. So we, because it's fun, but it's also something we need to be focused on are constantly debating what the future of credit looks like. And let's really mean just the next couple of years because we don't want to fool ourselves that any of us can predict too accurately too far in the future. So when I think about the future of credit, I think about leaner teams that is not necessarily a function of automation. It's a function of cost pressures of the industry and all sorts of, all sorts of other challenges that investment markets face. But it seems like a continuation of the trend over the past five or five or over 10 years, we would expect that trend just in terms of staffing to continue. We see a lot more machine-generated analysis qualitative and quantitative and relatively fair value pricing system. So this is something that is really a big focus for our company. Being able to assign transparent model-based pricing for debt, leveraging market data and fundamental data, which is really intended to apply more precision and reliability to work that credit analysts have done that their whole careers. We would expect systematic credit strategies to grow as a present of the overall market. We would expect electronic market-making to be more common down the rating spectrum. New issues will be less resource intensive as certain processes become more standardized and we can talk about that more in a minute. And then two things that maybe are a little bit provocative, but again, at the point of this is trying to inform or debate, we really see LPs focus more on differentiating relative strength in tech infrastructure and data sophistication as a means of making asset allocation decisions. And so going back to the point about credit data strategy, when you're thinking about what is the motivation behind this, there's a lot of immediate benefits for your business. You could respond faster. You could cover more of the market. You could be more accurate with your analysis. You could save costs, etc., etc. But then there's the secondary motivations, which are equally important. If your end investors are coming to you to have data-oriented discussions and you're not leading on that point, your business can be really at risk. Relatedly, making sure investment professionals have best in class data systems will also be a differentiator in recruiting and your ability to source top talent. So on the whole, the industry is going to shift a bit as more oriented towards data and tech because that's how all knowledge industries are transitioning. And going back to our motivation from Cognitive Credit a decade ago, when you asked the question, why was that the right time? Besides the other anecdotes I gave you, that was the time when lots of different analytical fields were really being transformed by this technology. So none of this is necessarily capital markets specific. All sorts of fields are benefiting, but also going through a pretty meaningful change because of this technology. Yeah, it's interesting because to your point, a lot of this sort of feels like it's all moving together and maybe has a little bit of a co-dependency as well. Systematic obviously is going to have some dependencies on electronic market trading and then to the degree that they can lay or in learnings from something like the Cognitive Credit, all that seems to pull in the same direction. But maybe we just closed because you did allude to the dynamics around new issues and while you feel those could be less resource intensive and sort of maybe some standardization there. Like maybe just closed it out on sort of your thoughts in that place because obviously we're in like a crazy new issuance market today. So yeah, well, so I think sometimes people say, well, you're not going to fully standardize the credit market. So how robust is this argument? And what I would say is you can have a lot of the market get more standardized, more automated, more algorithmic and still have a lot of growth in the overall credit market and the remainder of the credit market still end up very bespoke, very customized, and that's not a contradiction. That's, again, if you kind of took a macro view, look how credit has evolved over the past 25 years, that's really what's happened, right? And I think some of the technological change we're talking about either accelerates that and or concentrates it, but that is why credit is ultimately different than equity in the first place. You can have a individually negotiated $500 million loan to a ring fence project finance asset where specific cash flows are being paid on specific development events secured by specific assets, and it's all very customized. Well, that's the beauty of credit, right? That's not something that's necessarily well-financed by equity, but it's perfect for a structured loan. That example will continue. That in no way contradicts the fact that in a couple of years, people probably treat double B telcos a lot differently than they did 15 years ago. And you don't need humans typing numbers into spreadsheets and calculating formulas and dragging rows across and adding columns, and that's really not necessary to analyze that risk. And so I think you can have both those things be true at the same time, and that's the future of credit. - Rob, thanks so much for your time today. I'd like to, I mean, yeah, I'm not going to disagree with any of that, but I do also want to think our editing team that's Mary and Metadidia, they help pull the final product of these podcasts together to our listeners. We do hope you enjoyed this conversation with cognitive credits, CEO and founder. That's Rob Slater. 'Til next time, this has been Credit Crunch. (upbeat music)