Transcript: Can Treasury Tech be Enjoyable to Use?

Joseph Drambarean:
All right, welcome to FinTech Corner. We’re here at AFP. I’m your host, Joseph Drambarean, once again one of many hosts this year. I’m joined by Bruce. Bruce, we had a panel, was it yesterday?

Bruce Edlund:
Yeah, it was yesterday.

Joseph Drambarean:
It was on AI, and the topic of today’s conversation is the future of TMS, which I feel like AI has something to do with that. But before we get into it, I was hoping for the sake of our audience, you could introduce yourself. Tell us a little bit about your background, what you do today. How do we overlap? How did we get to the point of sitting down having a podcast after also having a panel together?

Bruce Edlund:
OK. Let’s see. So let me go through my background a little bit. So I’ve been in corporate treasury for, like, 20-plus years, in smaller companies and also a very big company called Walmart. And that had a big impact on me because I spent four years at Walmart, and I did two Swift Service Bureau implementations while I was there. Plus, they already had a lot of host-to-host connections. And what I came away with from that experience was what really hit me was we in treasury sort of get tasked with these connectivity projects like that. But we didn’t seem to get a lot of the benefit from it.

It was more like the AP team is getting their batch payment file done, the accounting team’s getting their previous day bank statements, and the treasury team, are we getting current balances? Well, we could order intraday bank statements. Are we doing payments? Yeah, but they don’t really work all that well. I really came away from that experience, like, not liking it that much and thinking, you know, hoping there’s something better. So then, fast forward, I went back to Citrix after four years at Walmart. They asked me to come back. We had no connectivity, and, you know, I was going to do sort of the standard thing that everybody does. This is like sort of the late teens, so like 2016, 2017, then the project got delayed.

And anyway, when someone finally explained to me how an API works, it really clicked with me because I remember the file process and all the stages that these text files go through, like being on a network somewhere and then going through the service bureau in and out to the bank. And so many times, like somebody saying, “Where’s the file? Did the bank statement come in?” And the API is just on. And it’s going back and forth all the time. And that made perfect sense to me. And so I have a strong bias for APIs because I really think it’s the right technology.

What I really want… I mean, we can talk about a lot of, you know, different features of a TMS. But what I really want, I think, is so basic. I want like a generic bank portal experience, right? We have our bank portals. We have so many of them. We log in, and the data is pretty much real time, and we can do a payment. It’s basically updating all the time, and you log in, you do something, and then you see a result. And that’s the kind of experience that I want. And I feel like we can only get that with APIs. So that’s kind of, like, why I’m… why we’re sitting here.

Joseph Drambarean:
You also happen to be a customer.

Bruce Edlund:
I am a customer.

Joseph Drambarean:
And I have a feeling APIs had something to do with that.

Bruce Edlund:
It was a very, very big part of it because, you know, APIs are kind of a buzzword that’s been going around for a few years. And I feel like other providers sometimes have maybe tried to latch onto it, but maybe they’re not really there. And others have kind of blown it off, like, “I don’t really need it, whatever.” For me, it’s very big. Like, I feel very strongly that APIs are the solution for the reasons that I talked about. And so obviously, Trovata being built on APIs was a big factor for us. But, I mean, there are other factors too. I mean, it’s a newer interface, which is, I think, a really great user interface. But also, like, implementation was super easy. And I talked to your customers about that, and that really impressed me. We did, as you probably remember, we did a proof of concept. It was only three weeks.

How many TMS providers could you do a proof of…? I mean, you couldn’t get up and running anything in three weeks. We signed some forms…signed some forms with one of our major banks, and we started getting data. The implementation being easy, the interface, and I think a good value in terms of cost. Really all those things.

Joseph Drambarean:
You know, it’s funny hearing all of this. It’s obviously a very validating experience for me personally, because we made so many big bets at the beginning of Trovata. A lot of it was based on a fundamental principle. It actually had nothing to do with APIs. It had nothing to do with any of the technology. It was actually this principle of, “Can we approach this problem from a human-centered design perspective?” That sounds really obvious.

