AI Is Making Mistakes in Finance for CFOs to Fix Data Problems and Improve Analysis with Nick & Jain

Welcome to another episode of Future Finance. I'm Paul Barmhurst, the FPNA guy, and I have here with me my trusted FPNA eye guy. Glenn, how you doing, Glenn? I'm good. I'm good. Good to see you, Paul. Now, you're not in your normal location. Where are you this week? I am doing what all smart business people do, and that is combining business trips with checking off family obligations. So I have a client in Greenville, South Carolina, which is where my in-laws happen to live. So I brought my wife with me, and we're doing client work and hanging out with the in-laws. And thankfully they have good internet. You never know when you're traveling, what we're going to end up with. But it feels like we're kind of cooking here. Got it. So we're mixing business with pleasure today. As every day, yep. Alrighty. Well, we have two guests with us. Glenn, why don't I give you the pleasure of introducing our guests this week? You know I hate reading intro's, Paul, but I'll, I'll devend and give this a shot. Hey, Nick and Dan. I'll go ahead and introduce you to our audience. And I'm really looking forward to talking to you guys. Our guest today are Nick Jane and Dan Catell, co-founders of Eagle Rock CFO, a tech-enabled fractional CFO firm that uses proprietary software to automate core FPNA work for its clients. These two have pretty different paths to the same company. Nick is a Harvard MBA out of McKinsey and Bane Capital, who then went the operator route. He was CEO at IdeaScale, a B2B SaaS company, where he nearly quadrupled EBITDA margins at 18 months. And he's been CFO at a logistics company and to e-commerce startup. Dan is a Stanford Chemical Engineering grad and also a Harvard MBA, who spent almost seven years at Prime Camp Management, which for those who don't know, it's one of the most respected and secretive fun shops in the business. Before co-founding Zimbado, a VC secondary trading platform that grew to over a billion and eight years ago. He's also an active broke equity investor in AI in Frontier Tech through GreenSans equity. Together, they launched Eagle Rock earlier this year to bring AI-powered analytics and fractional CFO services to the mid-market. Nick, Dan, welcome to Future Finance. Thank you for having us, Glenn. I think we'll start here on this question. We'll start with you, Nick, and then Dan. We'll let you add anything you want. So tell us just a little bit about yourself and how you started building Eagle Rock CFO how that came about. Sure. So Dan and I have spent our careers both being investors and operators. And AI started becoming a thing a couple of years ago. And as Dan and I finished up our last kind of professional opportunities, we connected last-- we've been friends for about a decade. We had never really worked together. We started talking to AI last summer and said, "Hey, there's something here. Where can we go build something really cool and exciting?" And an obvious area seemed to be, "Hey, AI is really good at analysis. And dealing with complex kind of multi-part situations. And we're pretty-- we understand how businesses make money. Let's stick those two together and launch kind of a AI automation business for the FPNA world." So yeah, they are running for our job, Glenn. Exactly right. Dan, did you want anything to that? Yeah, I mean, I think I've just been blown away by this whole process. Probably the part that scares me the most is just when I can be the most, the most. When I compare the tools that Nick has done most of the engineering work on our side, what he's built, when I think about comparing that to the chat bots, I interact with as a consumer every day. Oh man, the accuracy rate is just so much better. And Nick can go into the details later about why it doesn't hallucinate, but it's just a different world. What we're doing for the FPNA function than I see with CHAPGPT day to day. That's the world that I'm living in. And so my day job when I'm not doing podcasting, I'm doing AI implementations and they're all bespoke. Every client I have is a snowflake. We haven't productized anything. But also, so I do implementations, but I also do a lot of training. And I was working with our head developer, we're building an LMS system to train for a Fortune 500 company to train their team on how to use AI. And I talked to the developer and he said, you know, this may be the last thing we build because the agents are getting so good. If everybody can use them, they don't need us anymore. And it's super it and like you guys were saying, really, I would say with the last four months, because everybody's been talking, calling everything that they build in AI and agent in true agents, haven't really been a thing until very recently, where they could go off and do long range functions and all that. The people were just calling their modified chatbot and agent or whatever. And so, but we're just with whether it was the open claw experiment or even what, especially I guess what claw it is doing right now, what anthropic is doing with agents is just amazing. And I guess all that preamble to talk about this, and I'll throw this to either one of you. But if your tech handles the heavy lifting, I'm wondering, because we were talking a little bit before we