There may be errors in spelling, grammar, and accuracy in this machine-generated transcript.

Isaac Heller: Hey, everyone, this is Isaac. I'm CEO at Trillian. And today on AI, accounting, intelligence. We'll be talking with real people about real things related to AI and the impact on the accounting industry. Stay tuned. All right. Hey, everyone, this is Isaac Heller with the Accounting Intelligence podcast. I'm pretty excited today to have a fun conversation with [00:00:30] Chris Bernardi, CEO, co-founder of Hudson Labs, formerly bedrock, for all you fans. Uh, Chris, it's good to talk to you today.

Kris Bennatti: It's really great to talk to you, Isaac. Excited to be here.

Isaac Heller: Yeah. I think we're going to get into some fun conversations about entrepreneurship and maybe some like philosophical accounting and audit world conversations as well. But before we jump in, um, Chris, I saw like last week, I think, or the past few weeks, [00:01:00] like an article in Forbes or somewhere of the of the woman who predicted the supermicro kind of failure, which was in the news recently. What is supermicro and how did your company or your team kind of predict it? What what went on there and how did you get called out? Pretty big.

Kris Bennatti: Well, Supermicro computer is a publicly listed company that was recently added to the S&P 500, I [00:01:30] believe, uh, but they've seen insane growth, uh, in part driven by the AI hype machine. Um, and also what looked like real numbers. Uh, but at the same time, they were very clearly disclosing some very abnormal business practices involving a complicated related party Pretty network. And to anyone [00:02:00] who has experience with fraud, looking at how fraud is perpetrated, looking at how earnings management. You know, I don't I don't want to say fraud was actually happening at Supermicro. But when you look at how aggressive earnings management is perpetrated, what that generally looks like, the way that Supermicro computer was set up and was set up in a way that looked like it was meant for that, uh, to be, uh, to be frank. And [00:02:30] we have an AI driven platform that assesses both forensic risk, the likelihood of an SEC enforcement action related to fraud, and also highlights concentrated related party risk. Supermicro computer was flashing red on all of those screens for a very, very long time, for very obvious reasons. You know, I think any forensic accountant going through those financial statements would have come to exactly the same conclusion that our models came to. And we publicly [00:03:00] highlighted, uh, those risks multiple times over the years. And then when the Hindenburg report came out, the UI it sorry, a short report came out on the company by a famous short seller named Hindenburg. Uh, and the stock price cratered. Then their auditor quit or resigned, um, and didn't do it in a way that was particularly polite, uh, which reinforced the allegations of the short seller. Um, and I [00:03:30] published something on LinkedIn that just said, hey, look, guys, you know, I this isn't news. Uh, and that LinkedIn post got picked up by Forbes and later by, uh, CNBC.

Isaac Heller: That's awesome. Chris. And, um, you know, for the record, I've followed Chris and, you know, Hudson Labs and Bedrock for a while. It's a small community, but, um, you know, following a lot of your stuff on, on Twitter and LinkedIn. So we'll get into that. We'll also get more into the entrepreneurial [00:04:00] journey and what your company actually does. But, you know, it's funny just to start with Supermicro. So you mentioned Supermicro is an AI company, which is all the rage right now. Um, you're an AI company too, which is pretty cool. So it's like AI versus AI. Um, but so there on.

Kris Bennatti: The hardware side.

Isaac Heller: For the hardware.

Kris Bennatti: Side, yeah, they do. They make the servers, data centers, that sort of, uh, area of the, the AI infrastructure.

Isaac Heller: Very nice. Okay. So we're so we're almost all [00:04:30] powered by AI in a different way these days. Um, but it's interesting because the way you talk about Supermicro is almost like, well, hey, it felt pretty obvious that something was going on here, right? And, you know, I, Chris, from my point of view, was able to see that with a little bit of AI and maybe even with some, uh, basic reasoning yet a Big Four auditor. A layer below was not able to identify that. And so, you know, there's there's a [00:05:00] question here because, you know, the big four is is well regarded and, you know, earns a lot of money maintaining the, I guess, balance of financial reporting and language of business. Like is it kind of do you think that accounting and audit is kind of in an equilibrium right now where they don't really want or need to kind of catch the fraud or be overly scrutinous on the financials? Was this a one off an anomaly, or are [00:05:30] there a few more super micros, you know, lurking and a a, a I would call you a casual observer versus an auditor who knows the financials. A more casual observer, um, could be able to catch it over an auditor. Is this a one time thing or are we looking at more? And is it just in this complacent world right now For us.

