Treasury's Groundhog Day: Why Automation Isn't Sticking

Joseph Drambarean:
Welcome to Fintech Corner. I’m Joseph Drambarean. I’m joined here by Paul and Valerio. I’m excited about this conversation because we’ve already been talking about AI translation, why KYC is so annoying, why eBAM will never take off. It’s already an amazing podcast, but I wanted to kick off by maybe just introducing why we’re here together. So, Valerio from RSM—Paul you know him already. Maybe I am the third wheel in this conversation. Why don’t you just give us a little bit of context on how this conversation came together?

Paul Bramwell:
Sure. So I’ve been in Treasury for about 34 years, passionate about Treasury and Treasury technology. I joined a consulting firm several years ago; Valerio was already there at that particular company, and we just got on went very, very well. I think our views are very similar, etcetera. It’s about trying to do the right thing with technology for clients. It’s not technology for technology’s sake. It’s there for a business reason. It’s got to solve a use case to be of any value and benefit. Subsequently, I moved on to different places. Valerio’s moved on. And interestingly, Valerio turned up at RSM, a very great company, great firm, and they’re starting to get into advising their clients on treasury.

Brought Valerio into the team and started the conversation to help potentially RSM’s clients get better use out of technology, or if they’re not able to manage the technology, having a partner that can manage it and bring that to them. It’s incredibly important, so hence this conversation.

Joseph Drambarean:
Valerio, what do you have to say about Paul? A very glowing review!

Valerio Trinchi:
I think probably I’m in Treasury like 34 years, minus four compared to Paul. So it’s like always a step behind. I had the pleasure of working with Paul for several years. As he said, we share a passion for efficiency, a passion to leverage technology, and to enable further automation and cost-effective processes in every company for the treasury. Sometimes, this process is resisted, is not as intuitive, or is not adopted as smoothly as it could be. And so in that sense, it’s how you package a message as you preach the word out there in order to build a business case that can be leveraged by the Treasury department internally and be approved to deliver more competitiveness and an advantage to the status quo in their respective companies. 

So I welcome the invitation. It’s always a healthy confrontation between Paul and I, but I think we share the same passion and the same goal: to bring, to enhance treasury processes to the next level.

Joseph Drambarean:
One of my favorite phrases Paul uses is “Groundhog Day.” He uses it every now and then to describe when something has existed for a while. The topic that I’m thinking of right now is just the concept of automation and efficiency, something that has been talked about in treasury for a very long time. And it’s almost like Groundhog Day whenever you hear about it. Why is it still so difficult to achieve these goals? You know, we’re sitting at obviously AFP right now, and everyone is providing their insights on what the best way to do this is. Why is it still so difficult after all these years?

Valerio Trinchi:
I think it’s a combination of reasons. Of course, there is specific peculiar situations in each organization. There is, of course, the typical naysayer. I mean, the way they do things is the only way to do it. There is like, priority, so in like some organizations there are some other priorities that need to be taken care of before they get to specific treasury needs.

There’s budget, cost, opportunity—it’s a combination of reasons. I think there is also, I mean, different stages and levels at which companies approach this exercise. There are people who require something to be mature and tested over and over. Some other companies are more aggressive and try to stay ahead of the curve, and everything in between. So I think you find different characters, thankfully so, because otherwise, if we had to wait for something to be cooked and overcooked before it was adopted, it would be difficult for some of us to make a living, right?

Joseph Drambarean:
One of the things that is a recurring message here is that technology, and specific technologies like the use of APIs or the use of platform services are going to transform the way that treasurers and their corporates interact with making services, but also are able to implement tools that make them more productive, give them more capabilities, et cetera.

How have you seen the proliferation of this technology over the years and is it actually working? 

Valerio Trinchi:
I think, mean, this technology specifically, you mentioned API is something that if you are in the field, you’ve been hearing for some years right now. I would say that there was a huge expectation from the start. I would have expected a faster adoption. But as we were discussing earlier, sometimes there are some, I want to call it barrier of entry, but some initial resistance because the investment required could be significant, not only in the technology itself, but also to sunset the legacy procedures and also the training that comes with that.

