WE CONDUCT WEBINARS AND PANELS WITH LEADERS IN ACCOUNTING AND FINTECH!

Speaker1: [00:00:00] Peel back the onion. There’s a researcher. Professor Bender has this beautiful term she has for this thing. It’s a stochastic parent or kind of a way of parroting back all the things that it’s learned in a sort of mathematical way. And but to us kind of interacting with it and you kind of reach the limits as you play around with it. It’s not perfect. It sometimes has, you know, a factual errors and but but man, it can do a lot and when trained for a specific type of domain. So, for example, if you fed it in maybe all the tax laws, maybe if you fed in all of these sort of tax plans and whatnot, it couldn’t get even better in that particular domain. So there’s a little bit of extra fine.

Speaker2: [00:00:57] Welcome to AFO Wealth Management Forward, a podcast about finance, accounting, technology and entrepreneurship. We apply our decades worth of experience and insight into what makes businesses work so we can help others grow, both personally and professionally In this ever evolving marketplace, we help accounting firms and financial advisors grow their practice through the adoption of holistic wealth management services. Learn from industry leaders and subject matter experts to unlock the secrets of their success. A podcast that shows people and companies the transformative power of technology so they don’t fear it, but instead harness it. Don’t fight the robots. Team up with them. And here are your hosts, Rory Henry, director of business Development and CEO Rob Santos of Arrowroot Family Office.

Speaker3: [00:01:41] Write Hello, everyone. I am joined by a special co-host. The man’s name is Seth Fineberg. Seth, how are you doing today?

Speaker4: [00:01:49] Doing great. Thanks for having me on, Rory. I really, really appreciate it.

Speaker3: [00:01:53] Thank you for being with us. We also have someone that I talked to, and he’s got a pretty interesting background. He was at IBM. He’s been in sales and project management. He’s been a chief marketing officer. He is now the co founder and CTO at Tally four, which is automating the book-to-tax process. And we’re here to talk about A.I. and this big thing called ChatGPT. So without further ado, let me introduce our guest, Ben Wen, and also Peter, his brother, who is also the co founder of TallyFor. He’ll be in the background like Howard Stern. He’ll be like the Benji in the background. Ben and Peter, welcome to the show.

Speaker1: [00:02:36] Hey, thanks for having us.

Speaker3: [00:02:38] All right, let’s get started. Can you kind of give our audience a background, Ben, of of what you do and what led you to tell me for sure?

Speaker1: [00:02:44] Let me give you the nickel tour. My background actually is in computer science rather than I have done sales at IBM. And I’ve also been product manager and chief marketing officer for tech companies and marketing this and basically jack of all trades and master of a few. Maybe it depends on who you ask, but because of the background in computer science, I got really interested in what is happening in AI and stuff like that. And then to get to the four part and to accounting is that Peter here, who is my brother, came to me and, you know, had these grand ideas for how to make tax software more easy to use and better and just basically suck less fun. That’s taxing. But yeah, that’s that’s.

Speaker4: [00:03:40] Basically all we.

Speaker1: [00:03:41] He bugged me. That’s my line. The he bugged me I like to joke for two Thanksgivings and a Christmas And finally I said, fine, here, here, have some prototype software. And he found someone to actually pay money for it. And I said, Don’t sell that. That thing is just a demo. And then fast forward a few years and now we’ve got a bunch of customers that actually use this tally for software in the cloud to make their tax lives easier for CPAs and enrolled agents. And for anyone that has clients that pay taxes, which for the accounting spaces is pretty much everybody. Yeah.

Speaker3: [00:04:18] And I once set the chime in here because we’ve talked about this before, Seth, about AI and the profession’s been hearing about AI for the last decade or more. Can you kind of talk about and give your input on what you’re seeing in regards to AI here?

Speaker4: [00:04:33] Sure, sure. And it’s interesting. So when I, I, I’ve been covering the accounting profession, people who know me see me around for the last 20 years or so. And when I even tried to kind of get out for for a little while and when I kind of came sort of back in full time, it’s around 2015 or so, I was just kind of catching up on like what’s going on in cloud, what’s going on with tech, because I was the tech editor excuse me, I was tech editor for years at Accounting today, and business tech has always been something I had been drawn to for pretty much a lot of my business editorial career. So really I started hearing about things like bots and AI, machine learning, things like this coming into the profession. And of course, you know, the gloom and doom Bell was was ringing loud at the time, like, this is going to take your jobs. And I’m like, You know what? Everyone just take a beat because, you know, it’s not taking anyone’s jobs the same way that the cloud didn’t take it didn’t it didn’t completely collapse. The IT department did it. Now, there was a reimagining of of some jobs, you know, like the i.t professional of old, the people who are watching your servers and making sure you know things. If something something broke, they’d move you out of the way, know, or they would they would just have to shut the system, you know, whatever these you know all of a sudden, oh it’s cloud, you don’t need servers anymore.