Kris Bennatti: So I'm going to actually come in defense of Ernst and Young. Okay. So [00:06:00] I'm possibly a big fan.

Isaac Heller: Sorry. I just want to say also.

Kris Bennatti: Big.

Isaac Heller: Biggie. We had truly and work with you. We we we love you. Why? They just happen to be in the news on this one. And you know, all the big four get their shots. So we are both defenders of E. But go for it.

Kris Bennatti: No, but more specifically, I want to defend the audit profession because I think EY did their job, which was to enforce transparency [00:06:30] at Supermicro. You know, obviously EY wasn't there going, hey, this company was previously subject settled with the SEC and continues to have aggressive sales practices, but we don't pay EY to monitor the aggressiveness of the sales practices of their Uh, their client. We do pay them to make sure that the related party note is robust and detailed. And it was, you know, I think that the risk [00:07:00] was appropriately communicated to investors. Uh, the fact that investors chose to ignore it is not a wise problem. Um, and I actually have, um, you know, I think if we're trying to fix fraud or reduce fraud, I don't actually think that, uh, improving audit or changing the ways that audits happen is going to go a long way in achieving that goal. Um, [00:07:30] and in fact, one of the first articles I wrote that got attention in this space was called Why Fixing Audit Won't Fix Fraud.

Isaac Heller: Very interesting. Okay, good. So, um, you know, it sounds like, you know, from your perspective, ey did its job to bring the the risks to the surface and where you sit. Chris is almost more as a peer to the investor community and the the folks that are looking and researching these companies and trying [00:08:00] to look for signals, which is really interesting, and it kind of makes us complementary in our in our entrepreneurial past, because we truly and we're like trying to get into the data at the corporate audit level and trying to like, you know, help auditors and accountants become more efficient, which is actually getting into the data, whereas you're at the layer of looking at those numbers and interpreting, you know, whether they're accurate or whether they tell a kind of a different story. So either way, very interesting and very cool that you and your team were able to [00:08:30] pick up on it. So for anyone listening, they might not even know, uh, that much more about you or Hudson Labs. So, I mean, before we keep going, I tell tell us a little bit about yourself, your entrepreneurial journey really to to bedrock and now Hudson Labs. What's it been like?

Kris Bennatti: It's been fun.

Isaac Heller: You're a you're a KPMG alum.

Kris Bennatti: I am a KPMG alum. Yeah KPMG Toronto. I worked at in mining shout [00:09:00] out mining company.

Isaac Heller: Shout out mining to the mining accountants in Toronto. Big shout out here.

Kris Bennatti: Yeah yeah yeah yeah. Uh, do you have an accounting background?

Isaac Heller: No. Busted. I have an accounting passion. I fell in love with my GAAP and IFRS classes at a nighttime MBA program. Right. Shout out to Suresh, my professor in Dallas. Um, I do not have a CPA, but I did an ASC 606 overhaul for a pre-IPO tech company. And [00:09:30] I feel like that that gave me some kind of badge.

Kris Bennatti: That's fantastic. Well, I also no longer have a CPA designation. Um, I chose to to stop paying my dues. So, you know, we're the same. I also often feel like at this point, my, you know, it's been it's been a fair few years here since I actually worked as an accountant. And my actual accounting expertise, [00:10:00] it can barely be considered CPA level at this point.

Isaac Heller: Very nice. Well, you could get you could get back on track with 149 hours left after this, uh, this podcast. It would be. It would be nice if you could, like, transfer or sell your CPA license, like those New York taxi medallions, like, I would just, you know, I would just take that. But anyway, so so you're you're not at CPA level anymore. You at some point you [00:10:30] how did you get like twist and turn from KPMG accounting CPA to now CPA to CEO.

Kris Bennatti: Yeah. Um, so I was a CPA working at KPMG. I quit More or less as soon as I got my CPA. As many people do, um, which I think was surprising to everyone because I did I did actually enjoy the job, which I seem somewhat unusual these days. Uh, but I had already had a background in stats [00:11:00] and econ, um, and wanted to get more involved in data analytics, data science, uh, which was emerging at the time. Uh, I went back to school for a master's program one year and learned a lot during that program, but also did a lot of self-teaching around learning to code, learning Python, and got a job as a data scientist coming out of that one year master's, which was great. [00:11:30] Uh, and also got a job teaching Python programing, uh, at an AI bootcamp, which was was really fun, but it was through that role as a data scientist. Because I was working at a corporate governance advisory firm. I had the background, the corporate background, where I really understood securities filings. I really understood the corporate space well. So because of that joint background, I ended up specializing in corporate text processing. [00:12:00] So Pre-language model, natural language processing, um, primarily AI and got a lot of attention for that work. Uh, despite being aware of all the incredible things that were happening with language models and AI at the time, and the way I ended up founding this company was, you know, really looking at the gap between what I was [00:12:30] doing that was considered state of the art and being aware that Grammarly existed and Autosuggest existed in the state of the art looked very, very different. And there was a big opportunity to take take the language models being used in consumer technology and apply it to my space.

