Relearning Treasury: Whatever You Know is WRONG!

Craig Jeffery [0:05]
Yeah, so it sounds like a really edgy title. Everything you know, is wrong. You know, is that really true? That everything you know is wrong? No, it’s not true. But if everything you knew is wrong, you’re in the wrong business. But the idea that there’s going to be some ways to rethink what we know about treasury, particularly how technology is impacting Treasury is the key point. So apologize for the edginess. If any of you are offended by that you probably wouldn’t have come. But if you are, or if you’re not, or if you’re here, I’ve got some tchotchkes for you. I’ve got mouse pad writing pads, I’ve got a mouse pad that you can clean your screen or your glasses with, as well as some multi charger, whether you are the House Divided with Android, or, Apple, so you have those options. So those will be those will be available to you. In addition to the concept that maybe we need to rethink what’s going on, how is technology changing? There was a number of sessions, where we talked about that concept we heard of people weighing in. So let’s look at that. So what do you think of when you see that picture on the left hand side? Anybody? You think of a game rhymes with Chess? You think of chess, what’s going on a chess and I’ll tie this into some of the things that are changing technologically some of the things that are changing with HighRadius, and what are how HighRadius is changing the environment, how different technology providers are making changes and how we as Treasury professionals need to adapt how we think about things how we think about technology, our structures, our processes. So in 1996, who here was alive in 1996? Excellent, excellent. That’s good. In 1996, who is the leading chess master at the time, started with a K? It wasn’t Karpov it was Kasparov. Right. How many people liked chess? Okay, six people. Okay, excellent. So maybe I shouldn’t go into the minutiae there. Anyway, he was the chess master. He played deep blue, which was made by IBM in a series of six games. And who won. Kasparov won 19 96 He defeated deep blue now in 1997. Deep Blue won 3.5 games to 2.5 games out of a six game series. A tie is half a point. And what happened after that Kasparov left defeated, there were some allegations of cheating. What was the cheating that the deep blue team was using grandmasters to influence how the game was played? Right? They were using human intelligence to reprogram the games in session to make it make it go better. That whole system was shut down. They never found out if there was some cheating or not. But that had happened that was machine beats man in a game. Now as time went on, in 2014 2016 2018, Stockfish was a computer game. How do you play games on your phones that are like the challenge your mind? How many play Wordle? Okay, how many know what Wordle is? Why am I addicted to that every day, I’ve got to play it. Anyway, that’s off topic. The Wordle is but Stockfish was designed. And what they did with stockfish was they took an inventory of all the historical games of chess, and loaded it this inventory to build a giant repository of useful information. For computing power. It was open source, they added massive computing power, and could play and for the most part, could beat any human and beat any human easily. And so what these tournaments are is computer chess games, playing computer chess games. Now, does that seem kind of geeky? It’s like, we’re gonna have our computers play computers. Anyway. Well, that’s what happened. So each year Stockfish has done quite well, in those tournaments. In 2017, a chess program called Alpha zero, was created and this used artificial intelligence. And so instead of loading up historical games, it set itself to playing itself. The rules were programmed in and it started playing itself. Now, how many times have they been played Chutes and Ladders with their kids? And after the third game, you are ready to go in saying you think you’re a terrible parent. You know, it’s like watching Blue’s Clues or what’s that? What’s
that Australian TV show? Is it the wiggles, it’s not the wiggles? It’s got the blowing that’s what it is. I swear that’s that That’s actually that one’s pretty good. But

