Transcript
Mike Matchett: Hi Mike Matchett with Small World Big Data we are here today talking about something a little bit the same as what I love to talk about, which is AI, and a little bit different from what we're talking about, which is accounting in fintech. Why am I going into accounting? Well, it's kind of a cool application of AI. It has a lot of it impacts as well, and how someone might support this from within their organization and really help a company accelerate what they're doing with accounts both payable and receivable, I should say, as well as a few other financial processes that are going on. Stay tuned. We've got Auditoria here to explain what they're doing. Hey Nick, welcome to our show. Nick Ezzo, VP - Strategic Accounts & Market Development: Thanks, Mike. Great to be here. Thanks for having me. Mike Matchett: All right. So let's just let's just I just want to hear a little bit about your background. First of all. So we're talking about using AI. And we'll get to the AI parts in a bit. That's what's definitely where we're going. But we're talking about accounting Primarily. What drew you into the world of accounting and finance? How did. How did you get drawn into that? Nick Ezzo, VP - Strategic Accounts & Market Development: Well, it's been a kind of a passion of mine for a while because the, uh, for lack of a better term, the accounting and finance team are kind of the stepchildren in the organization. You know, as a marketing guy, I've got a lot of cool tools, salespeople, a lot of great tools, developers, a lot of cool tools. The finance department has gotten left behind historically. And, you know, I felt it was important to, uh, to help these guys kind of knock off work early, maybe take a vacation, take their family to Disney World or something. They don't need to always be in the office. Those people you live with, those pictures on your desk, you want to spend more time with those folks, right? Mike Matchett: All right. So so the accountant, those poor, those poor accountants with with their eye shades on and the ink stained forearms balancing ledgers and stuff. All right. So there's I mean, we look at a company, there's there's there's an ERP, there's a CRM, there's a bunch of, there's a bunch of software kinds of tools there. Uh, but none of them really usually have the kinds of answers that accountants are asked to produce like it was. So an account's job was to really pull a lot of different information together, do an analysis, make sure the books balance, and then answer some questions to the CFO and so on, and to to even to the board about how the company is doing and where it's going and so on. So there's a lot of integration going on there. It seems really ripe for AI here. So tell us a little bit about Auditoria. How how Auditoria saw opportunity. What were what were some of the big problems that you guys set out to really solve when you, when you, when you dived into this. Nick Ezzo, VP - Strategic Accounts & Market Development: Yeah. Well there's a few like very transactional problems that we were trying to solve. One is that the accounting team is beleaguered on all sides by requests from their suppliers, from their customers, from their internal stakeholders, from the board or from their investors. And a lot of those come in through email. And we determined that if you take a an AI chat bot, smart bot, whatever you want to call it and apply that to these questions that are coming in, you could probably use a large language model to handle 80% or more of the things that are coming into the accounting team, and we've done that today with dozens of companies and been very successful in automating those email requests that come in from suppliers, customers and others. Mike Matchett: I think I think I've even submitted emails to many different companies to the AP app. So in such a company, or are at such and such a company and use those mailboxes, right? So I can only guess at how busy and flooded those boxes must be with stuff. And to have an automated process that can actually interpret what's coming in and maybe kick off the right sub processing seems like a real benefit. But you know, we're where do you where did you start? I mean, this is there's so much going on in what could possibly in an email to AP and ah, uh, where would you, where, where did you really say here's the bigger value. Nick Ezzo, VP - Strategic Accounts & Market Development: Well, believe it or not, there's there's a long tail there. Um, and in the, in the supplier questions that come in the vast majority are people asking for money. Where's my money? Where's the, you know, is the check in the mail? What? Am I getting paid? Um, I sent you this invoice. Did you get it? Um, and so that's the number one question they ask. And we've, we've figured out there's there's a million different ways people can ask for money. And we pretty much know all of them because we we use an LM to figure out, like, oh, this is a person asking about this invoice or these 5 or 10 invoices, or maybe there's a CSV attachment that has 100. Um, so the bots are really, really good at interpreting those types of questions. Then it kind of does a long tail like, hey, you know, why did you short pay this amount? And I'd like to change my banking and billing information or, um, you know, can you can you send me this information? Can you, um, you know, I want to update my, you know, my my 1099. So there's a lot of other kind of weird requests that come in. We handle 80% of those. And what we don't handle, we enable the accounting team to actually handle more efficiently, because the smart bot summarizes what the person is asking for and says, hey, the person asked me to do something. Here's what they asked me to do. I'm not allowed to change banking information. Human. Please do this for me. So that's kind of where we got our start. And then, you know, Digitizing invoices, digitizing other documents that came in was a natural adjacency, and then reaching out proactively to suppliers to ask for information or reaching out to customers to ask for information became a very another like onerous, laborious thing that human beings are not very good at. And we applying authentic AI to it seemed like a very quick win and it has been. Mike Matchett: So let's just dive into that. So you say a genetic AI, this isn't just an LLM. This is this is this is something that has ability to talk to multiple data sources. Let's say what do you combine together. What does this what is your AI system able to do on the back end on behalf of an accountant. Nick Ezzo, VP - Strategic Accounts & Market Development: Yeah. So