Isaac Heller: That's awesome. And I think later we should, if we have time, we should get into some AI. And, you know, kind of what you're saying is there's these big broad AI and now open source LLM models, and then there's these nice little slices of the world where [00:13:00] you need more specialized and dedicated kind of AI expertise. Right. Even within the accounting world, there's dozens or maybe hundreds of different slices where you need to kind of go deeper, but we'll save that for another time. Just trying to think. So you're in you're in you're in Toronto, right. When you when you go from KPMG to data science, you're, you're in you're in New York today. Fair to say.

Kris Bennatti: Yes.

Isaac Heller: Okay.Got it. And so in Toronto has like a huge AI community don't they. Like [00:13:30] Waterloo there's some university that cranks out the the the. Is that right? Is there like a is it like an avant garde like early AI I community in Toronto that kind of you gravitated towards. Like, was there an entrepreneurial. I led community that pulled you in. How did you just jump in?

Kris Bennatti: Yeah, no, that's exactly right. Toronto is an incredible place for for AI and AI research. And that's how I met my co-founder, Suhas Pai, who is the LLM guy. [00:14:00] Uh, someone stopped him on the street the other day and went, hey, are you the LLM guy? Really?

Kris Bennatti: Uh,

Isaac Heller: That's like his his name.He's the LLM guy.

Kris Bennatti: That's his. He's the LLM.

Kris Bennatti: Guy. Yeah, I mean, he Suhas.

Isaac Heller: Yeah, we'll have to.

Isaac Heller: Yeah. We've got an LLM guy at truly and we'll have to you'll have to have you guys talk. But that's awesome. So you and Suhas meet in this cool community of AI lovers.

Kris Bennatti: He was leading the language modeling natural language [00:14:30] processing, specific stream of the AI research community. Uh, there's this research community called Aggregate Intellect. Socratic circles. It's a bit of a mouthful, and I. Yeah, it's it's a great community. A lot of people just get together, read through research papers, etc. and Suhas was leading that. And that's how we originally connected and how I was a lot more involved in [00:15:00] seeing and experiencing the really incredibly cool, actually state of the art projects that other people were working on while I was still a relatively junior data scientist.

Isaac Heller: Interesting. I'm curious, like there's there might be some people listening here who are starting their accounting career, going into audit or maybe in audit and thinking about the next steps. Um, thinking back to when you were in those KPMG auditor shoes, do you think that it's it's different, much different today [00:15:30] in terms of like, I don't know, the choices, the opportunities. Meaning would you would you tell people now get out of audit orbit faster. Spend more time in audit. Specialize differently. Go to consulting. Don't go to consulting. Like, have things changed? Now, a few years later.

Kris Bennatti: I was talking to a partner at one of the big four yesterday at asking all sorts of questions about how how audit has changed. But frankly, Isaac, [00:16:00] I don't know. I feel pretty, pretty disconnected from the actual day to day, uh, lives of accountants and the opportunities out there. It does seem, you know, from where I'm sitting, high level accounting expertise, really smart accountants. I is not coming for your job. So it's a great place to be specialized. I think there's going to continue to be really great, uh, [00:16:30] opportunities. And because accounting and audit is so painful, you know. If you like it, if you love it, if you're into it, stick with it.

Isaac Heller: There you go.

Kris Bennatti: That's that's my my piece of advice.

Isaac Heller: I like it so. Nothing new, nothing new under the sun. Just AI is the next technology. Excel was technology. You know, databases is technology. And AI is some level of technology that will kind of settle and still maybe [00:17:00] need a lot of that, um, specialization, I guess you could say. But Chris, I think it's awesome that you learned Python and data science. I was talking to Nick Boucher. He's a finance influencer. Shout out to Nick like AI Finance Club. Really cool guy. And he's now teaching courses Python for accountants, Python for CPAs. And it's getting a lot of traction. And it's almost like more of a response to AI. Right? Like the [00:17:30] AI is here and you could potentially be more powerful leveraging the AI. If you could just turn the corner and learn to some Python and data science. Do you think that holds meaning? You know, do you think that there's a stronger and stronger demand for Python experts that are CPAs as well? Or again, same. Same as before?