And at the same time, will require a massive campaign for the final beneficiary, the companies. So it’s not unreasonable to observe that it may take time for these things to happen. I would say there has been an acceleration in the last few years. I would say in a couple of years right now, it’s becoming mainstream to talk about API. And now the companies, without even you proposing it, saying, or we want that because now they hear from the banks that is available. 

And so I think we’re getting there. So I think we passed the maturity stage or whatever. I mean by that, but so I think in that sense, now we are at the, we got to a mature stage. That’s what I’m trying to say on the API, but I don’t know if.

Paul Bramwell:
I think it’s kind of interesting. I’m a geek and a nerd, so I have to know why everything works from a technology perspective. I don’t particularly care about the technology, I just need to know what it’s supposed to do. And I think an API is something that you consume, you don’t go out and buy one. No treasurer got up in the morning and thought, I better go and buy an API today to improve my life. So what they do is they buy systems to automate inefficient processes. And very often what they’ll do is automate an inefficient process because they don’t do a process redesign before they put the platform in.

Treasury system has been around for 40 years. So I actually used one of the first treasury platforms. There was a DOS version of the product way back in 1989. It was built in 1983. And ever since then, since the treasury technology has been around, people have adopted it. But very quickly, people get to the limit of what they’re able to do with technology. Because maybe you’re not curious, you’re not a geek, you’re not a nerd, you don’t read the manual. Nobody reads manuals anymore. So you buy a piece of software.

Software by its very nature is complex and you need to invest something to get the best out of it. The most powerful treasury system in the market is Microsoft Excel. It will do anything that you could possibly want it to do, but it’s going to take an awful lot of effort to get it done. So I would say 95, well 99% of people use 1% of Excel functionality because they don’t know how to do it. And I think that’s where there’s been a barrier to effective use of technology to automate because you’ve very quickly reached the limit of where people are intellectually curious to get the best out of what they’ve got. 

And so what you see is this constant repetition every five years where every company goes back out to RFP for a new treasury system to replace the one that doesn’t do what they need it to do. But it probably does. But nobody’s read a manual. Nobody’s got training. Staff turned over. They all left. So you end up in this vicious cycle, so to speak, where you are condemned and doomed to keep repeating history, because you never actually do enough to get the best out of the technology. Which is what people like. I like advising companies on how to get the best out of it. And that’s where RSM come in as well. 

Valerio Trinchi:
And if I may, mean, one comment and one consideration of what Paul said. I will probably characterize Excel not as the most powerful, but the most popular. And in terms of consideration, I think going back to the API, there is also an element of a – is the market ready to take full benefits? So API, what it does is basically allow more near real-time data to be delivered into your IT infrastructure. Is your IT infrastructure ready to receive and take benefit of this information? Because if you’re running to PulsePoint on manually message Spaghetti Monster Excel spreadsheet, probably you will not take full benefit of it because you’re not going to have the capacity to refresh this data set or reports, for lack of a better word, enough in order to exploit the best out of having new data all the time. 

So I think there is also an element to give time to the market or to the companies or the adopters. You know what I’m trying to say, to be ready to now fully exploit the capability that this technology is delivering at your doorstep. 

Joseph Drambarean:
It seems to me like there is a war happening between the consumerization of services and concepts just generally on how we should do our work and how the reaction in the enterprise, whether it’s banks or corporates, are taking those principles and adopting them. An example of this is APIs have been around for a very long time. They just happen to be new in the banking world. 

But literally the entire technology ecosystem of the world is powered by programming interfaces that are exposed through HTTP, right? The entire internet functions on it. So it’s not new and how you interact with them is very well documented and easy to understand. One problem though, it does require a high degree of complexity in implementing those interfaces, right? It’s not the same thing as receiving a file on a, you know, an expected timeline, same spot every time, right? 