Speaker4: [00:06:11] Everything’s in the cloud. You don’t have to do anything. You know, you’re going to be out of the job, pal. Wrong. Same thing, really, with the death knell of, like, you know, these pretty old school practitioners saying, Oh, AI and bots are going to take away your job. So that that was the initial conversation. I want to go back probably to probably around that time around 2015 or so. And I even I even sort of wrote a whole piece and did a whole discussion with a lot of folks who were in core accounting at the time, Sage, zero, Intuit, all of them by about 26. 18 or 17 or so. They all had incorporated bots into and and and machine learning into their software into what they do. Now it seems like to me it’s just another level of, of automation. You know, it’s just it’s just really kind of the next level of like, here’s how things are going to be automated. Do you still need a human being in front? Absolutely. 1,000%. It’s just I think the folks who are coming into the profession obviously are a lot more comfortable with technology. They were born and bred in it. You know, unlike like my generation was kind of more OC Like, we got to see the birth of the Internet and, you know, and technology, we kind of sort of grew with it.

Speaker4: [00:07:34] We grew off the desktop and into the cloud. Versus kicking and screaming that it was here at the time. So we’re a little more welcoming than maybe some practitioners who’ve been around a bit longer. So things like AI, they’re not necessarily afraid of it. You know, Megan is not coming to do your taxes and, you know, and throwing out a window. It’s not it’s not happening. So, so so with this, you know, so you obviously this is the hot new toy that we’re going to talk about now. It’s I see accountants actually kind of having a little fun with it. Yeah. They’re like, can I do this? Kind of do Let’s set up a chart of accounts. Okay. Boom. There you go. Can it just kind of let’s see if it can just kind of reconcile these things. Oh, it can. Like with a little bit of playing around with it. So I would just kind of, you know, again, I’ll let you guys talk, but these are just my $0.02 about it. It’s like, if I were an accountant today, I would get very excited at the things that I no longer necessarily have to do and just you kind of work alongside of it. You work along with it and you have things that you really don’t have to think about anymore, which again, was total promise of cloud of all kinds of automation that is available to you in the first place.

Speaker3: [00:08:51] Yeah. All right. Our tagline is Don’t fight the robots, team up with them. And I don’t think I.

Speaker4: [00:08:56] Know bot keeper, you know, built this whole business around that whole idea and plenty plenty of others that don’t have bot necessarily in their name.

Speaker3: [00:09:04] Right. All right. Let’s kind of go Ben if you can kind of talk about what generative AI is and ChatGpt Yeah, what is the G, the P and the T stand for Sure.

Speaker1: [00:09:15] Yeah. And then number one, I also welcome our robot overlords, I think much you know, if you think you can take take the real back just a little bit I think Seth you are right on and I think Rory with that tagline you you’re right on if you look at the way that we use technology as human beings is that we get our time back in little bits. And if you look at even something like the printing press, yes, sure, there are fewer scribes that are that are copying books by hand. But if you think of all the innovation that is released or all of the knowledge that is released by having a printing press, it’s pretty amazing. And if you think a little bit more recent, what would our jobs be like if there weren’t Microsoft Excel or Google Sheets or things that help our brains do less of the actual calculations and things like that? You know that that the original job title of a computer was actually somebody that took calculations that had to be done by hand. Usually a card shares the cards. Not not yeah, not even punch cards. It was it was actually human beings that would do calculations before you had spreadsheets.

Speaker4: [00:10:30] That then got to the computations. Right.

Speaker1: [00:10:32] To do the computations. And typically there are often women that would toil downstairs in the basement to go and do the calculations that would get sent back up. Yeah. So it’s the.

Speaker4: [00:10:43] Original outsourced model. Exactly.

Speaker1: [00:10:45] Exactly. Outsource, you know.

Speaker4: [00:10:46] Right. So it was that you would send you would send out the tax return and that’s where it would go.