Kris Bennatti: Oh, I think it's hugely beneficial to be learning to code as an accountant and most other professions, but I don't think that's necessarily relevant [00:18:00] to AI per se. You know, you can achieve so much with web scraping, more powerful at data analysis, being able to do work in pandas, which is the Python equivalent of Excel, can really accelerate your your work and your role. And back in the day where I had people from KPMG coming to me going, oh, hey, you run an AI company now, how should we use AI and audit? And [00:18:30] I was going you have implemented zero automation software. Why don't you just start there and then think about AI? You know, when I left audit, we were we're still looking at an invoice and looking at this sheet over here and taking this number and putting it over there. Why not just automate that with very simplistic software before you think about advanced state of the art AI.

Isaac Heller: Interesting. Yeah. And that's I think that's a lot what we we [00:19:00] felt it truly and and that exact use case we have like a data match automation. Um, okay. Interesting. You mentioned first of all I heard pandas. Is that like a that's the python. It's a.

Kris Bennatti: Library.

Kris Bennatti: Within Python. It's the data frame library. So it's Excel in Python. So you're working with another version of a spreadsheet, but it's just a faster and more powerful and generally [00:19:30] connected with your data source directly so you can do more.

Isaac Heller: Got it.

Isaac Heller: And for those of you who are afraid of Python like me and jumping in. Knowing the lingo like pandas is really good because you can hang out with a wider range of people and sound dangerous or smart. So maybe we'll glean a few more of those along the way. But you mentioned Python. You you mentioned web scraping, a few other things. I don't know if you told us what Hudson does, [00:20:00] like if if I'm listening right now or watching, like, I'm going to check out Hudson, but what should I, you know, what should I want to do with it? Hudson labs. Right. Hudson labs, com.

Kris Bennatti: It's hudson-labs.Com.

Isaac Heller: Amazing. Okay. And what do we do with it?

Kris Bennatti: So we specialize in finance specific language modeling or LMS, where we're extracting and organizing Complex information in securities [00:20:30] filings and other types of financial disclosure. So one example that's more on the accounting audit side is where the only complete real time provider of Sox weaknesses, where the only complete real time provider of going concern warnings or, you know, bankruptcy risk warnings. I always say that because my audience half the time doesn't know what a going concern warning is. Um, uh, but we extract and organize [00:21:00] unstructured data in financial disclosure and make it accessible to corporate stakeholders, which would primarily be investors, hedge funds, asset managers. But we also work with insurers, plaintiff class action firms. Et cetera.

Isaac Heller: Got it. Awesome. So I mean, look, if, you know, there probably could be some, you know, creative or innovative use cases for, you know, advisory firms that are looking to understand, you know, control weaknesses, [00:21:30] material weaknesses, basically signals, right.

Isaac Heller: For for clients.

Isaac Heller: That may need help.

Kris Bennatti: Yeah, that is where we do overlap with accounting professionals is we do help accounting software providers, risk advisory professionals with business development. So if you, uh, specialize in, um, business combinations or transactions accounting, we can give you a real time feed of every [00:22:00] single public company who's reporting having issues or a stocks weakness specific to business combinations.

Isaac Heller: That's awesome.

Isaac Heller: Very interesting. Okay. So so, you know, we'll have to we'll put some link to to Hudson after this. But you know I learned about Hudson by the way if I say bedrock for anyone listening it was previously bedrock. So I guess I'm like an OG fan. Is that fair to say? So I would you know, I would go on Twitter and I would, I would follow I would be interested in some of [00:22:30] the stuff that you guys were putting out and you specifically. And I think that's like a I'm starting to learn myself like it's a big part of entrepreneurship, which is, you know, building the brand and using your kind of unique insights or skill sets to, to to shout, right, and tell people what you're doing and what you think and stuff like that. I mean, did you ever did you ever think about building a brand for yourself or curating this, what they call like influencer [00:23:00] baseline or whatever? Or did you just do what you did? And it kind of went from there?