came on error two, about deterministic versus probabilistic, what we, you know, the trial balance has to balance, what we know we can automate, sure, but we don't need generative AI to come in and weigh on it. So I'm wondering for you guys, how did you make those decisions? And what is your stack without giving away any, you know, secret sauce proprietary information? But what's the approach you guys took to this is hardcore just Python or math that we're doing standard here. And this is where we're putting AI in. And or are you just leaning more heavily into letting agents run? So no to the latter half, if you let agents run, they basically make stuff up or head off in the wrong direction, or my favorite thing, sometimes they miss the forest for the trees, and sometimes they miss the trees for the forest. And they do that consistently. So look, let me attend the tech stack, and then I'll go back to kind of some of the design principles. Our tech stack is really five pieces, only two of which use AI in any way, and not a, honestly, not a gentick AI, although a Dan and I personally do most of our work using agents, our client work in the software and technologies we offer to our customers tend to be, basically software solutions that lever J.I. rather than agents. So our tech stack is basically five things, right? It requires a data ingestion engine. How do we eat data? A little bit, very little A.I. is involved there. That's just traditional old school code. Secondly is compressing data, because one of the things you run into with A.I. is if you drop a giant spreadsheet or a hundred thousand word book into it, it starts hallucinating in very predictable ways. So you got to compress that hundred thousand words or that spreadsheet with four million lines down to something smaller. So we do a little bit of data compression. That's still old school technology. Thirdly, we do a little bit of tagging of the data, still old school technology, Python scripts, JavaScript scripts, whatever. Okay, level four, the fourth piece is where AI comes in. So we now have this huge chunk of, or the smaller chunk of data that is well categorized and organized. And then we ask AI, hey, help us figure out what's going on here. And we asked that very, very tactically across thousands of questions like, hey, what was, what are the major expenses? Hey, are there vendor consolidation opportunities? Hey, are there, is there a counting fraud? So rather than just asking, hey, what's up? We'll ask it an entirely vetted targeted list of thousands of questions. And that rely on domain expertise. And the fifth piece, it's not fancy. We leverage a little bit of AI. We make it look pretty by creating a presentation layer. But really like the cool stuff is what happens in the kind of levels, or phases one through four data ingestion compression tagging, and asking the AI to make kind of synthetic or synthesis decisions. The presentation layer is just make it look cool so clients pay for it. You guys are solving for what every finance team, whatever, whatever CFO's office is trying to figure out right now. And I guess that probabilistic versus deterministic and your answer made perfect sense. But that's what everybody, if you, if you've never done anything with machine learning, and you're not familiar with the technology, it just feels like magic and people are just thinking, oh, take whatever process I'm doing and just sprinkle some AI on it and boom, it's fixed. And that's, so I loved hearing your approach to it. And yes, we are in an AI era, and it's all driven. But the fact that you go back to leaning on just the fundamentals of deterministic coding, and this is, you know, this is what we're creating. This is the same for all businesses. Glenn, it's almost like two plus two supposed equal four. Yeah, unless you ask ten or two AI, right? Yeah. I had a roommate in college that had a shirt that said two plus two equals five for larger values of two. You also had one that said, if you can read this, you're overeducated and it was in Latin. And about ten others like that that he wore every day. So it made for fun. All right. So talk a little bit about how your kind of fractional CFO model works. Obviously, there's a technology behind it. There's a human. And so how are you managing the workflows? Where does the human step in, you know, kind of take us through a little bit of the process here. It gave you a sense for how the system works. Most of the time, the technology can handle just about everything. I think there's some reluctance. It's not just an FBNA. It's kind of everywhere. People to trust AI. Nick talked a lot about the data tagging and the organization of the data that we do before we ask questions. And that's really kind of the key to reducing the hallucination rate, along with some verification layers. But we still do have the human element there, essentially a handhold with the clients and get them comfortable with the answers that the system has figured out. Nick has told me that, you know, maybe one out of a hundred questions, the AI will still get wrong and we do do a little bit of human cleanup. But from what I've seen from the AI answer is it seems to be close to 100%. 