right now we can do everything that basically that a junior accountant can do. You know, we can reach into any number of data sources in the back end, whether it's an ERP system, a CRM system, a procurement system, a ticketing system like ServiceNow. Um, we we have access to all that data. Um, and the smart bots, what we call our agenda GUI, they can access all that data and provide the right answer to the the the requester, whether it's a customer or a supplier. Mike Matchett: I know, I know, the people watching this are going to have this question. So I'm just going to I'm not trying to embarrass you or anything. I'm sorry, but security and making sure we just aren't going to hand out money. Uh, people are saying like, hey, as far what I remember is about AI is that they can hallucinate. So am I going to have an AI agent out there who just decides to give away lots of money and on behalf of my company and do it automatically? Does that happen? Nick Ezzo, VP - Strategic Accounts & Market Development: Good question. We get a lot of questions like that. That or is I going to take my job away? Um, is it going to run amok and change all my data to my system? Um, I have to remind people, and I'm sure your audience knows this, Mike, is that, you know, AI is nothing more than a computer program that somebody wrote to do a very specific thing. And we don't allow our our smart bots what we call our agenda. We don't allow them to acquire new skills on their own. We don't allow them to freelance or write their own emails or anything like that, and we certainly don't allow them to change or update banking information. So if you know anything about finance, accounting, people, they're very by the book. Everything has to kind of fit into its boundaries. Everything has to tick and tie. And they love predictability. They do not. They disdain unpredictability. So we go to great lengths to make sure that our users understand what they're what we're asking the bots to do and what they're not allowed to do. And what happens when the bot is asked to do something like change banking or billing information. Somebody says, hey, send an email and please change my bank routing number and checking account to this offshore account are not allowed to do that. Like like Roomba who sometimes gets in trouble, you know, on the carpet over here. And I got to move Roomba into a new position. The bot will say, hey, look, I'm in trouble, human, please help me. And you know, the the human operator would then look at the transaction and say, okay, we can do this or not. So we actually act as a little bit of a buffer. So to your question, are the bots going to put money out the door that we are not wanting to do? Actually, they do the opposite. They prevent human beings from accidentally sending money to people that shouldn't get it. Mike Matchett: It becomes an extra security layer rather than than an extra layer of possible leakage. Right. We actually are enforcing laws using the agents, enforcing the rules a bit more, which is pretty cool. Um, so, you know, it does sound also like I think there's an opportunity here to use AI to understand more than one, more than one language. And by that I also mean the end users language in the accounting language first of all. Right. So we translate all those different emails coming in into account account if you will, but also for the LMS to now actually help a company, uh, sort through maybe going globally or expanding their regions or starting to provide multiple language support. Nick Ezzo, VP - Strategic Accounts & Market Development: Yeah. So to tell your audience a little about how the secret sauce is made, we have underlying LMS that we, we access. And those could be like llama, Gemini, OpenAI and our own proprietary specialized language models. Now those those third party external models are already multi-language and multi-currency. So by default we support over 300 languages and 300 currencies without even breaking a sweat. So we didn't have to do that heavy lifting. So that's one of the benefits of these advanced llms that are out there, is that it does allow companies, to your point, to become global entities. They can support requests from all over the globe in any language somebody wants to speak. Invoices sent in any language or any currency. And the bots can respond back in the appropriate language of the person using predefined templates that your team would create in that language. Mike Matchett: All right. So it's kind of kind of in the first blush. Looks like, you know, a chat bot help desk kind of situation. But really they're doing something very productive, which is helping people do their apar and their, their clients or vendors, uh, get to the next step in those transactions, uh, much more efficiently. And, uh, it sounds that's. Nick Ezzo, VP - Strategic Accounts & Market Development: True, but there's more, because AP and R are very transactional. Things come in, things go out. Yeah. Financial planning analysis or FP and a and a lot of your audience might support the FP and a team or the CFO. They want to know what's going to happen in the future. So right now we're cooking up some some native functionality within our bots that you can ask questions like, you know, if the interest rates change over the next 12 months by X and our cash flow, what was our cash flow look like, first of all, and based on that, um, what are the risks associated with our business plan? So and the bots can now be trained to access both internal ERP, CRM, ticketing system data as well as external sources like Bloomberg, Moody's, Morningstar Insights and others. So that's kind of the direction we're heading into, which is less about the transactional stuff and more about like helping the CFO become a more productive contributor to the organization. Mike Matchett: Really sort of a healthy financial analysis where where he can, with just some natural language queries, ask things that might normally take weeks or even months of lower level people to assemble and get more of a direct response. As the AI agents and their AI agents and their friends, other AI agents go out and assemble, assemble an answer for him. I mean, I've been pretty impressed lately with some of the authentic AI tools I'm using in my world. On here, I can imagine that for someone who's in an accounting world, that may be lagging a little bit just because of security concerns, but if you guys are able to bring that forward and say no, we've put boundaries around it, we can ensure nothing bad's going to happen, that there's going to be a lot of interest in that. Nick Ezzo, VP - Strategic Accounts & Market