Kris Bennatti: I personally have not thought I would prefer not to be an influencer. I if if humanly possible. Yeah, but I do recognize that it's an important part of growing a business, etc.. So most of what we did early on was coming from the company rather than from For me and or coming from our employees rather than than from me. But you're [00:23:30] right that putting out our thoughts publicly and being very open with sometimes spicy takes was huge in getting early attention. You know, our first customers were all from a Substack interview someone did with me about whether audits auditors can find fraud actually interesting. And and, you know, all of our subsequent press generally came not from us [00:24:00] pitching stories, but from journalists seeing the work that we did and wanting to write something like that, or write a follow up piece on it or something to to that effect.

Isaac Heller: Interesting. Yeah. I mean, that's a I think that's a good point. I think I think some of the best, um, kind of content, let's call it, is from people that aren't trying to be influencers, because it kind of self-selects for stuff that you just find interesting [00:24:30] instead of, you know, kind of forcing one to push out ideas every day, which may dilute, you know, the the ideas or distill the ideas. Um, but like at this point, I mean, you're, you know, you're CEO of a company, your figurehead, like, do you have to remind yourself once a week to put something out there, or are you just straight up if I have a good idea that day or 2 or 3, I'm going to share it. Or is it like an alarm? Is it like an iPhone alarm? [00:25:00] Like, you know, look for fraud today and post about it Tuesday, 3 p.m..

Kris Bennatti: No, frankly, we had more of a content social media focused strategy in the past, and at this point, I it's not a core part of our strategy. Maybe that will change, but I'd love to hear your experience. I feel like that's something that you do quite well. I mean, you posted something about me that got an extraordinary amount of engagement.

Isaac Heller: Did it? I, I [00:25:30] don't know.

Isaac Heller: Um. What can I say about this? I agree with you that I'm not trying to be an influencer. Um, I have, um, a couple colleagues that are marketing gurus that encourage me to share my thoughts and post, and so that that helps, um, boost it. And, um, sometimes it's just fun, like, like I just realized it's fun. So, like, so what we're doing now, like, I accounting intelligence, I didn't, you know, [00:26:00] wake up and say I wanted to do a podcast. It was like, hey, Isaac, this would be a really good idea to do it. And now I got to meet you, and I got to meet Larry, and I got to meet, you know, all these other cool people and have conversations with them before or after, stay in touch and stuff like that. And so in retrospect, like, you know, kind of putting putting myself out there, you know, to try to help. The Julien brand has been positive for me because I get to, you know, meet some interesting people or [00:26:30] you'll get a LinkedIn message. Hey, I saw you talked about, you know, Supermicro or whatever. I love that. And then you make a friend. So yeah, it's kind of been like, like a, like a rewarding process. Um, and but but I mean, I have I have no science to it. In fact, I will, I will post things and then delete them an hour or two later because I just like, think about it again. I'm like, that's that's stupid. Don't you [00:27:00] know that's nobody cares about that. So, you know, whatever, whatever makes it is kind of a crapshoot, I guess you could say.

Kris Bennatti: Yeah, that is the part that I find challenging is it tends to take up like how much headspace is taking up versus the value, right? And social media has a way of taking up more headspace than you plan for. Like I [00:27:30] go, okay, I just posted this. I'm not going to check the angry things that people are saying until tomorrow. Um, and, you know, I think one interesting part of our marketing strategy in the past was in order to get people to engage with our posts, we would say controversial things about a specific stock.

Isaac Heller: Right.

Kris Bennatti: And the point [00:28:00] is to drive engagement, right? We know that that's going to elicit a very positive response from a select group and a very negative response from a select group. And the negative response is going to include people who are looking at your Air LinkedIn there, googling your name there, trying to find dirt on you and see the meanest thing that they can possibly come up with. Which in theory is simply amusing, but it does take up a bit more head space than just sending [00:28:30] out a whole bunch of of email or LinkedIn outbounds.

Isaac Heller: So can you think of like one time where someone sent you a DM like you posted something and maybe it was accounting or fraud, someone sent you a DM or post and you were like, whoa, I didn't, you know, I didn't I didn't get into this business.

Kris Bennatti: There has been a few the highest profile one, this is so high profile that Harvard Business School reached out to write a case study about it.

Isaac Heller: Okay. [00:29:00]

Kris Bennatti: And I didn't realize that it was so notable, but, um, what happened? Uh, there is a former hedge fund manager that has a very large following on Twitter. Now a large retail following. Even though he started out as sort of a more short focused hedge fund manager, he now is really excited about specific stocks [00:29:30] and gets in very bitter fights. So we knew that this individual would a has a huge following and B is easy to engage.

Kris Bennatti: So we.

Isaac Heller: Got Dan Loeb is. It. Uh no it isn't. Okay.

