Jason Jones: From Investor to Builder: Launching Ventures and AI in Accounting

Speaker A: And I just allocated to. Basically the strategy was go long the Internet and short anything that gets disrupted by the Internet. Great strategy. And it just continues to work to this day. Like that anyone who gets a credit card gets an interest rate of 20 to 25%. Whether you are a, you know, a recent college graduate or you're Bill Gates, if you apply for a credit card, your interest rate is 20 to 25%. And what I realized is that, you know, the blockchain was the next protocol. We'd done the communications protocols, we're now doing the money protocols. Right. 40% of all public company audits, or 40% of the ones that they reviewed, um, have a type 1A deficiency, which is the most serious type of deficiency. You do a procedure and you compare it to what the rules and regulations are. And if there's an issue, you raise it. Like, that's, That's a model that works in every, every vertical.

Speaker B: Welcome to another episode of a Growth Ventures podcast. Today we're hosting Jason Jones, co founder and CEO of teln. Jason is an entrepreneur with a proven track record of building businesses across fintech, AI and blockchain. In this episode we'll discuss that his journey from investor to entrepreneur, his approach to launching and scaling ventures, and how Talon is helping accounting firms with AI solutions. Jason, welcome to the show. It's great to have you here.

Speaker A: Hey, Hamlet, thanks for having me.

Speaker B: Of course, of course. So, Jason, uh, I mean, I'm looking through your CV port, uh, profile on LinkedIn and just to give some people some contacts, you went to Babson in 91 to 95 as an undergrad. If I remember correctly, Babson is like one of the, at that time was, if not the top or one of the top entrepreneur programs in the country. Is that correct or is. Yeah, memory.

Speaker A: It was number one in entrepreneurship and it's still number one in entrepreneurship today. Um, according to US News and World Reports. I went there because it is focused on entrepreneurship and they've done a great job just continuing to double down, triple down on entrepreneurship in their educational program.

Speaker B: And you were the class president. So does this mean you were the top entrepreneur in the top entrepreneurship program? What does that mean exactly?

Speaker A: That meant I was pretty good at getting beer for people.

Speaker B: Uh, so what are some of the things during your university days that you kind of learned that have kind of resonated throughout your career that you kind of can recollect and share with, with the community?

Speaker A: Just in general. Yeah, just not, not absent, but just in general, like.

Speaker B: Yeah, uh, exactly.

Speaker A: I mean, I was, you know, some people go to college to discover themselves and I was, I was born to be an entrepreneur. I knew it from, you know, when I was 12 years old and have always, kind of always enjoyed starting businesses, starting projects with friends and kind of being creative and kind of creating, creating a new business and then, and then, you know, making some money and having fun, uh, building along the way. So, um, to me the career has always been about starting new ventures. Um, and uh, my career just happened to, um, you know, I was, you know, had a newspaper out and had a, I actually had a moving business and a painting business when I was a kid. Uh, and then when I went to Babson, I, I ended up coming out and I wanted to do something around business fundamentals to begin with just so I could kind of get my footing as a, as a young, you know, professional m. So I did go into the investor business for a while, but it was always with the intent to get back to starting businesses. Um, and that's what I've done and I've kind of pursued this long, long career of starting, starting businesses, back of the back of the napkin type of thing, um, kind of going through that process which we can, we can dive into here on this podcast.

Speaker B: Amazing. So, so like 95, you, you graduated at Babson in 95. I graduated from SC at 2000. That just for some of the younger audience here, this is. We weren't doing pitch decks back back then. We were doing, you know, the entrepreneur community wasn't as developed as it is today. Uh, we were writing business plans and scrambling to find funding and things of that sort. At least I did while I went through school. Was it similar to, for you as well too? Or is that, is that about.

Speaker A: Yeah, I remember, I remember writing, you know, 100 page business plans, um, for new ideas, definitely, um, a lot of detail and I would say a lot of it carries over to today. I mean just they may come at different stages and, and maybe delivered in different forms, but the fundamentals are the same. You know, what is the market, what's the opportunity, who are the competitors? How do you get it? How do you get it to market? Etc? So a lot of the, a lot of what I learned there, I, I carry through to, to today and I feel like I just rinse and repeat, keep, keep going, keep doing it.

Speaker B: Amazing. So, so you graduated from school, then you started High Step Capital, which was a, uh, hedge fund, venture fund focused on technology businesses. Can you, can you tell me a little bit of how, how you did that, and you did that for a while. I mean, we, you went out for seven or so years or so.

Speaker A: Yeah, well, first I, you know, when I graduated, I, I went to some big businesses. I worked at Fidelity for a while. Um, I worked at Cambridge Associates where I helped endowments and foundations allocate their capital to venture capital funds and basically met all the v. I was 22 years old and I was meeting, you know, all the great legendary VCs of the time, uh, and then giving them money, basically being the gatekeeper between them and capital from endowments and foundations. Uh, that was really cool. And I got to see all the, all the, I met all the VCs and networked with them. But I also, my job, my day to day job was to call the portfolio company CEOs and basically asked them how much value the VC added to their business. So it was kind of like the gossip network. I would call and say, how did John Doerr do at Kleiner Perkins? How is he as a board member? What is his strengths and weaknesses? And I'd find out that he's really great at marketing and not so great at computer science or something. I'm, um, not saying that about John Doar, but just generally that's the feedback I'd get. And then I'd also find out about all the other VCs that are on the board as well. And we'd hear like, you know, this, this, this one vc, he's amazing. Or he or she's amazing. They, they're, they're so thoughtful and attentive. So it's kind of like doing due diligence on all the VCs while doing due diligence on one. And I'd also hear things like that VC never shows up at the board meeting or he falls asleep during the board meeting and he's just completely worthless. And so you get the good and the bad, right? So once I had that network and I kind of, I knew which VCs, like really added great value. I could then like, the most fun part of that job was finding the next, next generation of VCs that were starting new funds and knowing which ones.