Development: Yeah, for sure. It's an exciting place to be right now. This technology did not really exist five years ago. So it's a it's a it's a very exciting and interesting place to be on the frontier of a genetic AI for finance and accounting. Mike Matchett: Right. And I can imagine then beyond the accounting, if the CFO can answer these questions rapidly, I can imagine there's going to be other CXO types who will be like, wait, I want to ask some questions of the system. Do I want to get I want to get more immediate answers. I don't even want to ask the CFO. I just want to go direct to the source. So brave new world going on here. Uh, so just just to finish off just in general, I know it's a difficult statistic to sort of throw out, but if someone has got sort of a medium sized, medium firm enterprise accounting team, and they go to employ this. What can they expect for returns? What what what are they looking in terms of acceleration or ability to execute? Nick Ezzo, VP - Strategic Accounts & Market Development: Yeah. So we, um, you know, a standard implementation takes less than a quarter a month, two months, something like that. And we we fine tune the application and start to finish. And once the system is up and running and tuned appropriately, um, we, we shoot for 80% automation, 80% fully autonomous transaction processing, that additional 20%. You still need a human being. Those edge cases, things like changing banking information, changing billing, billing, or quite frankly, just customer and vendors supplier service where they need a human being to talk to. So we're not going to replace all these people in the back office. Now, if we freed up those people that you had answering the emails to do other things, the question is, what could they do to create actual benefit to your organization? Nobody went to accounting school to be chained to a mailbox all day. Um, you know, CPA does not stand for copy, paste and attach. Um, you know, nobody likes to do data entry. And, you know, kids coming out of a county school certainly don't want to call you on the phone and ask for money. So there's a really there's a whole band of transactions that can be done by this guy. There's work to do, more valuable and challenging work. Mike Matchett: And really get them to that future looking for. You're talking about where they can start to add value to the company and actually help with planning and making decisions about what the company should be doing going forward. Uh, which is awesome. Um, on there. Uh, let me let me just ask this, uh, on here, uh, and this is sort of a best practice question. Before I ask you one last question. What would you what would your advice be to someone who's got, uh, an a, a team and they're sort of going, this could be us. I mean, we've got mailboxes overflowing. We're missing things. What would be your advice to them on where to start. Nick Ezzo, VP - Strategic Accounts & Market Development: Yeah, I would say take a small step. You don't have to, you know, use the term boil the ocean. You know, find an application like like a small point solution, like a helpdesk application where you could see quick returns, make your team happier, make them more productive. And it's a very low risk, um, you know, application. You're not you're not changing a whole lot. You're not writing back into the ERP system. You know, you're just answering questions on behalf of your team to an external party. So I would say find a small, um, but but manageable, high impact projects to start with and then build from there. Once you get the confidence in the system, you start to see it perform. Then you'll you'll find all kinds of other applications. I've seen that happen over and over again. It becomes a little bit of an addiction for people. It's like, oh, I've automated that now. I want to automate this and then this. And so pretty soon you start to go around the horn and automate all your AP transactions, your, ah, transactions, um, and consider us to be part of the solution, you know, your broader solution along with your ERP, CRM and others. Mike Matchett: I mean, I'm just impressed, uh, looking at this. That to me that says if there are a company out there and you've got an AI initiative from the top down your board saying like, why aren't you using AI? So here's a use case. You can put your hands around, you can really box out and you can say, this is really I almost a templated in terms of, of protection, uh, and start there and, and actually document some good returns from implementing AI. And you can become an AI organization by doing something you're already doing just better. Nick Ezzo, VP - Strategic Accounts & Market Development: There's one more point to that, which is we're doing we're automating something that people hate to do anyway. Like it's just like if you say, oh, we automate the mailbox where the suppliers are asking for money. People are like, yes, we have a team that does that and they hate it. Yeah, no one likes to do it. We can take turns doing it and none of them like to do it. So that's also a nice benefit because it takes away the piece of work that no one likes to do anyway. Mike Matchett: Awesome, awesome. Yeah, I do it. Kick the I do. Uh, so if someone wants to get a little bit more information about Auditoria, they want to check it and they want to like read up a little bit more on it or get in touch with you guys. Obviously you've got a website, but what would you recommend they start with? Nick Ezzo, VP - Strategic Accounts & Market Development: Yeah, yeah. So our website is www.auditoria.ai. You go to our website, you can request a demo. You can ask for information. We'd love to show you how the product works. Um, we also have a lot of prerecorded demos on the site. So if you don't want to jump into a full demo with a with an account executive. You can just watch our prerecorded demos. Our solutions consulting team puts them together. They're they're kind of low production value. That's the real product. We don't do a lot of editing or trickery to them. So you can actually see the AI agents in action. So we'd love to to talk to anybody who's interested. Mike Matchett: All right. Thank you so much Nick for being here today. I look forward to seeing what you guys come up with as you take this forward with that and a part, as well as some of the other things that you might be able to expand this into. I mean, there's a larger organization there with lots of other ERP functions going on. So, uh, thanks for being here today. Nick Ezzo, VP - Strategic Accounts & Market Development: Thanks for having me, Mike. Mike Matchett: All right. Take care.