Speaker1: [00:10:52] Exactly. Right. And so if you if you look with that sort of bigger lens and you can even go further back like is writing stuff down a type of automation that we don’t have to memorize stuff. Plato is actually sort of against like having to write things down because it would soften the mind, just like, does it soften the mind to use a calculator in a test, you know, for certain tests, Yeah, you don’t want to have it because my seven year old needs to actually learn how to add and subtract and all that kind of stuff. But for the more complicated things, of course you want a scientific calculator. It’s better than a slide rule in a calculator. You know Excel is better than that.

Speaker4: [00:11:25] Don’t throw out that ten key just yet.

Speaker1: [00:11:28] I got that. Where’s that HP I have at HP 41. Got it. You got it. But nice. But anyways, so but but to get to where we are today with this chat thing which kind of exploded on the scene in the in November, you know, I tried it on the day that it came out and and I asked it to write a limerick about. Chicken in love and.

Speaker4: [00:11:51] I beco it went great.

Speaker1: [00:11:53] I actually my my 13 year old is is applying to high school so I actually asked to write a 350 word essay for a parent with a 13 year old applying to a progressive private high school. And it nailed it. So all the people that are reading those essays, I didn’t actually write it. I did actually write it. I only used it for a couple of shits and giggles. Sorry, everybody. I don’t know if I like this. Where are you going? There. Good. Anyway, I think we’re good anyway. So what is this sort of a GPT generative pre trained is based. Yeah. Yeah. Is, is basically this algorithm that uses a huge data set to create a model of for, for language for these tokens and tokens are different parts of a language model. And the particular ChatGPT, I think they’re on version three which is open I Sam Altman runs that came out of some of the work some of the folks over at Y Combinator has like an 800 gigabyte model with the tokens and the entire I don’t even know all the components that are in there. But essentially what it does is it looks at the frequency of words as they kind of come together and in some sense is guessing what I should be saying. But it guesses in such a way that we human beings, when we read it, we’re like, whoa, is that thing? Is that.

Speaker4: [00:13:26] Thing? We are we have it on our phones, right? It’s predictive.

Speaker1: [00:13:28] Text. Exactly.

Speaker4: [00:13:30] It doesn’t always doesn’t always get it right. Like, No, no, no, don’t say that. Oh, did I just send that to my wife?

Speaker1: [00:13:36] Undo, undo, undo it. It is it is sort of the next generation of that. But if you peel back the onion, there’s a researcher. Professor Bender has this beautiful term she has for this thing. It’s a stochastic parent or kind of a way of parroting back all the things that it’s learned in a sort of mathematical way. And but to us kind of interacting with it and you kind of reach the limits as you play around with it, you know, it’s not perfect. It sometimes has, you know, a factual errors and but but man, it can do a lot and when trained for a specific type of domain. So, for example, if you fed it in maybe all the tax laws, maybe if you fed in all of these sort of tax plans and whatnot, it couldn’t get even better in that particular domain. So there’s a little bit of extra fine tuning that you can do with with these models. And, you know, I suspect that there will be a whole ton of product managers that now look at this that say, okay, what can I do in my domain that makes it easier for my users to effectively we get an extra an extra brain, if you will, to help alongside you like like an assistant. Perfect analogy. And I don’t know if you guys remember in Star Wars, and I’ll stop talking here in a second. There was one character who I think was on on the one of the starships had this thing around their head.

Speaker4: [00:15:08] Yeah, that was yeah, that was what he was part of the empire. But yeah.

Speaker1: [00:15:13] It was, it was a brain enhancer. I remember I even had the little Star Wars card for I somewhere. I have the name of a.

Speaker4: [00:15:19] I think that’s exactly what it was. And then later in another series, it had a, one of the one of the clones also was fitted with the, with one of.

Speaker1: [00:15:29] Those you know, we walk around now with headphones like you guys doing and these things in our head and it’ll have Alexa or chat or whatever it is in there. So, you know, and hopefully it’ll won’t mess up the.

Speaker4: [00:15:42] Brain enhancer or hair.

Speaker1: [00:15:44] Exactly. But all these things are going to become some type of brain enhancer, just like we already have brain New Hampshire’s, we call them computers or iPhones or Microsoft Excel. Today, there’s going to be more and more of this and the better that.

Speaker4: [00:15:55] They learn over time. Right. Would you say like this is something I don’t want to get lost, particularly on professionals who are like, yeah, I don’t know about this. It’s like, yeah, you feed it. The more you feed it, the stuff that you that you want it to learn, the more it is going to learn.