100% correct. You know, we're doing a lot of extra work with our early clients just to make sure they're really happy. So there are some cases where a client has asked us to go and help find a loan or help buy insurance policies. And we haven't developed any AI bots that will take phone calls for us yet. So if we're talking with another party about financing or some sort of outside service that we need to get to the company, then we're still doing a little bit of that human monkey work ourselves. Who's calling our job a monkey work? (laughing) Aren't we all these hairless monkeys? Some are more hairless than others. - Yeah, I was gonna say Paul is not a hairless monkey. (laughing) - One thing I wanted to hit on though is you're go to market and I think you're hitting a pretty interesting segment. And I think this is just one of the segments you're hitting, but going after the fractional CFOs, I know I see so many of them out there, single, single, fractional CFOs, we're trying to come in and do a turnkey solution. And that means they have to be an expert in a lot of areas. And it could be, you know, maybe that, if it's a small enough one, maybe that fractional CFO is actually doing some basic bookkeeping and doing controller work and more strategic 13 week cash flows and all that. How did you nail down that market? What did you say that said, hey, fractional CFOs need to do it like this? - Well, let me take a crack at it. I think it's three things potentially. Number one, what do Dan and I actually have expertise on? Right? I don't have any expertise on, you know, let's say the creative side of marketing. So maybe Dan does, but like that's probably not be good area for us because we couldn't speak credibly or acquire clients or seem intelligent to them. So finance is an area where we have domain expertise. He's specifically on the FPA strategy side. Number two is where are people still doing a lot of manual work either on blackboards or on Excel or pieces of paper? Obviously FPA is another kind of category for that. And number three is where are their missed opportunities for value creation and how companies run themselves? So for example, in any normal company over probably five or six million bucks in revenue, no one is going through every single cost line at them every single day because people have lives to live, right? No matter how big your finance team is, they're not looking at every single line at them. Guess what, for an AI tool, that's a penny of cost, right? So as you think about the conflicts of those three factors, what are we good at? Where are people still using legacy tools that can be empowered or made better or using better tools? And where is there a lot of real hard value creation left in how companies are leaving money on the table in their operations? And that's where we came together and said, look, FPA is an area that is ripe for automation and AI powered tooling. And it's an area that every company really has, right? Like almost every single company has an FPN and department, CFO or CEO or somebody who is basically looking at the numbers to try and figure out how to make the company better every single day. And that's an area that Dan and Dan are passionate about to kind of add icing on the cake. And I guess the follow up on that, if you're talking about fractured CFOs, I've lived this world and I've seen it a lot. That means they're coming into companies that I've got quip books or maybe zero and they might have HubSpot or some other pipe drive or something. They've got some disparate systems, data's a mess. If it's a founder led company, their chart of accounts could be that like the quick books chart of accounts and there's sort of a blend between cash and accrual accounting. I mean, that's what sort of shingles small firm CFOs are dealing with. Was that part of your build? Was how can we help them clean this up? Because it's always so painful when you first come in and you realize, oh, that's going to take me 60 days just to understand the chart of accounts here. - Well, I think that's exactly right. If you look at a big company, they already have the team of analysts and data science people who can help them get the answers. But when you're talking about that single shop, that PNA, fractional CFO guy, he's got to figure out how to do it all himself. So giving him the tools or her the tools to do this the way that a big company could as a single individual, I think is a big part of our mission too. - Makes a lot of sense. But I'm curious, what do you guys do? How does the tool manage that? If the chart of accounts is a mess, do you go work with the company to clean it up or how do you deal with that? 