Isaac Heller: Dan's one of our investors. So I want to make sure I wasn't, uh. Uh, good.

Isaac Heller: So some. It's not an.

Isaac Heller: Activist. Got it, I understand.

Isaac Heller: Okay. Yeah.

Kris Bennatti: But he no longer [00:30:00] runs a fund. This person.

Isaac Heller: Okay.Got it.

Kris Bennatti: yeah, I so we responded to a positive tweet about the specific stock analytics and highlighted some risks at that stock. Uh, he we expected to him to get a little bit mad. He did get a little bit mad. That tweet did. Well. What I didn't expect was he would get in touch directly. [00:30:30] He ended up getting my number. He ended up calling me. Um, but when you get get on the phone with me, it's so much easier to get mad at someone on Twitter. You get on the phone with me. I'm a reasonable person. You know, he that conversation deescalated very, very quickly. And we ended up having this nice cordial relationship for, for a long time where we, you know, text each other. And he'd tell me I needed to update my models because they were all wrong, because they were rating some of the [00:31:00] stocks he liked as hot risk. Uh, but it was very cordial. And then.

Kris Bennatti: For a while.

Kris Bennatti: And then one day, the stock that we had been critical of or not necessarily critical, just raised some risks. The stock it's a very volatile stock. So the stock price ripped. It went way up. And he all of a sudden hadn't heard from him in a while. Retweeted our original post about the risks. [00:31:30] Um, saying something along the lines of best tear apart your models, darling. Something mean? Yeah. Sent me a bunch of DMs.

Isaac Heller: You know that that old darling snarky comment, you know? Yeah.

Kris Bennatti: Uh, sent me a variety of very nasty DMs. Uh, and Harvard Business School, uh, I, you know, I was looking at this [00:32:00] sort of smiling to myself, uh, because it was incredible free publicity. Um, And he's made so many enemies, and all of his enemies started following us on Twitter, which are mostly people we want to sell into. So it all.

Kris Bennatti: Positive. Uh, but Harvard Business School, I guess, saw this high profile tweet criticizing us and decided it was a risk to our business and wanted to see [00:32:30] how we overcome that risk, etc., which I found found funny.

Isaac Heller: This is I mean, this is hilarious. Like, you never know what you get into. Have you read the book Moneymen about the Wirecard, uh, scandal or.

Isaac Heller: Watched the Netflix?

Kris Bennatti: I should, I should, I haven't.No.

Isaac Heller: I'm trying to think who. It's in my office. One second.

Isaac Heller: Uh, the author is.

Isaac Heller: The author is Dan McCrum. Dan McCrum was a was an ft. Financial times reporter. He is he's a he's a great reporter. And just for those of you watching, [00:33:00] uh, it's the movie. The book is money Men. There is a Netflix movie about it, and it is about the Wirecard scandal in Germany. You know, Wirecard was, uh, you know, hot fintech company up and down. And there were just a lot of people on both sides, short sellers and bold investors. Talking about Wirecard, and the reason I mention that is because if you read the book, it like goes into Russian espionage and like, you know, armed men confronting [00:33:30] you at the door and stuff like that. And so that's, that's an extreme. Okay. But it is kind of funny to like, uncover this world of, of, you know, people take this stuff, uh, pretty, pretty intensely, but I agree, I mean, social media is like another world, but when you I mean, when these things happen, do you, do you step back, maybe just just moving on to like Hudson Labs and in general, like, do you step back and say, like, what is my purpose with Hudson Labs? Like, what am I really trying to [00:34:00] do? Am I trying to drive more transparency? Am I trying to, you know, build a successful business and create some jobs and have some fun. Like what's what's the mission that that anchors you?

Kris Bennatti: Oof!

Isaac Heller: Yeah.

Kris Bennatti: Not a hard question at all. But no, I this is something that we talk about Celeste and I weekly, you know, what's are we are we achieving our mission? What's what's the most important thing for us to be focused on right now? [00:34:30] Uh, but I am so interested and concerned and excited about, uh, how companies communicate to stakeholders and how that information is consumed. I think there's a very interesting balance between, uh, the burden of disclosure and [00:35:00] the burden of preparing that information. Um, and it can definitely become too much and it's costly. And given that we are investing so much, it's so important that this information is actually making it in front of the people who are supposed to be using it to make decisions. And a lot of people don't realize how poorly that process is going right now, and how much information [00:35:30] is getting lost and isn't necessarily being consumed by the investor community or other corporate stakeholders. Despite the fact that we're spending so much time and effort to make sure that everything agrees and looks perfect. And that note 22, uh, is compliant. Uh, and I is such an incredible opportunity to extract, organize and simplify what [00:36:00] is a very Complicated and hard to understand disclosure regime. And I also has the opportunity to go beyond just what's in the SEC filing, but also collate and collect all of the other information that's being communicated in a non-regulated or formats as well, which is hugely necessary. And people always come to me going, oh, I is such a huge risk. [00:36:30] We have not innovated in this space in such a long time. There's so much opportunity. The risk is not in using AI for extracting and organizing and, and, you know, doing what you're doing on the internal side, the risk is in not doing it because we're in a situation now where there's so much complexity and we've seen a lot less innovation, a lot less focus [00:37:00] in this space compared to consumer technology, for instance.