Speaker B: The emerging VCs, the emerging funds, emerging

Speaker A: VCs, the ones that would spin out of the Sequoia and start their next fund. I would know which ones actually provided a lot of value and allocate them. So anyway, after doing that for a

Speaker B: while, um, by the way, incredible, incredible. First job, right? Like, or one.

Speaker A: Amazing. Amazing.

Speaker B: So accidentally this happened or like accidental?

Speaker A: Not at all. No.

Speaker B: Okay.

Speaker A: My first job was Fidelity, and I was in the back office processing trades. It was basically a, a white collar, blue collar job. It was a, uh, it was a pro. It was a processing line that get the trades in. I checked to make sure there's a, Wherever there was a trade error, I would have to fix it and then ship them out. I'd get a couple batches per day. I'd have to fix them as quickly as possible and ship them out and then wait for the next batch. I hated that job, but it got me into the door, right? And I knew I wanted to do something entrepreneurial and something around venture capital. So I did my informational interviews at night and would just ask anybody that was in the VC space. You know, could, could you talk to me? I'm, I'm a kid out of school. I want to be a vc. I want to be an entrepreneur. Can we just talk? Right. I did a bunch of those kind of informational interviews, asked friends, asked friends of friends. And, um, and that's how I landed the, the, the Cambridge Associates job. And it was, it was an amazing experience. Um, I was young and it was kind of weird allocating to these huge established firms, but it was also just a great, great experience for me.

Speaker B: And then what type of, uh, are we talking about? Like 50 million, 100 million, 25 million? What type of allocations at this age, at this time you were doing?

Speaker A: I mean, for instance, when Benchmark Mark launched their first fund, which was the fund that invested in ebay, I represented 40% of the capital that went into that fund.

Speaker B: Amazing.

Speaker A: Yeah. Uh, I mean, it wasn't my capital. I was consulting for the endowments and foundations. But my clients, they put a lot of money into that fund. Another One, SoftBank Fund 3, which invested in Yahoo and Groupon. I'm sorry, Yahoo and not Groupon. Um, E Trade. Yahoo and E Trade, back in the day. So these are Web 1.0 companies. Um, and I caught some really great venture funds when they were super young. Um, and then from there Goldman, um, came calling and they were starting an Internet research group. It, um, was kind of the first one on Wall Street. Uh, and I kind of knew all the VCs and all their portfolio companies, so I was a good match. And while most people kind of have like 10, 20 interviews before getting into Goldman Sachs and you have to be kind of top tier, just my network, um, opened the door for me and I was able to do it. One interview, got the job, and had spent another few years as an investment Banker slash research analyst at Goldman, bringing a bunch of Web 1.0 companies public. Um, and that was great.

Speaker B: These foundational, that's like a complete full circle. Right? You invested in when they were like precede and analysis and you're bringing them to public and you're kind of going through the full life cycle of the journey for the startup.

Speaker A: Yeah, well, I invested in the funds and the funds, uh, in them, so I didn't invest at that point.

Speaker B: I, I know, I, I know, but I'm just saying.

Speaker A: Yeah, then I did the public offering thing and then I went to the hedge fund side. Well, actually when I was, I was at Goldman, I was like, I'm going to be a VC. Like that's the idea. And then, then one web 1.0 crashed and all the VCs got destroyed and all the hedge funds who were my clients at Goldman continued to do really well. So they did well on the upside and they did well on the downside. And I was like, that's a good business model. They're always making money. No matter what the cycle is, they're making money. Um, so I ended up becoming a hedge fund portfolio manager and analyst and kind of spent another eight years, 10 years kind of on the buy side, um, investing in technology companies. Um, first for a couple big hedge funds. And then I launched my own fund called High Stuff Capital, which you mentioned. And I just allocated to. Basically the strategy was go along the Internet and short anything that gets disrupted by the Internet. Great strategy. And it just continues to work to this day like that. It's this, the disruption. It's the disruption strategy. Right. So we'd uh, also use a lot of data, come in data coming out of the Internet to kind of pick stocks. We could go into a lot of detail about like some of the early, early use cases that we would, early ways.

Speaker B: Uh, so let's, let's talk about that. So, so obviously you at this point have been in finance for about a decade. It's about a decade or so. 8, 9, 9 years or 12 years

Speaker A: in total where I was on the, on the buy side or in the investment world, I would say 12 years.

Speaker B: 12 years. Okay. And one of the biggest things that you just kind of alluded to is what Clayton Christensen talks about is it's really disruption theory. Right. Like what is the new technology that comes into place? What is it going to disrupt? How is it going to allow a much more nimble, faster company to be able to really take a bite out of the apple of this new technology? That's come around and really take away market share from existing. It's kind of the, the philosophy. It sounds like that you were, uh, approaching your investment thesis around. Am I summarizing that well or.

Speaker A: Yeah. Yep. And I was. I, I am to this day. I'm a fundamental research analyst. I believe in finding the big trends, finding the white spaces and allocating to that space. Right. Either as an entrepreneur or as a investor. Right. I'm finding, I'm finding where the puck's going to go. Right. Wayne Gretzky, right. Finding where the puck's going to go. And, and then I want to find the teams that can actually execute and build towards that big white space. Right. And it's always big white spaces. You don't look for little ones, you look for big ones that are going to change the world. Right. So, um, that's always been my thesis.

Speaker B: How do you do that? Like most people. Uh, so, like, this is great because we're already jumping into some of the practical things, right? You've done it so well over the years and so many, you know, being fortunate as long as we've lived and we've seen a couple of cycles, a couple of times around now at this point, right. Uh, how do you, how do you, how do you say, okay, I, this is, this seems like to be a disruptive technology. What are the industries, how they're going to get impact? Can you elaborate a little bit into that?