Speaker1: [00:16:11] Yeah, absolutely. And it will be like almost having a perfect memory in, in maybe some sense. Right. We can search our email.

Speaker4: [00:16:19] And then you tell it to like when this isn’t correct.

Speaker1: [00:16:23] Right. Exactly. Exactly.

Speaker4: [00:16:24] I go back and.

Speaker1: [00:16:27] Right and remind me what I need to do this. And there’ll be all these sort of things that you would expect an assistant to kind of do and do it in that sort of dialogue based.

Speaker5: [00:16:37] I have a question. So, you know, you say ChatGpt Right. And you talked about the three letters afterwards, but why, why chat and how is it different than Google Search? Right? And I obviously ran a couple when I first came out. I ran a couple of things I wanted to ask you, but I’m curious, you know, is there a is there is there a is there a leap into the conversational part that makes it more accessible to to regular folk? You know, what what is the actual language part that is.

Speaker4: [00:17:07] Like voice activated or.

Speaker1: [00:17:10] Yeah, no. Good question.

Speaker4: [00:17:11] Yeah.

Speaker1: [00:17:12] I think, you know, when you the way that chatbot was built, in some sense there is this thing called the common crawl. And the common crawl was built, I don’t know, several years ago when they realized that actually crawling the Internet is a useful feature. We should make it open. Now. Google itself and all the search engines to a large degree have their own crawlers and they have their own proprietary corals, and then they have a gigantic index, effectively with some a little bit of training and hinting. And and, you know, for Google, the original one was just to look at the way the backlinks connected with each other.

Speaker4: [00:17:47] And it was it was the search engine for Yahoo! If you want to go back.

Speaker1: [00:17:52] Yep. And and and now it’s a search engine for a lot a lot of different.

Speaker4: [00:17:57] Everything There are a few that.

Speaker1: [00:17:59] Are trying like DuckDuckGo and.

Speaker4: [00:18:02] Yeah.

Speaker1: [00:18:03] But but for the most part they’re a pretty straight index and you know here’s the top hit for this set of words. Now there is some parsing around the language, trying to figure out what are the keywords and whatnot. Now, in some sense and Google has a project, though its name escapes me right now, land or something like that is trying is using a GPT in order to create from it. But the experience that we have as humans in approaching this sort of conversational bot is different because we ask it, you know, a pretty open question and it extracts out what we think it wants the answer to be, which is different from the straight. Like, Hey, here’s the top list of things that people got when they search for Toyota, such and such and such and such. Right? And I don’t think that a chat function is going to replace the search function. I think there will be some augmentation. And there’s also the understanding of what is important to my relevance search that is ultimately going to be very, very human. And that I don’t think is going to go and I’m sure there are people over at Google trying to figure out like, well, what’s the best way to go? And and now take this lambda thing to the next level.

Speaker5: [00:19:20] But well, to nerd out. So I put in a couple of searches, right. And I said, you know, talking about tax, I was like, hey, what is what can let’s test this thing about some tax knowledge? So, you know, I put in a couple like softballs, like one of them was is my air B and B income taxable? And it gave me a pretty good answer, right? It told me, and you can all do your own search like that. But then I’ve asked I asked him something more nerdy, like what is new in the Internal Revenue Bulletin 2020 Dash 34. I’m doing a tax research question. Yeah, and it actually gave me a summary.

Speaker4: [00:20:02] Of that particular tax.

Speaker5: [00:20:03] Code. Yeah. So I’m like, Oh, that’s better than Google in a sense, because Google, you’d have to go to it would give you the link and it go into you read the, you get the PDF of the IRB and then you’d have to read it and understand it.

Speaker4: [00:20:18] Yeah. Unless you had like checkpoint or something like that.

Speaker5: [00:20:20] Yeah. Well exactly. You know, I’m just trying to see as a tax professional.

Speaker4: [00:20:24] Yeah, you probably have your CCH or.

Speaker5: [00:20:27] Something, but can this give me that extra lobe of my brain and say, Hey, I put you down the path one step further? And, you know, I actually thought it was quite interesting. So.

Speaker1: [00:20:37] Yeah. And I think that’s the actually one area that we’ll see some pretty interesting use of these GPT engines is to take a look at either the tax law in particular, but more likely the analysis that comes from various places that you subscribe, whether to checkpoint or what have you from the other providers. So I think there’ll be some interesting things there. You’ll almost have an assistant that you can ask kind of questions. And what I think is super interesting is that because it sometimes is wrong, you will definitely.