Isaac Heller: Interesting.

Isaac Heller: I like it's kind of a it's kind of a nice arc in terms of the mission of Hudson, because it starts out like this stuff is interesting to me. Like this area is interesting to me. I'm passionate about it. And then you start to dig in and you're like, people are spending money and there's cost and there's risk and and there's appetite. And then you start playing with the technology. I [00:37:30] write AI is a generational technology, right? It's there's things like this every once in a while. And this is one of those things that has some, some superpowers and you start playing with it. And then it's like this angst of like, what more could I do? Why isn't everyone else using it? Like, okay, I know there's risks, but I've looked at the risks and I agree those are risks, but those aren't risks. It could actually be very, very valuable. So it's kind of like the arc of I would say a lot of, you know, entrepreneurs and innovators these days is it just starts like [00:38:00] you're having fun. Or at least for me, like I it was this ASC 606 thing and I was like, this is crazy. We're reading all the contracts and we're putting in the spreadsheets, and then the auditors aren't checking it, or having to pay more to the auditors for a new standard, you know, because they have to audit outside of the ERP. Why doesn't it go into the ERP, you know. And you start asking questions. And then there's IFRS and ASC and Sox and ESG. And these questions keep coming up. Um, and so I don't know it's just it's just self fulfilling. But [00:38:30] it's you know, it all goes back to just having a little fun. But I, we would, we would.

Kris Bennatti: Do you feel still feel like you're just having a little fun Isaac.

Isaac Heller: Oh, yeah. I mean, I've lost some hair over the years, but, like, I'm having I'm having a lot of fun, you know? That's great. Um, I think I think that's the most important. And, and I would say the other cool thing about AI is AI has been like this, I don't know, centrifugal force or like this torque. And what I mean by that is, [00:39:00] you know, you and I, we've been at this for about five years. I saw on your your profile like when at least for, for truly and when we jumped into AI this was 2019. The first use case was related to OCR, right. It was it was OCR for documents. I'm sure you and Suhas have been there, done that. And that was cool because of the new compliance standards which brought contracts onto the balance sheet or made them more interesting, you know, and contracts, meaning like the the elements of contracts [00:39:30] became more interesting for accountants through um, and leases and, and you know, Sox and all these things. And so OCR was like the talk that propelled us for a couple of years. And then 20, 22, uh ChatGPT splashes and there's like this talk about gen AI and like text based chat and, and that gave us like this other, you know, exciting thing to chase. And now Agentic [00:40:00] is is kind of bubbling up. And so we've we've looked at this a while, but new agentic models, theories, advancements or really the advancements of the LMS have enabled Agentic type work. And so we're you know, we're in the workflow world. So I'm saying it's like a third kind of talk of AI momentum. There's a lot of a lot of questions around will people pay for it. Do they care. Is it spooky? Does it provide value? Um, but it's it's definitely fun. So I guess we got we got to talk a little bit about AI, [00:40:30] like, um, you know what? What's what's your AI like? What are you guys doing with AI? Right.

Kris Bennatti: Yeah.

Kris Bennatti: I'm going to take a really quick second to, um, put in a shout out to my co-founder and his book, uh, which is doing really well. Uh, it's called designing LLM applications, I think.

Isaac Heller: Okay.

Kris Bennatti: On the O'Reilly platform, it's an exceptional textbook. If you're [00:41:00] interested in designing applications, using Llms, using AI and.

Isaac Heller: Amazing, by the way designing. Okay, I got it. Designing large language model applications a holistic approach by Suhas Pai Pai. And it's got a 30 day return guarantee. So you guys should definitely, um, check it out. Okay. Got it. So tell us about the book and then [00:41:30] about how you guys are thinking about AI.