Speaker A: Yeah. Number one is like, you talk to navc, they'll say the same thing. Pattern recognition. You do this for a while. Any old vc, they'll say the same thing, right? They've done this for a while. They've seen this cycle before, They've seen this play out before. So I feel the same way. Like I keep seeing patterns repeat over and over again. Um, so you kind of know how to play the game, game that's being played, the new technology, new trends that are out there. Um, so pattern recognition is huge and I would say training, uh, in kind of fundamental research how to really understand, um, not only the part of it is understanding the finances, the accounting and finances, which Wall street really, uh, uh, uh, trains you. How are companies playing with their numbers? Um, what is the real performance of the company? Why are they showing you an adjusted EBITDA number instead of a profit net income number? What are they hiding? Why are they showing it that way? Um, and really backing to numbers. So that's one thing on the Wall street side, but then on the Silicon Valley early stage side, it's More like, okay, what is this technology and what is the opportunity with this technology? How is it going to change things? Um, and if it is, where's it going to change things most? And I can go into my latest business, which is an AI business, and how we kind of got to where we got to with that one.

Speaker B: Yeah, I love that. So obviously you saw it with the Internet. AI is the technology disruptor of today. Uh, so let's. Before we get into all of that, um, can you, can you kind of. When, when you think about. And, and I'm trying to kind of frame this for everyone that's kind of listening in and how you've had the pattern recognition component of it. Have you seen this? You saw that in the inner era. Uh, and then what was another area that you saw this in as well too? If you can kind of give this so people can see the storyline of why now you're going after this particular space.

Speaker A: Okay, so the Internet gets started. It's a disruptive force. People say it's the greatest wealth creator of all time. It was also the greatest destroyer of wealth of all time. Right. For every Amazon, there is 20 or 50 circuit cities. That was once a successful business. That's no longer. There's newspaper industry that got decimated. So, yeah, it did create incredible value for some of the top technology companies, but it also destroyed incredible value for so many companies and industries. Right. And I've seen this pattern over and over again. And I would say that Web 1.0 was all about the communications protocols. It was like newspapers. It was, it was also retail, like these, these different areas that really got where the, they really crossed with the Internet. Like the Internet really just decimated a few industries and I would say transformed those industries. There are some winners, there's some losers, but it really transformed those industries. Newspapers and retailers are two good ones, and there's a bunch of other ones as well. Um, I saw it. You know, I invest, I've invested in so many online marketplaces over, over time. I just love, I love that, that business model. I mean, the Internet's so great at matching, um, buyers and sellers, you know, workers and hirers, people who want to come together to kind of make a market. Right. That's a great thing about the Internet. There's so many online, online marketplaces out there. So that was a key theme for me in all of my investing. And, um, eventually in 2010, I found a company called Lending Club, um, which was only a couple years old at the time, was matching borrowers and Lenders. And um, I just went down that credit rabbit hole. I was an equity investor. I didn't know anything about credit, but I just found it so interesting that somebody who had credit card first. I found it interesting that anyone who gets a credit card has, gets an interest rate of 20 to 25%. Whether you are a recent college graduate or you're Bill Gates. If you apply for a credit card, your interest rate is 20 to 25%. So it doesn't matter how wealthy you are or how much financial risk you represent, you're going to get a high interest rate on your credit card. I found that fascinating that it's not risk based pricing. Everybody gets the same high interest rate. And what lending club was saying is like, okay, well if you are uh, a Bill Gates and you do carry credit card debt for whatever reason, I don't know why he would. But if you did carry credit card debt, wouldn't it be better if somebody could come along and provide you with a loan at a lower interest rate so you could refinance your 25% debt down to 12% or 11% or whatever the risk. If we could risk based price those highest credit risk people wouldn't um, be great to lower that, that, that cost of capital for those people that do have, you know, great credit. And I found that, yeah, that would be great. And that person's taking control of their financial life. I love that. And on the other side, as a, as another, as a, as an investor, I could give them the money so that they could do so and I could make a 10 to 12% return on that. I was like, man, this is a great return I can make if I can do this. It's the same type of returns I'm getting on my market neutral hedge fund. But the credit market is so huge and it just, it brought me down this rabbit hole and I ended in the credit industry for, you know, quite a few years. And uh, I built, built a few businesses around that. So I built a robo advisor that was in the peer to peer lending space and became one of the largest robo advisors.

Speaker B: And this was called Lending Robot, right? Is it that one?

Speaker A: Yeah.

Speaker B: Okay, nice.

Speaker A: It did have a couple of iterations in the names, but the final name, the name today is Lending Robot, um, still, still in existence. It was uh, we sold it in uh, 2017 but uh, you know, it's still operating out there. But um, that was ah, that was an interesting business kind of allocating to consumers who were refinancing their debt, uh, Allocating to small businesses like dry cleaners and pizza shops that were, you know, taking out small business loans. Allocating to real estate, um, uh, investors who were doing fix and flips on, on, on, on uh, residential homes. Um, we found all sorts of assets that we could kind of lend against, um, and get some really interesting yield. Um, so that business was great and we kind of built these consumer portfolios that were low risk, medium risk or high risk. Um, you know, yield would match the, the risk level. Um, and we were able to kind of figure out how to digitally onboard the investors, um, allocate their capital all kind of electronically, um, show the kind of investor balances, you know, all digital and then, um, and then withdraw them, you know, just all digital. So we basically built a workflow that was completely online and went um, from amazing a fund to a robo advisor. Right. Kind of built that up.

Speaker B: Was this your. So this was now your first transition from being an investor to being into fintech and now building your own startup as it was. That's, that's phenomenal.

Speaker A: And it was gonna be, it was gonna be a credit fund but we turned it into, instead of a fund, we turned it into a completely digital experience which made it a robo advisor as opposed to a fund.