Craig Jeffery[5:02]
there’s some ones that are, it’s insane. Yeah, that one kind of captures the adult. So that was a bad example. It’s like, I won’t go into that. But anyway, I’m off topic quite readily. But what it what it did is it played against itself for nine hours. How many games do you think it was able to play in nine hours, continue to play and learn, play against itself, keep learning, adding new rules, adding new, adding to its algorithm using AI? How many times like I think over 9 million games, I think, I can’t remember, but about 9 million games that played in nine hours. So next time, your kids want to play Chutes and Ladders, maybe there’ll be an Alpha zero for that. But after that, nine hours of training, think about that nine hours of training, then it played. Then it played Stockfish, which was allegedly the best computer game at the time. And out of 100, games, Alpha zero, and Stockfish came to a draw 72 times no winner alpha 01 28 times and lost zero. So think about that, it says has nothing to do with chess in terms of where think about here, but the use of artificial intelligence to rapidly learn and this ability to iterate using high compute power, you know, rapidly learning and moving forward dominated the ability to load all of the historical games that they could find from Stockfish. Now, if you’re interested in following more, there’s other information and there are some grandmasters most of them won’t play the computers. But if they do, and they put really restrict time limits. Sometimes the humans can win, because there’s ability for humans to recognize patterns, especially grandmasters to recognize patterns and come out winning sometimes. So that’s a side job if you want to look at that for some period of time. So anyway, so very, very significant difference in terms of the compute power, it’s raw compute power, but it’s this ability to learn and iterate rapidly. And the right hand side, what does that stylize picture look like? A jet fighter, right? Anybody remember Top Gun? Anybody? Wasn’t a like, older than 10. When I saw it, I remember seeing that right. That was cool. Who was in that? Tom Cruise? Was a Kelly, somebody? Kelly McGuire? I don’t remember the names. Who is it? Kelly McGinnis? Okay, at least I got a couple of letters. Right. So that was a that was a big movie and top guns or what? Anybody know, people are in the Air Force that are fighter pilots. Are they low or high on Pride? It’s very high, it would fill that stadium and then we wouldn’t get sunburned. Right? It’s very high because it’s very, very competitive. And they do extremely well. Well, there was a alpha dog fight trials in August of 2020. So less than two years ago, there was a fight trial. It was using a simulator. It was on a computer, it wasn’t in the air. Right. That’s probably not the best thing to do to crash when you’re trying to see how the computer program works. And the fighter pilots are quick to point out that the alpha dog the computer program, didn’t have to do takeoffs and landings, which is kind of like, okay, so they didn’t have to do takeoffs and landings. And what happened? They did five rounds of fights. And who won? You know what the answer is, right? Who won, it’s going to be the computer, the computer program won all five, all five rounds in the Top Gun, the top have fighter pilot about over 2000 hours of experience, said those twisting moves, we were not prepared in our training to handle those. So I imagine that’s helping with their additional training right now. But this idea of, we can use technology, artificial intelligence to help move things faster, in a more autonomous manner, that we can leverage much, much more powerful tools to do our work is something that impacts all have business, it impacts treasury and impacts other areas. When you think about, maybe whether you think about it or not. If you look at auditors, what’s happening in the audit field? What’s happening to new staff positions, very few are being hired for auditors, right? It’s becoming far more automated. It’s more that people are programming use the machines to program the activity for auditing can go through the data much more rapidly, use better rules and be controlled by faster, smarter people. How many people have auditors come into the treasury area?

Craig Jeffery [9:48]
And how experienced are they? Like they come in and audit cash. They’re usually like one year on the job. And they say is this like a checkbook? And they say, you know, show me some Wire transfers and the authorizations. And it’s like, it’s not even a real. It’s not even a real audit. If you didn’t tell them anything, they wouldn’t know much at all. And so what we see that’s a little unfair if anybody was an auditor, is anybody an auditor here? Did I offend anybody? Okay, anyway, so that’s what happens, right? You’re new you go and audit cash. But that’s replacing the vast majority of audit functions right now. And same thing with this automation is happening. It’s clearing out significant amounts of activity and AP, and AR in different areas. And so it’s going through, particularly with high intensive areas that can be that where automation can be leveraged. I say that not to scare everybody. But to say, we are getting more and more powerful tools, we don’t shovel with it, we don’t scoop dirt with our hands. We don’t use a shovel. We use giant earthmoving equipment, that’s very quick. Same thing with computer technology, we tend to leverage these tools that can make us move far more earth or far more data more rapidly than ever seen before. So there’s a couple of themes from that. So we saw chess, we saw a fighter pilots, here, this is why you should decide your order, and not change your mind. How many people like to change your mind? I do like to change it a lot. So but a couple of things here, what’s going on with data? Data is doubling every couple of years, it’s growing 40 to 60%, a year, which if you do the math is doubling every couple of years. How many people have Wi Fi at home? How many? Which is a dumb question, right? Because if you don’t have it, how are you surviving? How many people have some sort of nest or configuration that makes it really smart and binds everything together? How many people have other devices connect to it, maybe have a refrigerator, your TV, other devices, what other devices, you have security system? Everything you probably have more things connected than I do, I want to ask if you set up a separate channel, for your Internet of Things for security purposes are not because somebody will probably be taking a picture of this. But that’s also a good idea. Right? Put that on a separate channel. But the point is data’s doubling, we’re seeing that we have we have security cameras, we have machines that capture in and out we have RFID scanners in and out of companies. We create all this exhaust data from our systems. We have accounts receivable data payables data, massive amounts of data that was previously wrapped up in systems that were inaccessible. But now there’s more data. What’s happening with the cost of computers? What was the cost of your first computer? Please, someone give an example of that. The first computer you got?

Say what it was? Nobody remembers? 5000 You got a nice computer? When did you get that?