What we got with host to host or Swift or whatever it might be is effectively the digitization of a paper copy of a statement being delivered to you by mail, right? And instead of that being the case, now it’s just being delivered via SFTP or, you know, before that FTP. APIs is actually a, if you want to think about it conceptually, a quantum leap from that to something else altogether, because a programming interface implies that you are touching the database directly, right? And you know what to do with it. And you know where you’re going to put it. And you know what you’re going to connect it to then from there. I think that that overhead was probably massively reduced in terms of the marketing of the APIs being the greatest thing since sliced bread and probably impacted the adoption early on, right? 

Because if you’re sitting, for example, at RSM, I’m sure that you saw this early on this is awesome, real time, secure, more data. And you’re sitting there saying, all true. One problem, none of your apps can connect to these APIs. None. And by the way, if you’re going to implement the middleware, make sure you have a tech team that you’re going to hire that will maintain that thing for the next 10 years. Otherwise, I don’t know what you’re doing with an API. 

Valerio Trinchi:
I think that’s exactly right. I think you said probably better than me, maybe not better than Paul, but Paul and I used to joke a lot about like a border example or comparison is like if you have like Formula One tires to put on a tractor, what you’re going to do with it. Not much to your point, right?

One or the other is not ready to exploit the best out of the technology or performance that a specific tool can deliver. So I think there needs to be a sort of realignment in order for the two parts of the equation to benefit from each other.

Joseph Drambarean:
Yeah, and obviously I am extremely biased in this topic because the approach at Trovata is APIs are not the end. They are a means to an end of getting better data, more secure data, et cetera. You need to make that easy to consume through something, through software, through an experience, whatever it might be. And so what we’ve done is we’ve almost like an Apple style, reduced the cognitive load of worrying about APIs at all, and said, isn’t it great that you’re able to get better data, more real-time data? And those concepts are made available through user experience.

That’s been my fear all along, you when I walk from booth to booth and I see, you know, pamphlets about, have this new, you know, payment API that can do all of these things. The first thing that I think about is, well, does anyone in your organization know that you have to register two sets of certs, those certs encrypt data very differently. You have to save it in a very specific way. The sessions last only 60 seconds. Make sure that you implement that in a way that it’s, I just laugh when I think about the complexity from an IT perspective.

And IT is not engineering, right? IT is trained to manage systems, not to build brand new systems.

Valerio Trinchi:
I mean, just to expand to what you said, and I know, Paul, maybe you have your own take here, but I think the good news and the bad news is that we get to a level of widespread adoption and availability of this technology that non-adopting becomes now a competitive disadvantage. 

So I think, I mean, there was a grace period of some years we can, I mean, things are two years, five years, what have you, but right now, not adopting certain technology, including API, which also require investing on all the rest, enable to exploit the capability API will deliver in a company, becomes a competitive disadvantage as more and more organization step up and build up and leverage these new capabilities. 

Paul Bramwell:
There has to be an answer to the question, so what? So I think you have to point out one of the practical things that it’s solving for that you need an API to do. So as a treasurer, as a treasury team, am I going to go out and implement an API, do the two calls in 60 seconds? No. Do I even care? No. So what I want is data. I want to be able to get data more accurately, more efficiently.

And so it’s table stakes now that if a vendor has API connectivity, why is that better? What does it do more of that is good for me as a company? And once one vendor does that, somebody else has to do it. And so with the banks, for example, if the banks have APIs that makes it easier to connect to them for the treasury system, the ERP, et cetera. Well, you know what? That’s going to be on my RFP when I go out to look for services. So I’ll ask the question, do you have an API for X?

And if the answer’s no, that’s a text mark that you’re not. So I likely, could be, as you say, it’s a disadvantage to you as a bank not to have it. And I think when that starts to become painful, where people don’t select you because you don’t have the API, which means working with you as a bank is more difficult, then it becomes more of a commodity and people are expected to be there as table stakes. But again, it’s that, as a treasurer, do I care?