Speaker B: And you're so humble. You're like. At the same time you were also building sort of a, a fintech media company or. Yes. No. Uh, was this also occurring too or.

Speaker A: That was like, that was a side project that turned into a pretty business. Basically. When I was allocating the fund, I wanted to meet these peer to peer lending CEOs to figure out which ones to put my money with.

Speaker B: Right.

Speaker A: Uh, so I was like, I'll, I'll create a meetup. Um, it was called Lend it. We said we'll make a meet up and you know they'll, and you're being so humble.

Speaker B: Lend. This is like one of the, one of, one of the fintech's premier sort of uh, meetup slash conferences that happens. Is it twice a year, once a year or how often is it still happening?

Speaker A: Yeah, well it's been sold now. It's now called fintech Meetup. But um, that business was ah, um, in its heyday it got to be really big and way, way bigger than I ever expected. I thought it was going to be a meetup and it turned into like this huge um, uh, fintech event and would have it in the uk, we'd have in the us China. We had a huge presence in China. I had a 12 person staff in Shanghai. Um, yeah, it was crazy. It just kind of expanded my worldview kind of globally, um, and running. I got to see the power of the media as well. Like what, like really how, how the media can expand your presence just globally so quickly.

Speaker B: Um, so how are you doing both? Right. Like here you are, you're building Lending Robot, let's call it, and at the same, a startup and a robo lender or advisor. And then you have this media company that's also starting to grow. So you obviously context switching a little bit here and there, back and forth on two different types of businesses. Uh, how are you figuring this out?

Speaker A: It's your personality, right? Some people just need to be really focused on one thing and some people just gonna do one. Some entrepreneurs will build, have one idea and will build one thing their entire life and perfect it. Right. And I totally respect and admire those people. Other people like myself, I love creating new businesses. I love the zero to one experience. And I'm happiest and most productive when I'm doing, you know, one or multiple things. Right. There's always another one in, you know, in the pipeline. Um, which makes me a little bit like a vc, but more like I've kind of figured out what my role in the world, which is basically a serial, a serial entrepreneur. That's, that's really what I am. I love building them and you know, built up some great relationships on the VC side. So, you know, I can get them funded sometimes too.

Speaker B: That's amazing. So, so now let's talk about AI. Right, so, so you, you went through, uh, when was the first time was it that you saw the potential of AI and you said, man, there's a lot here. Was it three, four years ago when OpenAI released their APIs or when were you a part of this? Okay, there's something happening here in the world.

Speaker A: All right, So I said the trend. In 2017, I sold my Robo advisor, um, and I, I saw two paths. I could either go down one path, which was blockchain, or the other path that was AI. I realized these are two kind of fundamental technologies that are the next wave, right? They're going to change things. Um, on the AI side, this was, this is, this decision was made in 2017. But going back to like 2014, 2013 in credit and lending, those businesses aggressively adopted machine learning early, very early in the AI cycle. So I spent a lot of time, even in my own, in Lending Robot, we spent a lot of time thinking about how we could. When you, when you allocate to a loan, like uh, to a borrower. You get all sorts of variables on that borrower. Like you're not only saying like, what is their FICO score, but you're also seeing things like how many credit cards do they have? Are they married or are they single? You know, do they live in a home or are they a renter? Like there's like literally like 100, 150 different variables for every single loan. Right? And the lenders originate like 4,000 loans a day. So we were building this technology where. And by the way, when all those loans drop, the first buyer, whoever gets in first, got to buy the loan. So we would build this technology that was kind of like a, ah, high frequency, uh, buying, that would ingest thousands of loans, review every variable and then figure out which variables were most meaningful, which ones would give you the biggest signal that this borrower wouldn't default, and what machine learning would do. So, so then you start doing regression. You'd say like, well, a person with a fico score above 700, that's a homeowner that has at least three credit cards, that person has a lower likelihood of defaulting than, than, than others. Right? So you, you mess with the data, basically do all this data analysis.

Speaker B: So this, this wasn't necessarily an underwriting model. This was more of a loan acquisition model. Am, am I, am I hearing you correct?

Speaker A: The loans had already been underwritten, but now they're being made available to invest in. So it's like underwriting, which uh, is loan acquisitions like loan purchasing. Right. So we were like looking at all these variables and we realized that traditional logistic regression would take one variable and regress it against two or three other variables, which, that's what, like American Express and Capital One. This is what they were doing. And then the whole industry moved into machine learning and instead of regressing one variable against three variables, with machine learning, you could regress all variables against all other variables. Uh, so you'd have this huge massive network of kind of, um, correlations happening kind of in the black box. And at the end of it it would come out with, okay, these are the loans that are most likely to not default. That was the early stages of machine learning, one of the first applications of machine learning. Um, so we had adopted, our industry, had adopted AI 2013, 2014. Um, it was really early on. Um, I mean it's obviously gotten much more sophisticated, but that, so I spent a lot of time in the machine learning space as um, a result there, um.

Speaker B: Is this what folio was. Or which company was this?

Speaker A: Or this is still lending robot. This is still.

Speaker B: This is still lending robot. Okay, got it.

Speaker A: So then we get to 2017, and I had to make the decision, do I. Do we go down the AI path, or do I go down the blockchain path? And what I realized is that, you know, the blockchain was the next protocol. We're done. The communications protocols. We're now doing the money protocols. Right? It was just another. It was an extension of the Internet, right? It was figuring out how people can. How people can collaborate, um, in a protocol. Protocol is a system of collaboration, right? It's a bunch. It's a bunch of people that are getting together and following some rules that allows them to collaborate in order to. To do something together. So to me, it was like another evolution. It's just the Internet. It's just another. It's another layer of the Internet that's being built. Um, so I had to make the choice. I chose blockchain at first, and I built two DeFi protocols in the blockchain space. But I still always loved AI, so I kind of gone down both paths. Um, we can talk about the blockchain side. We can talk about the AI side. Really? Whatever you'd like to talk about.