Speaker4: [00:21:11] Want as human beings are.

Speaker1: [00:21:14] As human beings are as as the algorithms and the people that built the systems. I think it’s also important to have that expert have that CPA to have that enrolled agent and have that advisor to actually look at it and say, okay, that gives me a good hint. But for this particular client, no, that doesn’t matter.

Speaker4: [00:21:34] And this brings up the next point, Ben and Peter, that, you know, to to my original point is having it as your assistant or having it alongside of of the the accounting professional. The accounting professional is there to kind of interpret and really explain it. Like, I isn’t going to explain these things necessarily as clearly as your accounting professional will. And because things obviously you start getting into tax law, you do need a tax law professional to kind of interpret the law because there are interpretations of law just like there are. You know, as clear is clearly as the IRS is clearly as tax codes may seem to be written. And there are situations where this might not apply and it’s not as clear. And that’s why you have tax pros year after you’re pulling their hair out, because they’re waiting for explanations and just guidance, if you will, from from the IRS, from the governing bodies out there, from the AICPA, from whoever can weigh in on on whether it’s an audit or a tax code.

Speaker1: [00:22:49] Some tax That tax code is as clear as the bell, isn’t it?

Speaker4: [00:22:52] Yeah, sure is.

Speaker5: [00:22:55] A night time meeting.

Speaker4: [00:22:59] So you still the point is tax professionals out there again, just again, to just kind of keep you a little a little a little more calm and sleep a little better at night worrying about the the bots taking over yet again with this new iteration if they still need you possibly more than ever, more than ever, because it is not going to be able to do the things that you need it to do. Unless you it has the right information there.

Speaker1: [00:23:29] I actually would argue that to a tax professional that wants to adopt this, this becomes the cape that your tax superhero persona comes, right?

Speaker4: [00:23:43] You write that down.

Speaker1: [00:23:44] This thing.

Speaker3: [00:23:45] This thing used right, will make you a superhero.

Speaker1: [00:23:49] Well.

Speaker4: [00:23:50] It is the cape. You need a cape.

Speaker1: [00:23:52] It is the cape. It will make you fly even faster.

Speaker3: [00:23:55] That’s. Well, I mean, they did. I was reading somewhere, and they did this study with a nonprofit who was doing peer therapy. And the ChatGPT was responding to people who had some mental health issues and the response time was cut down by more than 50%. And the responses after they did the survey were much better than they received from the actual human. But once the respondents found out that it was the computer, their very turned off.

Speaker4: [00:24:26] Oh, I just read that today. Yeah. Actually it was like, oh, their response was like, Yeah, they enjoyed it. And even my my management company, like, I live in a massive, massive apartment complex here in the city to where it really has it doesn’t have its own zip code. It should. But we’re we’re about 24,000 people that live in this complex that I’m in. So we have a management company and they have tried to I, you know, sort of chat power. The request you can have on the phone. There’s a number you can just kind of text in for for maintenance requests. And you can imagine, you know, 24,000 people at any given time, you know, having some issue with their unit. You know, they so they’ve tried to incorporate bots to to help and they need work. You know, you ultimately you just feel more comfortable speaking to a human look at look at Amazon customer service. You know a lot a lot of the web, a lot of the Web facing companies that are out there, you know, they’re they’re they’re bots, by and large, that are dealing with these questions day in and day out. So they’ve a lot of a lot of them have learned enough to the point of like, okay, I can take a certain string of queries, you know, much like basic coding used to be, right? I don’t know what you guys sort of came up on if your your C plus plus people or 2600.

Speaker1: [00:25:51] Okay that was.

Speaker4: [00:25:52] The but you know the language that you that you learn, the coding language that you learn, it’s kind of like that in a way that that was my initial interpretation to of of t was like oh wow, this reminds me of like old coding. Like you pretty much are just sort of telling it what to do, what search.

Speaker1: [00:26:10] Yeah. And you know, next generation will look back and say, Oh, isn’t that quaint? And right, right. The Atari 2600.

Speaker4: [00:26:19] Because it’s just kind of read your thoughts, the.

Speaker1: [00:26:20] Nerd out. Yes, it will. It will. Great. We’ll see in our little pods And I call the energy out of us just like to make it.

Speaker4: [00:26:27] I could really use a smoothie. That’s right.