If I know that there is a benefit of having API connectivity rather than just SFTP, host to host, whatever, I’m going to put that on my RFP. Even if I’m not quite sure why, it’s going to be there because I know that’s future proof. And if it’s got an API, I’m to be able to connect it to upstream, downstream, to the banks, to do whatever more efficiently with more security and everything else. So that’s where I think it kind of bridges that chasm and then just takes off. And I think we’re at that point with APIs, yeah? And I think banks will start to realize, the smaller banks will start to realize that if I don’t have an API to make it easier to work with my clients, my clients will go elsewhere. 

Joseph Drambarean:
You know what’s funny? I’m thinking about everything you guys are saying. And I’m thinking about the session that we had yesterday talking about AI and how funny it is that we’re talking about AI and APIs in the same breath. And they are on such completely different planes of thought, because in many ways AI systems need data, right? 

In order to function at all to be beneficial in any way. And yet the problem of data has not been solved yet. Right? Spicy take curious, Valerio to hear your take on this. Do you think that we should even be talking about AI?

Valerio Trinchi:
I mean, can we not? Right? I think probably the answer is no, we need to talk about AI. What I usually say when I have a presentation and I do keynotes, there is no AI strategy without, to your point, data strategy. API is the highway to deliver this data to AI, for AI to apply the correct algorithm, appropriate algorithm, I should say, in order to return meaningful and valuable advice, suggestion, what have you. So I think they’re really intertwined. So I think we cannot take one in isolation from any others nowadays. It’s become more complex. And that’s why it’s important that organizations understand that not having a serious conversation internally now to basically enhance their entire finance environment is going to become the difference between surviving or dying. 

Because it’s really accelerating, and I see, you really sometimes even in our line of business, you need to, you know, put a lot of effort to keep up on everything that is happening and how you need to adjust the messaging and also the kind of services that you’re offering an organization to remain relevant, and informed. So I think, I don’t want to be too long hearing the answer, but AI lives on data. The data strategy, the data organization is probably not fully solved just yet. An API is the pipeline that delivers the data back and forth, and probably would need to be enhanced at some point in order to keep up as well. I don’t know how you see it.

Joseph Drambarean:
I wanted to paint two vignettes for you specifically in answer to his comment.

Valerio Trinchi:
He can disagree with me. You usually like to do that.

Joseph Drambarean:
This is an interesting setup for you. So yesterday in the session, what we heard was two stories. Two people came up and asked questions at the end. One person said, we already have AI rolled out in our organization and it’s in the form of Azure, OpenAI, ChatGPT. So it’s just available to us to use through Co-Pilot. We don’t have data for it. So what we’ve been doing is we’ve been taking our spreadsheets that we operate with today, and we simply give it to Co-Pilot and start asking our questions. So this vignette was an example of – loves AI, no data strategy, making do. 

The second vignette was somebody that came up and said – “I’m really worried because all of us love AI. We’ve been trying it, but our TMS doesn’t support it. And we can’t actually connect our TMS to our enterprise system. Does this mean that I have to do another implementation?” This was the question that that audience member asked. And I had to hold myself back from laughing because the answer is unfortunately – “Yeah”. It always actually meant that, you just happen to use AI as the reason. Hearing those two vignettes and kind of hearing Valerio’s response. How do treasurers even navigate this space at this point? It’s almost like cart before the horse, but horse before cart. Like, where do you even go from here?

Paul Bramwell:
So I think if there’s a very famous curve called the hype curve and the hype curve is really looking at, okay, these newer technologies, where do they fit, how do they work? You get to the peak of inflated expectations of what these things can do. Then you get to, I think it’s the trough of despair or whatever it is that basically these things fall off a cliff. And you’re looking for a use case for the technology. And I think we’re at that point with AI where people, companies, treasurers, et cetera, keep hearing it. So it’s constantly being talked up. Three years ago it was blockchain, before it was robotics, and everything else. And so a lot of the time it was these great technologies looking for use case. 