Speaker B: Yeah. No. So you were going down the blockchain tide, you're going down the AI, The AI path. Uh, so let's. Let's go down a little bit about the blockchain path a little bit and see. At what point did you decide, okay, it seems that there's more opportunity here in the AI path now? And let me kind of shift focus. Has that occurred yet? Or do you still think they're kind of. If you look at both of these new technologies that are occurring, do you think they're accelerating at the same pace? Or has AI taken. Taken first step forward now? And blockchain is about to start going. Where do you see this today?

Speaker A: Yeah, I would say that if you're a surfer, um, the blockchain wave is a big wave, but the AI wave is like the waves. What is it? Maui or like the ones in long. I forget what it's called. Half, uh, Moon Bay. Right. They have the giant waves. That's. That's like the. You know, it's like a huge wave. So anyway, you're. They're both really great weights and they're both got really great opportunity. Um, and in the blockchain space. Yeah, going down the blockchain path again, what I would say is that it's A system, a protocol. This is what gets me. It's a protocol, is a system of collaboration. Right. What that means is we follow protocol every day when we're driving our cars and we get to a stoplight, a stoplight, and there's red, yellow, green, and everybody follows the protocol that when it's red you stop and when it's green you go, and when it's yellow you proceed with caution. That's a protocol. We all follow it. It's a system of collaboration and allows us to operate efficiently. Right. It's all that blockchain is. It's a bunch of. It's a protocol. Right. So I, I think about like what protocols can be built in order to get multiple parties to collaborate and do something really efficiently.

Speaker B: So what you're really good at is spotting when new technology comes and it disrupts something. What would this protocol disrupt for some people who might not know about it?

Speaker A: Yes, that's what I think about, like, what's this? This is machines, new tool. What will this tool be good for? Like, what industries will it be good for? How can we use it? So for the blockchain, having just sold a credit company I was looking at and saying securitizations are really big business and they need multiple parties. You need to have a trustee, you need to have an issuer, you need to have a collateral manager, you need to have, uh, investors, you need to have custodians. You have all these different parties that come together so that you can pool a bunch of loans and issue a bond that's backed by these loans. Each one of those parties have different roles in a securitization. And if you could create a protocol that allows those parties to collaborate, once one service provider completes their task, then it has to move to the next service provider who has to complete their task and then the next one. If you could create a protocol, uh, to make that a really efficient marketplace, then you could drive down the costs of completing a securitization and issue more, faster, better, cheaper. Basically that kind of concept is not just for securitizations, but for like any financial transaction and all the different service providers, uh, that are in the financial services industry. Protocols can be used all over the place once you start thinking about it. So the one that I, that I attached to was the securitization one and I built a couple of securitization protocols as a result. And I think that's you keep hearing about kind of, uh, tokenization is going to get huge. It's going to get to billions or trillions of dollars and, and uh, all these big financial institutions are going to come, come online and real world assets is kind of going to be all tokenized. Right? I built that, I built that in 2000, 2017, 2018. My, uh, first protocol was called Center Futures, one of the first real world ass protocols. And that's still a trend that I think is there's tons of upside opportunity there to this day.

Speaker B: Amazing. And so now let's talk a little bit about AI. So same question back to that AI. I mean, clearly a lot of people are still like arguing, has it working? Is it disrupting? Where it's that, where is it at? Ah, I could tell you as an agency owner, we, we've been using it for now, three, four years or so, uh, very adamantly. And a ton of our different workflows from content creation to um, you know, everything from analytics to emails to, to even image creation to voice, to like, I don't think there's a workflow that we have that doesn't have AI inside of it or has an AI sop around it and, and more even excited about like this whole world of agentic AI and all of those different things going on. So, so I'm, I'm clearly biased towards this technology. Uh, but tell me about your perception of it and, and where are the, where are the major areas you would, with your incredible pattern recognition that we've talked about in all the years of things you've seen? Just, just kind of go. Here are all the places I, I know, Ken. And here's a utility for this and all the different places I could, uh, disrupted.

Speaker A: Okay, so it was, it was March 2023. I had to shut down a venture that I'd started, uh, called N Labs. Uh, if anyone knows about operation choke point 2.0. Um, it's been a lot of press recently. Basically my business was shut, forced to shut down after the banks went under and stopped serving crypto companies. Oh, well. Oh, well. So sad. You get, you get some losses sometimes. So sitting there in March 2023, um, with my CFO, who'd been my CFO for three different businesses in the past. Um, and before being my cfo, he was, uh, he was, uh, an auditor at Ernst Young UI. Um, and ChatGPT had launched in November 2022, so it was only a few months old. But that was the first time the public had seen the power of generative AI. Generative AI is a lot different than machine learning. Right? Machine learning is really about, uh, give A ton of historical data, like credits that I mentioned, repeatable structured data and let it figure out what are the uh, predict what's the future based on the past data that you're giving this past data, structured data stream that is machine learning, at least in my view that's kind of what the best use application of machine learning is. Generative AI is more about generating text, generating reasoning, uh, just figuring things out, um, basically being your co pilot or your assistant that you can kind of talk to and kind of generate new information with. Um, so when we run generative AI, if we're looking at a, like if we're looking at a financial statement or something, we may pull out all the text and read it through an LLM and then we may pull out, we may use other technologies to extract the tables and the data and run it through Python or something to do some calculations. So it's kind of like a mix that you know, with generative AI. Um, but we realized quite like most people when you first played with ChatGPT for the first time, kind of blown away by what it does and its potential in the world. So we were sitting there and um, basically by June 2023 we were thinking about our next venture and we were saying, all right, this generative AI thing, it's going to really impact knowledge workers. Just like Web 1.0 impacted retailers and newspapers, this one, this wave is going to impact knowledge workers. Right? So which of the knowledge worker industries does it make most sense for? And we spent some, some quite a bit of time thinking about that and we landed on accounting, um, and then we landed on audit within accounting. Basically accounting is the, the language of business, right? It's the fundamental business block of business. Right. And within that audit is the, is the ability, is the task of reviewing the uh, the language of business and attesting to whether it's valid or not. Right. Um, and we felt like that, that is really where you get, you're getting to a building block of how all businesses are built and uh, and something that, that, that everybody needs. Now you go peeling that on your back a little bit more. What does an auditor do? They get a data set, last year's financials that is static and they need to go run about a thousand procedures, accounting procedures on that data set. Each one of those procedures is defined by rules and regulations and they have to go through and do each one of those with tremendous precision without making any mistakes. And once they do that, based on the results, they can make a decision of whether that company's financials are, are valid, qualified or unqualified. Right. But they're there whether they are valid or not valid. Right. So, um, that's what they do.