Speaker1: [00:26:31] And what was the character in The Matrix? He just. He just wanted a steak.

Speaker4: [00:26:35] Oh, like I wanted a steak to taste like steak.

Speaker1: [00:26:38] Exactly. And there was a comment. Oh, the tax law interpretation, too. It’s interesting. You know, I asked tax, but definitely somebody put in a thing like, well, you should talk to your tax advisor. But, you know, and so there was a little.

Speaker5: [00:26:53] Disclaimer.

Speaker1: [00:26:54] Snuck in every time that it.

Speaker3: [00:26:56] Answered it. It could be somehow lead gen for advisors.

Speaker1: [00:26:59] Oh, totally. I’m like.

Speaker4: [00:27:01] Here we go. Now it’s got a marketing angle to it. It’s like, Oh, we’re going to just start. Yeah.

Speaker1: [00:27:07] There will be interesting stuff that comes out of this for sure. And I think it will be superpowers that that for the folks that want to adopt it and even the folks that don’t, I think it’ll help make your your experience with software also more efficient to you know it one of the big use cases that came out of an early use of GPT I think was GPT two is something called GitHub copilot which you know and I’m a soft and I was like, Hey cool, make my life a little bit easier and give me suggestions for code. And you would like you type in a comment in your comment language. I haven’t used a lisp like language called closure. That’s what we use over at tally four and you put in a comment. So this is kind of what I want and it gives you a pretty good at least to kind of look at. And certainly there have been things that that I’m like, oh, you know, I hadn’t thought about that. I should probably go read up on that. So it makes me know a little bit more about library or whatever it is that that might be used. That is just kind of out there. And I think that’ll happen with a lot of different industries. The product manager for that actually came out with this term called this I overhang, which is that the AI technology has run so far ahead that the product managers and the product developers like myself, we’re a little bit behind because. There should be ways to shake loose this extra value out of this technology. And we’ll see that, I guess, in the next year or so. I saw that for sure. Altman’s company is going for like a $29.

Speaker3: [00:28:39] 30 billion, I think valuation two.

Speaker1: [00:28:41] X of what it was prior. Och, that’ll be fun. So that might change things in terms of the economic downturn or or recession or whatever that we may or may not be going through in 2023. Knock on wood.

Speaker3: [00:28:55] We don’t. Yeah. I mean, you are you have a product you’re in the marketplace for for accountants. How do you envision utilizing something like GPT, Ben and Peter in regards to tally for.

Speaker1: [00:29:10] Yeah, I’m just the coder. I don’t know. I don’t know. I can make stuff up.

Speaker5: [00:29:17] Yeah. I mean, you know, we’d love to hear from our customers what they like to see. You know, I do think there’s right if we look back at what cloud accounting did for accounting, what we want to do for tax in general, some of the road stuff. But I do think that there’s a new horizon in many ways of of kind of the research. You know, how do you surface some of this information, just like all we’re talking about today, you know, whether or not we prioritize it sooner rather than later, you know, we definitely want to make it easier for everyone to do their jobs and, you know, have more fun. So that’s we don’t really have any concrete plans at the moment. So I’d love to hear what the audience would would like to say.

Speaker4: [00:29:56] I feel like it’s going to be an organic thing, Peter and Ben. So you’re really more to your point, it’s like, look, you guys are still continuing to improve what Tyler four is and everyone in the tax and accounting space is continuing to just do just that. And if they’re doing that well, meaning listening, listening to their customers or listening to their user base and being like, okay, what what could we solve for the best? What is what is your current pain point here? Sure. If I can kind of work in there and and it makes sense rather than just go, Hey, this is our new and shiny, which I think a lot of the a lot of the core accounting folks, you know, might have initially jump the gun a little, but now it’s just become like like, oh, you don’t actually have to go out and get this like I had. I remember having to write articles and have these conversations. It’s like just explaining, look, it’s there don’t necessarily have to go out and get an API or you don’t have to get anything additional. It’s just it’s built into what you do. And if it’s doing its job, it’s going to be pretty seamless. You know, obviously the front face of things is when you see like a chat sort of feature come up much in the way that that if you remember the old versions of word, you’d have the paperclip come up and.

Speaker1: [00:31:22] You seem to be typing a letter. Would you like me to know.

Speaker5: [00:31:27] There’s actually something.

Speaker4: [00:31:27] You know, it’s not all going to be like.

Speaker1: [00:31:29] That.