Craig Jeffery [12:59]
Late 80s. Okay, was it like a 3d six or something? 3d Six? Yeah, so it was it was like $5,000. And now you can get a really powerful laptop, that’s probably 8x or 16x, the compute power, way more RAM, way more memory, the cost just keeps going down. It’s usually having there’s some terms for these, which I won’t quiz you on. But compute costs is dropping very significantly. And so what has happened is we’ve shifted, we’ve gone from a lack of information to an overabundance of information, there’s just think back for six years, you’re talking for eight times as much information today as six years ago, that you’re dealing with, and you go back farther, it’s you have a massive amount of information, information coming from your bank, not just from the banks, information reporting systems, but all of their transaction systems that they manage, you have information that you’re tracking internally, you’re also matching that up with external information that you’re using to enrich your processes, information about your counterparties, risk, foreign currency, information, all of that needs to come together. And so we’ve moved from this lack of information to an overload of information. And I say it’s an overload of information, because you can’t handle that. Just keep rekeying stuff into spreadsheets and building your models. And so we need machines to help control the massive amount of data that we have to do our jobs, whether it’s transactional, whether it’s strategic, whether we’re looking at cash flows, or looking at risks or exposures. And then finally, we use tools like business intelligence, tools, analytics and machine learning, much of what you’ve seen here today through high radius, but also you’ll see that throughout, you know, throughout the organization, in many, many different ways and potentially put it to use primarily machine learning. We see as being applied in systems, not an independent machine learning tools that are justify it against data. But that will change. We see we see that happening where vendors are taking care of adding that functionality. So, if data is so important, it’s growing so much, we need to control it. What’s better than to say biggest small, smallest big, you remember that? That moto slow is smooth, smooth is fast. You use like an a shooter example or sports analogy or whatever, right? Go smooth, slow is smooth, smooth is fast. And then you think that doesn’t sound right, from a syllogism, because that means slow is fast. Those can’t mean the same thing. In the same way. You got an equivocation problem. And you say, Okay, I get the point, we need to be right. And so what is small and small is big, you know what’s happening with development. Anybody remember client server technology. Something is installed on a computer in your offices or your data center. And we have data centers anymore. Anybody used to have data center, you’ve heard of data centers. I was like, that’s our new thing. We’ll tell our kids we used to have data centers. And there was computers in a room that was cooled to manage that. Well, now we’ve already been talking about here that had a session here about are we at the end of SaaS, as software as a service, right, we were the session talking about how that’s changed. So as we move from client server where the stuff is installed, in your data center, you have two physical boxes, test and production or test development and production three, and then you continue to move. And then how often would those get updated? Maybe every 18 months would be a major update, there’d be little patch fixes, and everyone was scared, and it was a pain. And what would happen. Like, hey, we don’t care about this upgrade, we’re not going to go through the effort. And two of those past buys two of those past buys. And now you’re like 40 months and you haven’t done an upgrade, then the next upgrade happens. And what do you do?

Craig Jeffery[17:03]
You’re like, oh, it’s changed so much. Now we have to, we have to do three upgrades in a short period of time, that’s going to be very significant problem, and we let it go and it becomes obsolete. So the idea of small and big is what’s happened is we moved to the cloud, and moved to cloud native technologies we move to microservices, things stored in containers, they’re smaller, they’re able to be changed more rapidly, they’re more componentized. And so development happens more rapidly, we move more towards an Agile process, we’re able to move more towards an Agile process where there’s, we’re iterating faster, we’re making more rapid changes. And that’s changing the rate of speed, change the rate of change, it’s altering the rate of change that’s happening with development by what our banks are doing, what our FinTech providers are doing. And that’s, that’s decreasing the cycle time. So development cycles being compressed, instead of a major upgrade every 18 months in SAS world and move to every quarter. Now they’re pushing to every month. And so sometimes we’re seeing updates every week, because it’s it’s smaller, it’s faster, and so that this whole waterfall, one thing gets done. Next thing It started as moving and given away to lean. So that touches on the issue of what matters with Treasury. And so if you look at things like speed, or how, how slow are things? What’s getting slower in business, as you think about the last five years? or 10 years? Is anything getting slower? Is there any expectation that slowing down? I can’t think of any where there’s a slower expectation, I can only think of things getting faster, or much faster. Some things. If I ask, Who knew Who needs instantaneous data? Do you need real time data? Maybe not? Do you need current data, sometimes from your banks, other items might be necessary to see that. But nothing’s getting slower. Everything’s getting faster. It may not be instantaneous now, but everything’s moving that direction. Think back to when you first started your career. How long was it to close the books? Was it 15 days if you’re old like me? Was it 12 days? 10 days? Did people how when they said we’re gonna get to an eight day closed or a five day close? How many people here have a close in three days or less? All right, how about five days or less? 234 Okay, I’m not gonna ask how many over 20 But this idea of there’s speed and closing. Now let me ask this question. I want to if you show your hands of Speaker How many people record have reconciled your bank your personal bank accounts in your life. Okay, now put them down how many people still actually reconcile your bank accounts today? Let it show that nobody does that. Why is that changed. Have you gotten lazy I remember stapling pieces of paper to my statements, writing when someone overcharged on the tip on the credit card bill, I don’t know why I found enjoyment about that, how that came from accounting, but it no longer became necessary, because everything is reported, instantly, you can look at your statement, see what comes in? And yes, we probably are not going to say you shouldn’t reconcile. But the need is, there’s not this massive amounts of delay, this slow this terrible latency issue on your financial transactions that happens faster and faster and faster. So that’s what we see for. Whereas my cash position, you can’t use an accounting monthly cycle to say, Where’s my liquidity?