Speaker B: Right.

Speaker A: All ah of that is extremely procedural, um, uh, and just really attuned to something that AI can do. Then we realized that the number of CPAs in the industry is dropping dramatically. All the accountants are approaching retirement age and there's very few people graduating from colleges with the CPA, uh, focused on the CPA industry anymore. The number of CPAs are down 17% over the past three years. And as a result, there's just not enough workers to do all the work. Uh, it also means that the people that do the work get promoted faster and are, and this is an apprenticeship industry, so they get promoted faster and they're prone to, to make more mistakes than their, than their predecessors just because they haven't had the same amount of training. Um, while we were doing our research, the pcaob, which is the government entity that regulates public company audits and auditors came out with a statistic saying that 40% of all public company audits, or 40% of the ones that they reviewed, um, have a type one, uh, A deficiency, which is the most serious type of deficiency. 4, 0 and yeah. And we. The average S&P 500 audit costs something like $12 million. And they do 10,000 hours spent on that company to do an audit. Like if you audit like an Apple or an IBM or something, amazing amount of time and energy and dollars is spent to do an audit. And 40% of the time there's deficiencies like that blew us away completely blow us away. Why are there so many mistakes? And it comes down to two things. One is what I just mentioned, that the people doing the audits, they're getting promoted too quickly. They're just prone to making more mistakes because there's less kind of experienced auditors. Uh, on the other side, you have extremely complex regulations that get more and more complex every year. And it's just really hard to keep up by all those regulations. Perfect use case for AI Right. Read all the regulations. Read each one of the procedures and make sure that procedures follows the regulations to a T. And if it doesn't raise the issue, tell the auditor there's a potential deficiency here. Right. It's basically amazing.

Speaker B: Uh, so where, where's telling at? So, so you started telling and it looks like 20, 23. Uh, are we. How many, how many clients do we have? How many people are. Who are you selling to? Is it to the big Audit, uh, audit firms or like walk me through where it is in its journey right now.

Speaker A: Yeah, so, uh, it's been, it's been a good, a good journey. We, we launched, we came up with the idea in June 2023, October. We, we incorporated, we said, yeah, we're going to go for this. We closed Pre seed round, $500,000 pre seed round. Um, and we started building the, the building the base product. And uh, you know, we also went out and started selling before we had anything. Right. I think that's one key lesson that you know, that give to any aspiring entrepreneur is like, you sell from day one, whether you have a product or not. You start selling, you start, you sell vision, right? You sell vision. You want to get people that will use your product, that will talk to you, that will commit early. Right? So from day one we were selling. We, we ended up building the product, releasing the base version of the product in the spring of 2024. Um, basically right after busy season April 15th. Accountants are extremely busy right up until the end of April, at which point they free up a little bit. So we had our base version out ready to go by the time the busy season was over. Um, we then brought on, we kept of iterating on the product, we brought on our design partners, uh, early design partners. We kind of iterated a little bit on what exactly the, we knew what the platform would be, but we didn't know what the specific applications would be. We knew one would be an audit quality because of the deficiency thing that I mentioned. But we had to also kind of figure out what other products. So, um, we iterated a little bit over the summer. We landed on, you know, two really great products and we rolled them out in September of 2024. Um, MVP. And then we basically started marketing that in November. Um, so really we've started marketing since November. And now here we are in uh, February 2025. We have 10 clients. We uh, just signed our first top 20 client called Cone Resnick. It's a big, big, big accounting firm. We have uh, we have a number of kind of mid size firms that are kind of top 100 type firms. Um, everybody's so enthusiastic about how we can apply AI to their business. And uh, yeah, I'll stop there and see if you have any questions.

Speaker B: Oh my God, that's phenomenal. So some amazing lessons for the listening audience here. Right? So like you, you, you spent a good two or three months really validating the business itself and saying that, hey, and looking at the different data, the metrics speaking to people, figuring out what the real core problem was.

Speaker A: Product research.

Speaker B: Yes, a lot of research.

Speaker A: A hundred auditors in like three months. All we did was call as many as we could.

Speaker B: Amazing. And this, I'm sure your media background at this point and your early days of what you were doing as, uh, uh, all came into play here where you're not, you're not shy at all to pick up the phone and make the phone call.

Speaker A: So, so I do cold calls. I, I'll call anybody, you know. So it, yes. It requires you just to dig. Just to be hungry and dig and just find people that'll talk to you.