Speaker5: [00:31:29] No, there’s, there’s like cycles, right? And sometimes the laggard can kind of leapfrog, you know, like technologies that were kind of sitting around for a long, long time and haven’t changed. In some ways. They get to leapfrog because they kind of wait, wait, wait, hey, this is the next thing. And they can actually become a leader by adopting a little bit faster. And that’s what I think tax has the potential to do, because there’s so much that they haven’t done so.

Speaker3: [00:31:54] Because it is a laggard out there.

Speaker5: [00:31:56] Exactly.

Speaker4: [00:31:57] Well, yeah, I mean, it’s one of those things you don’t certain things you don’t want to change that much. Right. You know, I think some tax pros would wish that things would just kind of stuff changing. Just stay the same for a little while. We’re good.

Speaker5: [00:32:12] You know, they’re interesting. You know, we have obviously different generations using telephone. Right. And, you know, we get to see like what different generations expect out of software. Yeah. And, you know, I find it very heartening the younger generation is like, no, it should do this and please do this. Or, you know, why does it do this? And, you know, so that’s where we want to head and we want to be ahead of that.

Speaker1: [00:32:36] We’re going to have a tick tock interface so that, you know, but and, you know, when I used to work at IBM, one of the things that we used to do were ingest data from all different forms and all different types. And, you know, I see it here too. You know, I’m waiting for somebody to say, can you pull this balance sheet off of this paper, scroll something like that. And that that that is something that I was kind of waiting for.

Speaker4: [00:33:02] I guess it’s almost like the thing that Receipt Bank did and some of the others did with being able to scan in and interpret really the most crumpled up receipt. And this was the testing. You would keep putting it and putting it in system of you know and I think hub talk same thing you know just the worst crumpled up receipt that you could find and keep feeding it in the system until it actually read it well like I read what it’s supposed to be and that’s and pulled the data off of it.

Speaker1: [00:33:30] Yeah, I’m going to send something to my seven year old Drew on a piece of paper and see if that can get scanned in.

Speaker4: [00:33:36] See what? See what it does.

Speaker1: [00:33:39] Oh, $600 a Lego series.

Speaker4: [00:33:43] Damn. But yeah, I mean, I’m. I’m with you guys. I’m very excited to see sort of where it’s going to go. And maybe you’ve had some internal discussions you don’t have to give. Waiting trade secrets or anything. But about like, hey, where where can this type of this degree of I work in to tally for for what we’re building.

Speaker1: [00:34:07] Definitely. You know we we we like to dream and depending on how little sleep I’ve gotten, there’s always something interesting kind of kind of in that in that sleep addled brain of mine but but yeah no I think there’s going to be a lot of interesting stuff in tax and accounting in all these different places, anywhere where you type text and have a some sort of expectation of what would come back from that, you know, is, is a place where you can see something will happen and without yeah, there’s nothing concrete that we have, but there’s always things that we want to play around with, especially with fine tuned models and whatnot. And it also has to be, you know, cost effective too.

Speaker4: [00:34:57] Yeah, well, yeah.

Speaker1: [00:34:58] And how that all fits.

Speaker4: [00:34:59] Well, that’s a good that’s a good thing to, to actually bring up. I mean what are they sort of coding you know hours that would that it would that it would take you know to to work alongside a bot or to really to incorporate AI. That’s something I’ve kind of always wondered about too, like is it worth the time? Is it worth the time?

Speaker1: [00:35:21] It is. And you know.

Speaker4: [00:35:23] Like you can get again, you can’t get this off the shelf and just go, Yep, we got it.

Speaker1: [00:35:27] Now. It’s getting close though, you know, I mean I there are relatively affordable models for the standard GPT three and then you can fine tune the models. They get a little bit more expensive than how much you want to feed in there. And I there’s a parameter how many rounds you wanted to take and whatnot. It gets it’s about ten times, I think ten times as expensive as the standard model. So the question can you actually.

Speaker4: [00:35:51] You got to pass on those costs.

Speaker1: [00:35:53] Right? Exactly. So and then what does that look like at scale? And then what fraction of the things that you put into this thing you actually can charge for versus things that are like, Well, no, that’s just a what we call a trial, but don’t aren’t really providing the value So.

Speaker4: [00:36:05] Nice to have versus need to have. Right.