Craig Jeffery[20:45]
It’s not even good sometimes to say, Where’s my liquidity yesterday, we need to know where that is today, in increasing measure that may be different in your organization. But nothing’s moving to week time. It’s either daily, current day, or it’s moving closer to real time. And that’s impacting reconciliation, how we manage our various processes. So this is influencing everything we do. There’s no, give me two weeks to put together a slide to show how this impacted my supply chain is going to infect my cash flow, you have to be able to get that information quickly. So what do we need? This is a just a quick example. What are criminals doing? Criminals are at a course just like this probably at another stadium? And what are they teaching each other? Here’s how to catch up more people, by using automation, you know, is too much to just go out to the big companies to work on a business email compromise, we have to send out more phishing emails, we need to leverage our technology so we can catch more people the bigger net in the water to catch more people. And so that’s what’s happening. They’re using automation. And so what do we know about Treasury that we have to unlearn? We have to think about, you know, criminals are using automation. We need automation to defend ourselves to track and see are there anomalies? Are there quality control issues, quality control can be, you know, could be in a payment, quality control being how things are applied for accounting purposes. It can be in your production facility, but automation can help us point help point out issues or differences. And that is very useful for us to spot anomalies, for fraud to spot quality control issues, to use tech to point out those items as that management by exception type issues that we want to see what’s going on. That doesn’t look right. How many people have the disease, if you see a misspelled spelling error on a slide, your eye goes to it, and you almost can’t function. Yeah, and it’s a problem, right? It’s, it’s good, you see it, you just see it. And you can annoy people, you’re like, You got to fix that before we move on. Maybe you do, maybe you don’t. But that’s what goes on. So to build to build, we need to think about things differently. This whole linear approach to where is my data? How do I get it? How do I put it in a format? How do I use it? And then how do I report on it needs to take needs to change? Because what’s probably what’s a problem with that view, is we say, what’s our endpoint, we start with our endpoint. And we go back to the beginning to build to it. What is What are the questions that we have? So have you ever provided a report? Somebody asks you a question. You provided some type of report. And then when they looked at the report, within 30 seconds, they asked you a follow up question to it. Does that ever happen? And what did you think? Or what did you say? You said, Why didn’t you ask that?

Craig Jeffery[23:56]
In the beginning, why don’t you say that was a need that you had at the beginning? And if you’ve asked that question of somebody and someone said that to you, what did you say? You said, Well, I didn’t know. I had that question until I saw this information that informed my thinking. And so I want to know more. And so we can either continue to say, we have this straight line linear process of I know what I want, I figured out what my data is. I put it in places I can use it in structure, and I get a report. And I continue to do this Groundhog Day, or Sisyphus Rolling Stone up the hill only to have it roll back down to repeat that over and over again. Or we can think about how does technology like data lakes like massive repositories of data, leveraging BI tools that let us tap data that we have answer questions that we know about, but give us the environment? To answer those questions that come to us in Treasury. You have a lot of questions. Every time something hits the news. They ask you questions. They ask you to questions about, you know, Russia, Ukraine, the cost of urea for, you know, farm supplies and fertilizer, et cetera, et cetera. So there’s a number of these items that that we need to think about. So this, this changes our expectations, expectations, management has on you change have changed significantly since the financial crisis in 2008. Executive Management, and the board has kept higher expectations on finance and Treasury. There’s more expectations and it didn’t wane like other issues. Other times it’d be an issue and there’s a heightened sense of what’s going on for a few years, and then it dips off, it’s like, few we weathered that. They’re not asking about anymore. Well, it really hasn’t dipped, the expectations come to increase, because they see it in other industries. They see it in their personal lives. And we can accomplish those things in Treasury if we don’t think about them. All right. All right. Thank you.