Speaker B: I love that. And then once you had the reassurance that you were on, uh, there's something there there, right? You're like, there's something definitely here. There's a really good use case. You started pitching. You started. The product wasn't even built yet. You, you had a good idea of how you guys can build it, but you actually flipped it, flipped the script here and you started really pitching and starting to really see, uh, how much genuine interest you would get from potential clients and customers. And now fast forward. With the product being only out and really in market less than a year, it sounds like you have 10 clients already, one of the bigger accounting firms on board, and it seems to be living up to what you wanted to build. Right. So what are some of the current, uh, firms feedback that you're getting from them and how the product is performing? Is it, uh, what is the performance enhancements that they're. They're seeing from their accountants that are using it? Is that still too early to kind of know that or what is overall genuine, genuine feedback you're getting from the clients that you have?

Speaker A: Okay, so we're a little early for that because we built, we built the platform and now we're going to build the applications, and we're building the applications with our clients right now. So there's nothing that's like totally in market and usable yet. They're pretty complex applications, but the underlying platform is there. And uh, so I can tell you a little bit about that. And that gets into kind of agentic workflows and AI agents and that type stuff.

Speaker B: Uh, yeah. Why don't we talk about. Yeah, applications that you're kind of understanding now where you see all of this could potentially go. So like, as you're integrating more and more with the firms.

Speaker A: Okay, first of all, we are a vertical AI specialist. Right? We, we have one industry and we're going very deep in that one industry, our industry has extremely high data insecurity, confidentiality issues. Of course. Right. The uh, an accountant may have a thousand clients and every one of those clients is giving them their private financial information so they can kind of complete their accounting with, with that client. That's the information we do our analysis on. Right. So it's not like an accounting firm will say, hey startup, here's a thousand financial statements from all of our clients, go take a look at them and give us your analysis. And they don't ship their data to us. That's not happening. Right. So the first thing we had to realize is how can we structure a business? Because we can't be a SaaS business. They can't, they can't send their data to us and have us, you know, analyze it. So the first thing we had to realize is that we needed to set up uh, virtual private clouds for each one of our clients that's owned by them. They can drop their data into their own cloud. We would harness AI models and put them into their cloud and then we would build our applications in their environment. Right. So none of the data leaves their environment. That was lesson number one. That's a key to basically any enterprise that has sensitive data. The structure of the business has to change a little bit versus what you know, traditional SaaS businesses have done. Right?

Speaker B: Right.

Speaker A: Especially when you're young and you have no reputation. Right. Once you got the back end done, then we needed to figure out platform and applications on top of the platform. We called ourselves. We would say we're, we harness AI models, whether it's OpenAI or anthropic or llama or whatever. We, we take in the, the models and then we build applications that are specific for the accounting industry. And we're really good at that. We've got a whole staff of CPAs and we kind of build these applications that are specific for our industry. And that was interesting because we kind of built out this platform, this middle layer platform that would ingest the data or ingest the models and then we would build these workflows and these applications that were specific around what an auditor would do on a day to day basis. Might be a simple thing. The one I'm going to give you is like you're writing a financial statement and you need to create a, uh, footnote about a lease contract. Right. An auditor would read the lease contract from their client that would give terms like this is the start of the lease and the end, and the end of the lease and they, this is the monthly payments they make and you know, has all these kind of terms of, of, of, of the lease. Right. They'd have to take that and they would have to build an amortization table in Excel. They take it in a PDF and to build an amortization table in Excel that kind of present values back the total value of the, of the lease and then need to take that, that model in Excel and then have to write a, a footnote in Word that describes exactly what this company has done and describes the model that they built and ultimately ends up at a number. Right. A lease number that ends up in their uh, financial statements. The process goes data extraction from a PDF transformation in Excel, generation of a report, a footnote in Word. Right. Three set for steps. Right. That's a workflow. Step one, step two, three. Each of those steps has an agent that can complete that step. The first one is the PDF extractor agent. Their role is to extract data from a PDF and put it into a table. Right?

Speaker B: Right.

Speaker A: One is a M Excel modeler agent. Their role is to take inputs of data from a table, complete some type of formula based transformation of that data and output a new table that's going to kind of does uh, whatever it's supposed to do. The third one is the report generator agent that generates a new report based on the reading the lease contract and extracting the data from the Excel table and put together a pretty report at the end that describes what the process that happened. Right.

Speaker B: I love it. I love how you broke down agents into simple workflows and simple steps. Um, so what you guys are doing really is you're taking the whole process that a typical accountant would go to and you're really trying to figure out what is the bite size elements of each one of these steps that we could implement AI in and implement it within our platform. Uh, is this standardized across the firms or is it customized based off of the firm's workflows?

Speaker A: They all have to do the same procedures.

Speaker B: Amazing.

Speaker A: Yeah, I mean they all have like different templates and um, they may have slightly different methodologies, but generally they're all kind of calculating the same things. They have, they have to.

Speaker B: So the only variance here, really, here is that you're. Now as you mentioned before, it's, it's on their data, on their, in their cloud system. So the implementation is what's allowing you guys. So what happens when you release a new feature? It goes across all of the platforms and you're like, okay, here's a new workflow that you were supporting. Now it will just automatically show up in every single one of your clients workflows. It's that, that process doesn't change at all on your end, right?

Speaker A: No, I mean we're still uh, perfecting some of that. But you know that's the thing that launched in March was the platform, that back end platform that we can install into each one of our clients, uh, environments. The thing that we're working on now are the applications like the one I mentioned, that lease amortization application. That's a very small one, that's part of a, a bigger project that we're doing, but that's the type of stuff we're working on with the clients now. Um, what I would say though is like when we built this we didn't know what to call it. We just said we're harnessing these models and we're building these applications and but now the picture is, is, is forming and really what, what is Talon like how do we position ourselves? Who are we? And what we've realized is that we're an agentic framework. Right? We're just a very specific agentic framework. The name Talon means, comes from Middle English. It means to count or to count money. And the name a teller, a bank teller comes from the word tellen. And a bank teller was the guy that counted your money, go to the bank to count your money, uh, formed a financial transaction. Right? That's what our agents do. They are, they count your money, right? They're kind of in different ways, they're counting your money. That's effectively what they're doing. So what we realize is like we are building this specialized agentic framework specifically for financial services, specifically around counting around um, accounting around financial transactions. So all of our agents are financial agents that do some type of financial transaction kind of method or procedure. Right. Um, and once you have those agents that do these very specialized things like PDF extraction or model building or whatever, and you can build agentic workflows that string those agents together into um, a multi step process that uses multiple agents and multiple AI models in order to complete their workflow. And you do it in one category which is financial services accounting, you know, counting, you know, financial data. That is our, that's where we live in the world. So if you think about us, if we're kind of like, you know, the other agentic workflow platforms, uh, that are out there like a crew Crewai or an Autogen or a Landgraph, those are just horizontal plays that kind of go after any industry and do any Type of kind of, um, framework where we are specifically oriented towards financial companies. Financial.