Speaker1: [00:36:07] Exactly. But but somewhere along the line, there’s going to be something super interesting in the space. And, you know, I mean, for anyone who’s listening and has has a big data set that they want to play around with and tune, you know, I’d be happy to entertain some some discussions on on that around the area of tax and even broader bookkeeping and accounting. If you want to go to that next level and figure out how that might work. But yeah, there are a couple of things that I can imagine that we could do and have fun with, whether it’s not directly direct client facing, but probably for, like I said, the cake for the superheroes.

Speaker5: [00:36:44] Well, you know, Ben, when you mentioned some of that, you know, when I think about what you mentioned, Low-code no code, right? You know, I think there’s a equivalent or parallel in accounting, right? There is a standardization. There is a corpus of things that people are doing that you don’t realize they’re doing all the same until you compare it all. And we can’t even do that today with with stuff locked in, you know, one database here and there. But I think that first premise right, actually first getting that crack into seeing that data and then saying, hey, oh, you know what? You know, dentists do this and, you know, restaurants do this. And and it’s and then now you can almost have that low cost, you know, template type standardized kind of thing. Now, you know, then the automation comes, right?

Speaker4: [00:37:28] So you can go by industry like that.

Speaker3: [00:37:30] That’s I mean, I always bring this up on the podcast because we’re fascinated by it. But private equity’s involvement in the space really against 75 million baby boomers retiring by age 30, by 2031, like 5.1 trillion.

Speaker4: [00:37:45] It’s already happening, man.

Speaker3: [00:37:46] It’s 5.1 trillion business value to have that data. The potential for roll ups. Yeah, I mean, the accounting, you know, I always say this is probably the golden age for the accounting profession.

Speaker1: [00:37:59] Yes, it is. Yeah.

Speaker5: [00:38:01] Well, yeah, really. I mean, that’s, you know, it’s such a fundamental human activity, right? Business and taxes. And, you know, some people think. Right. Language and business and taxes co-evolved. Right? So it is a huge corpus of knowledge. And you know, the way you put it right, there’s a there’s a certain generation that’s moving on. How does it recycle back into the cycle? So it’s extremely interesting from that theoretical standpoint, too.

Speaker3: [00:38:32] I like it. Well, I think we should end there, gentlemen. This has been an exciting talk. Seth, did you have something?

Speaker4: [00:38:38] No, I was just going to I was just going to say, look, I. Now. I understand. I’m still I’m still punching, man. I was just going to say that whenever we we get rolling on a on a on a technology topic, you know, the pace that we’ve seen even in the last ten years, let alone the last 20 that that that I’ve that I’ve seen, it can be a bit intimidating. It can be a bit overwhelming. And I would just say, you know, to to the the tax pros and accounting pros at work out there like this isn’t the stuff that you need to be to concern about. Pay attention to the folks who are paying attention to it, like our guests here and and the like and the folks who are, you know, tinkering around with chat and other other things like it with with the bots that are out there and, you know. Let your let your software vendors, the vendor relationships that you have. Let them know how things could be better. I think this is you talked about the golden age. I think it’s the age, finally, where professional accounting professionals can really take a little bit more control over that relationship and kind of guide where technology will go rather than having it just simply done to them or for them or what have you. And and then just having to kind of backtrack a little bit. And that’s that’s kind of the biggest change that I’ve seen and hope to continue to see.

Speaker1: [00:40:24] Yeah, no, definitely hope.

Speaker4: [00:40:26] That makes sense.

Speaker3: [00:40:27] All right, Peter, any last words?

Speaker5: [00:40:29] Yeah, I’m good. It was really fun. Yes, You guys been talking about it.

Speaker4: [00:40:32] Thanks, Rory.

Speaker3: [00:40:33] Yes, Thank you, Ben. Anything else for the audience?

Speaker1: [00:40:36] Nope. Sorry to talk over there, Rory. Seth. Thank you very much, Peter. Awesome to do this again with you. And then. Yeah, look forward to.

Speaker3: [00:40:44] Let’s revisit this here in the middle part of the year. So we’re at 3.5 and we’re at. All right, guys, appreciate it.

Speaker1: [00:40:54] Thank you. Thanks a lot.

Speaker2: [00:40:55] And opinions expressed by Rob Santos and Rory Henry on this website, podcast interview are solely their opinions and do not reflect the opinions of our Family Office LLC or their company or affiliates, and may have previously disseminated on television, radio, Internet, or another medium. You should not treat any opinion accessed by anyone as a specific endorsement to make investment or a particular strategy, but only as an expression of their opinions. Past performance is not indicative of future results.