And I think AI has such a strong use case, but I think companies and people, certainly in finance, I think are struggling to find – where does it fit for me, and what can it do? Everybody hears, well, it’s going to take my job and blah, blah. It’s not going to. But it could if your job is a data analyst and you’re just pulling data and organizing it because it could. But what do I do with it? What do I ask it? In the treasurer’s world, you’ve got all these complexities of KYC, opening accounts, credit facilities, debt and investment. Am I going to put all my treasury data into an AI system and say “Should I issue CP today or should I issue a medium term note, depending on the SOFA rate today?”. Now, it might give you an answer if you program it well. if you actually trust that data that comes out and you go and issue CP or a medium term note, or you do an FX deal because AI told you to, you’ll get fired because that’s what we pay you the big bucks for to make these kinds of decisions to analyze all of that data. 

So I think right now it’s a technology that most treasurers don’t know how to fit into their domain to give them something which is useful but also is credible because if I use my AI engine and ask it a question about FX exposures, hedging, netting, whatever, and I get an answer or even forecasting and I act on that and the CFO says “That was a bad move wasn’t it? Why did you do that?”

“Because AI told me to”, you’re probably gonna lose your job.

Valerio Trinchi:
I would say, mean, it would prolong the conversation probably beyond the time that we have here today, but I think it comes down to the level of confidence and credibility that you can give in AI advice, suggestion, right? And it comes down to, okay, you ask Siri, you don’t really care much. Where do I turn? Here or the next one? Fine. When it comes to financial advice, you need to say, okay, what are you looking at? So it goes back to what we were saying – data. right? On what data analysis are you base your suggestion? it becomes, bringing this into the treasury world, maybe, I mean, let’s make like some more practical example, forecasting, which is the typical one that we can see applicability of AI or machine learning heuristic function. 

So how much historical visibility you have in that granular line items that you’re forecasting at different time horizons and what algorithm are you applying, and that you have refined over time, in order to basically reduce your variance, increasing your level of confidence between the bracket of percentage that you say I’m correct plus or minus X percent. And it comes down to iteration, that accessibility, and things that would give, to Paul’s example, a percentage measurement of confidence that that suggestion is correct. 

Are we there yet? I would say most probably not because, to what you said earlier is true. I mean, the entire data dilemma has not been solved. But I think directionally we know the direction we need to go. I think computation capabilities are increasing, I mean, month over month, and data access is increasing, and possibly we will get to that level of confidence, you know.

Paul Bramwell:
Yeah, it’s quite nebulous. Yeah. If you talk to a treasurer, how are you going to use AI? Probably won’t know. They might know that API is connectivity. I can get better data, quicker data. Therefore, I’ve got more data. I can act on it because the money’s in the data. But then how are you going to use AI? And I think it’s a nebulous concept. So you have to put it into practical terms. And that’s where it’s companies like ourselves, I guess, where we’re putting it in practical terms. What do you do with it? So it’s not a case of it’s great because it’s AI. Well, what does it do for me? How does it improve my life? How can I run it? 

Going back to the conversation we had at the start about things in finance and banking, we should ask ChatGPT how to solve the KYC issue. You never know. It may actually come up with actually an answer. 

Joseph Drambarean:
Well, it would have read every opinion that’s ever been written on the topic, so maybe it can come up with an interesting answer.

Paul Bramwell:
But it may short circuit, and actually think there is no solution.

Valerio Trinchi:
Capability to synthesize what is as read is a different story though.

Joseph Drambarean:
Or maybe that is the moment where it becomes Skynet because it realizes it can actually go…

Paul Bramwell:
It becomes sentient at that point. 

Valerio Trinchi:
That’s going to be dangerous.

Paul Bramwell:
It realizes it can actually go open all these accounts on its own behalf, transfer all the world’s funds and then it’s all over.

Joseph Drambarean:
Well on that bombshell, this has been awesome. Thank you so much for joining us, Valerio. Thank you, Paul, for joining the conversation. Until next time. This has been FinTech Corner. We’ll see you on the next one.