Kris Bennatti: Yeah. So the book only I think half the chapters are out. It's still flying off the shelves regardless. But I do think, uh, Suhas is, uh, going to be chatting about Agentic frameworks, etc. in the later chapters, so please look out for that. Um, he also does a lot of talks. Um, he. I'm sure there's lots [00:42:00] of recordings somewhere on the internet of him talking about agentic frameworks. Et cetera. Uh, but one of the things that, you know, we see come up a lot is, you know, models are so powerful. Now, when we started, we were using Bert. I don't know if you remember.

Kris Bennatti: It was.

Isaac Heller: Panda. I'm on a roll today.

Kris Bennatti: So Bert was a very popular open source [00:42:30] language model back back in the day, like, you know, came out in 2017 or 20 18 or 2016, one of those years. Um, when we were starting the business and it was hugely powerful, you know, language models were this incredible step forward. But they had lots of limitations. And our research at that time was overcoming the [00:43:00] limitations of Bert style language models in the context of securities filings. And now the base models that we use and train and work with are just so much more powerful. And frankly, you know, I've spent years saying, you know, math isn't something that language models can do. And [00:43:30] now we're in a place where we can have models, you know, a assign a task to a calculator in a way that's a lot more effective and actually do some basic mathematics and do comparative analysis. A lot of things that really weren't possible before, and that has completely opened up and expanded what we do, [00:44:00] uh, and made, you know, we now specialize in really nailing advanced multi-step mathematical calculations for specific workflows in equity research and investment research, which was never a place that I expected to be.

Kris Bennatti: So this is super exciting. Yeah.

Isaac Heller: That's awesome. So you said similarly jumping on the AI bandwagon early, getting accelerated along the way. You've got a great co-founder with this with [00:44:30] this new book to check out the last minute here. What's you know, what's next for Hudson? What's next for Chris? Is it is it more of the same? Uh, is it having fun? Do you guys have a another vision that you're another level of the vision? Is it agentic? Um, how are you thinking about the next few years? And by the way, you could say it's a secret. You have to check it out. You know, you have to follow us.

Kris Bennatti: So it's this is only semisecret. So [00:45:00] I a quick sneak peek here we are launching the Hudson Labs Co-analyst, which is a we refuse to call it a chat bot. Uh, but it's a specialized for equity research investment research workflows, particularly, uh, ones that require using both math and qualitative information in a single query. Um, and we're allowing our users to get direct [00:45:30] access to our infrastructure and our models using natural language queries, which is going to be really cool. And yes, it does, uh, touch on and use Agentic frameworks.

Isaac Heller: Very cool, very cool. Um, man, we should have done this as a CFA accredited because all of the all of the all of the I was going to say bros, I'll just say all of the all of the equity research, you know, men and women out there maybe is there like CFA accreditation? [00:46:00] I feel like the accountants aren't. I'm sure there's some accountants, some CPAs out there who love equity research, but I mean, that sounds awesome. I might check it out too. And you guys have Hudson Labs. You guys have, like more of a self-serve option as well. Like, if you're an independent equity analyst, you could go online and pay to use Hudson. Or is it only funds and corporations?

Kris Bennatti: We absolutely work with, um, single. We definitely work with investors who are managing a lot [00:46:30] of money but may not have large teams. So.

Isaac Heller: Interesting.

Kris Bennatti: Um, yeah.

Isaac Heller: That's awesome. Well, I mean, this has been fun today, Chris. Maybe, maybe part two will do like Agentic I looking back, you know, like one year later, you know, we'll we'll come back and share some some updates on what's advanced.

Kris Bennatti: Yeah, we should invite Suhas for that. He has a.

Isaac Heller: Let's do it. Let's do it.

Isaac Heller: Suhas and Suhas, I hope you get a couple extra books uh, here.

Isaac Heller: But, um, [00:47:00] I mean, look, Chris, uh, it's been good seeing you talking to you today. It's been fun. It's been insightful. Um, and I just appreciate the time and the insight.

Kris Bennatti: No, I really appreciate your time and, uh, you know, I wish we had more time so I could hear what the next steps are for.

Kris Bennatti: For Trullion.

Isaac Heller: Oh, wow. Well.

Isaac Heller: Okay, so we'll do that part two with the next step in. And, um, and then a check in a look back at, at Hudson. But until then, uh, [00:47:30] I guess. Thanks everyone for watching and listening to the show. And, you know, we'll see you next time.

Kris Bennatti: Fantastic.

Isaac Heller: Thanks.

Kris Bennatti: Thank you.