'Cause we all know in this small space, it's rarely clean from day one and Nick is laughing. - Oh, oh, Chuck, so yes, 100% true. We actually, one of our, you know, look, our tool does a bunch of things. One of the features that it does is it looks through all your data and says, hey, here's where it's a mess. And sometimes it says it's a mess because it's wrong for technical reasons, like one plus one is equal to three, which is we know an accounting shouldn't be true. And sometimes it says, hey, this journal entry kind of looks fuzzy instead, it should be over here. And sometimes it says, hey, your chart of accounts is a mess and you need to go fix it up. So generally we are not going in there unless it's a quick task and actually, you know, changing the chart of accounts. That is our clients work or their accountants work or their controller or their fractional CFOs work. But the good news that our technology does figure out when it is either technically or qualitatively an area for improvement. Because as fun as fun as it is for one of our clients, we said one of the biggest areas for value creation for them was actually just getting their accounting right because they couldn't figure out their P&L correctly. Well, we had one little, we saw equity going, owners equity going through revenue on a company. - Thanks, perfect sense. - Hey, yourself, then then put it back into the business and call it revenue. I love it. That's my new model. That sounds like, isn't that like a Ponzi scheme basically? (laughs) That's something like what Enron tried to do, but in a more sophisticated way. - Another question is kind of what to ask is, you know, we noticed you guys are doing some white labeling of the platform. So obviously that's quite a bit different than running a consultancy business. You have some human stuff. So how did you come up with the approach? What's kind of driving it? What's the uptake Ben like? Because it sounds like you're offering the software, obviously to other fractional services, but you're also offering some of your own fractional services with the software. So how do we walk us through all that? - Certainly, from the perspective of what we've built, it can be applied both to a company that has a CFO in place or to a company that doesn't have a CFO. So the white label is our version for fractional CFOs, for full-time CFOs. For somebody who wants to use this without our hand-boulding. And I think a lot of good finance people are capable of doing that. There are some people who just don't want to deal with the technology directly. Technology is not their thing. And that's where our consulting service comes in. And we can either work directly with the existing CFO of a company or we can work directly with the CEO if they're kind of at the stage where they're thinking about bringing out a CFO, not ready for a full-time person. So maybe it is kind of a fractional CFO consulting in its own way and not service. Or it could just be kind of the early stages of starting to implement some of those things as the company is scaling over time. - And what's your sense of, I know it's early days still, but what's your sense of how much time can be saved, how much more, you know, the real value proposition of this, whether you're going into a firm or through, you know, white labeling through other fractional CFOs, it's gonna be, what's the time savings, how faster are we closing the minds, what insights do we not have before? Like what's your sense of, like, cause again, spend a lot of time in this space, I just coming in where there was no CFO before or there was a glorified bookkeeper called, like head of finance or something, and it's just kind of a mess and getting to the other side. I think every, Dan, I think to your point, there's always that precipice of, we kind of need a CFO, we can't yet afford a CFO, what are we gonna do? And I think that's probably the, maybe a sweet spot for you guys, but what's your sense of the company that brings you guys in, what they're saving and what they're having that they didn't before? - Sure, I think our value proposition happens in two ways. Number one is if you have someone actually analyzing the book, so ignore the technical side of doing the accounting and closing the books, we do a little bit of work there, but really are focused on the FPNA side. For that FPNA side, if the person has time, there's probably spending somewhere between 20 to 50 hours a month, just looking, peeking around the number, saying, hey, I can create more value here, I can adjust pricing here. I believe our tool cuts that 20 to 50 hour exercise down to a few minutes of human time tops. So it is 20 to 50 hours per month of time saving strictly. But here's the big kicker, and a lot of companies, they're, people are just too busy managing the day to day, so there's not somebody who's stepping back often, and just thinking about the numbers and looking through the data and doing analyses. And for those companies, this is just a purely value additive service, where you don't have somebody trying to like go around optimize your costs or think about vendor consolidation or pricing, or whether you have all the right insurance policies, because people got stuff to do, and that's an area where it's not a time saving, it's just directly accretive to your bottom line. - Makes total sense. Well, Nick and Dan, we really appreciate you coming on, and very, I think you guys have hit a sweet spot. I wish you guys the best of luck, because I know there are so many fractional CFOs who are trying to figure this out themselves, and hope their clients do. So I know you got your own consulting too, but I think there could be a real sweet spot in having a tool that fractional CFOs could come in and be armed with. So super interesting to see how this shakes out for you. - Thanks for having us on. We really appreciate it. - Thanks for having us. - Thanks for joining, really appreciate it, Nick and Dan, and we hope you guys have a great rest of your day and good luck building. I know scaling is always an exciting and scary time at the same time.