Speaker B: I love that. I love that. That's incredible.

Speaker A: So, so stumbled into it. Like we kind of, we knew what we were doing, but we didn't really think. The pieces are just kind of falling into place as we, as we kind of go along.

Speaker B: That's awesome. Uh, so are you guys the only ones currently working with, uh, accounting firms? Or are you noticing other competitors already starting to emerge, or does this feel like a completely blue ocean for you?

Speaker A: When we go to the accounting firms and we describe how we're creating a platform, it's their platform where infrastructure, so they can build their own AI platform. And then we, we kind of build all these agents for them to do all these things most, almost all the time. We're the first ones that they've seen that does anything like this. And they know that the wave is coming. They, there is this, this fear, but also this, this greed that's driving them like, uh, on one side, it's like, we have to do something because this is going to transform our, our business and we have to be on, we have to be, uh, a leader as, as opposed to a laggard. Um, yeah, so that's kind of where, where they're going with their minds. Uh, so everybody wants to dip their toes in, or many, I shouldn't say everybody, but many of the firms we talk to want to get in. They want to figure it out and we give them an easy way to get in and figure it out with us giving them their own platform that they own, that will kind of, that will kind of, kind, uh, of evolve over time.

Speaker B: I love that. So, so we're getting close to the top of the hour here. What one advice would you give anyone currently working in the agentic AI space, especially trying to do enterprise sales? Uh, like what is one piece of advice you've learned over the years for all the other founders listening in this category who might be in a different industry? Like, uh, what do you, what do you think are some of the things you've learned that is helping you guys, you know, gain the early traction and early success?

Speaker A: You're saying, I'll give you two.

Speaker B: Okay, perfect.

Speaker A: The first one is when you're thinking about your business, be as close to the regulation as possible. Right. Everything starts with the regulation and goes from there. So figure out what, whatever industry you're in, there's some type of rules and regulations that are around that start there when you're building out your solutions. Because whatever your Client does, has to abide by these rules and regulations. It goes back to that whole thing about you do a procedure and you compare it to what the rules and regulations are and if there's an issue you raise it. That's a model that works in every vertical. It's a great place to start because that's the source of truth. The uh, rules and regulations are the source of truth. That's how you have to behave. So if you start there and build a bunch of agents around that, then you're going to be a trusted, a uh, trusted partner with that firm and you get going to get to do all sorts of other things with that firm. So I would say that that's the first one. The second one in enterprise is figure out how to use their own proprietary data. All of the uh, generative AI models are built on the massive waves of information coming out of the Internet. What it's not built on is the proprietary data sets held within the enterprises that you know, it's not available to anybody. So as a, when you go to that enterprise and you say I'm going to take your data and we're going to make it extremely valuable for you. Um, and now when you use your AI models with your data, you're going to have a huge advantage over anyone else that doesn't have the type of data that you have, uh, in doing what you do. So that's uh, again uh, that's useful uh, in any enterprise focused A.I. company.

Speaker B: Uh, I love that. So look at the laws, look at the rules, look at the regulation and uh, know that in and out, ideally even in a category, in a space, if it's changing a lot and has major impact, that's even better. Right. And secondarily uh, really make it very personalized to the organization's data set. So it's much more applicable and effective for all the members within the organization. I love that. Uh, Jason, it's been an incredible conversation man. I really enjoyed listening to your journey from graduating from school, getting your first job and uh, really dialing and listening into who the best PCs were and really being able to spot and find their emerging founders. And then from there, you know, going and working in private equity and really developing your pattern recognition skills over the years. From all the way from Internet to blockchain to even now AI. Uh, it's been a wonderful story and thanks for sharing that today. Um, it really showcases how you really look at every business strategically, uh, with data and also think about how you can potentially scale them.

Speaker A: Thanks.

Speaker B: Uh, for sharing your incredible things you're doing here at Talon as well, and some of the great lessons you've learned along the way and how you're really trying to transform the accounting industry and making agentic AI platform that can really enable the different firms to uh, really take advantage of uh, the challenges they're facing today, which is from the worker shortage and how uh, the opportunities for them to be able to provide more value for their clients. So if people wanted to, everyone listening in today, if they wanted to connect with you, what's the best way for them to be able to reach out? What should we put in the show notes?

Speaker A: Probably my LinkedIn. Yeah.

Speaker B: Yeah, perfect. Uh, thanks again Jason. It's really been great having you here. Uh, uh, looking forward to seeing what happens with talent and where this goes and uh, really excited for all the accountants out there and the future of accounting and what you've been able to build. Thanks for coming in today.

Speaker A: Hey, and Hamlet, I just gotta say, like, I really appreciate you, um, kind of um, providing a forum for people like me. Um, and I love the um, your guests all talking about growth in different ways and how they've kind of figured out their business model. So I really enjoy the uh, the, the content that you produce and I really appreciate you having me on the show. So thank you so much.

Speaker B: Thanks Jason. Thanks for coming on. Have a great day. Bye.