But when you think of other systems, they don’t always have that principle at heart, right? One of the main drivers of their design is how can they convert and extract as much value out of that sale through the process, right? So examples of that are the brain damage that you were just talking about, like going through the process of connecting. In many situations, that’s actually by design.

It’s by design because that allows for more parties to be at the table and provide their services, whether you’re a consultant or whether you are the actual provider that is able to charge a fee to go through that process. You flip it on its head to human-centered design where, you know, on the consumer side of the world, when you open an app on your phone or on your iPad or whatever it might be, the design is typically centered around your experience. Why? Because they want you to love it. They want you to be attracted to it and to winsomely use it because, at the end of the day, the more that you use it, the more eyeball time they can get, the more value they can extract out of you throughout the lifecycle of it. But that puts the entire bias on your experience being self-serve. You’re not having to interact with another human at all to get any value out of it, right? Those principles are important.

On the consumer side, and for whatever reason, they have not made their way into the enterprise. Right? I think there’s a very specific reason. And that was actually one of the big foundational bets of Trovata—well, what if we just do this fundamentally differently? What if we don’t charge implementation fees? What if we try to do it as fast as possible? What if we have this insane goal of near zero, right? If it’s possible that you could connect the day of, we’ll do that. If it’s not, we’ll work within the parameters of what the bank offers on the API side.

Anyway, the reason I’m telling this story is because I’m curious to hear this side of the story and comparing it, kind of comparing and contrasting notes—your career over the 20 years of having to see Walmart implementation, what happened at Citrix, what happened before that. What do you think about when you hear kind of this type of approach? Do you think that this is the right approach, first of all? And what do you think it means for the future?

Bruce Edlund:
Yeah, I mean, I think it makes perfect sense because, like you said, I mean, as consumers, we’ve got these great apps, and they’re super easy to use. Why shouldn’t we expect that on the Treasury side? Like, everything’s such a mess, like on the corporate Treasury side. And I feel like the TMSs over the years, you know, I guess they were built sort of maybe as best as they could have. I mean, they’re basically just big databases, right? And they have all this data, and they just kind of do their best to do the calculations that you need and do some reporting.

But it’s super easy. I’ve built out stuff in Excel, and access databases, Power BI, Tableau, and everything. And it’s super hard, like trying to combine a lot of this data. So, like, I get some of the challenges, but I just feel like a lot of the stuff that was created in the past was just, “We got this engine, and we’re trying to figure out how to give you a dashboard or some reporting or whatever.” One of the other things I really like about Trovata, besides just being a nice, clean, easy-to-use user interface, is it’s easy to get the data out and do our own reporting.

One of my pet peeves with TMS providers that I’ve given feedback over the years is, “You can think you have the best reporting ever, but we’re always going to do our own stuff.” I mean, somebody’s going to want something else. So we need to be able to get the data out easily. And that’s one thing that we found super easy with just the grid thing, copy-paste, or even using an API with Trovata and things like that. Just making everything easier.

Joseph Drambarean:
Believe it or not, that was also part of this whole thesis of, “Let’s actually respect the user and how they want to use their data. It’s their data.” That’s actually a crazy concept. It’s not our data. We are simply hosting it for them. We are a cloud service provider that is giving you this service. But if you want to take your data and leave, it’s your data. Just like an example of this on the personal side is, I always am wary of sharing my health data. 

I’m an avid runner, and I like to use apps like Strava and other apps that collect all this information and tell me things about myself. But then the thing that I worry about is if I put it all in there and I decide I want to go to this other app, are they going to lock me in? Are they going to say, “Sorry, you can only keep your data in here if it’s in Strava. But otherwise, good luck.” Right?

Bruce Edlund:
Have you figured that out, by the way, with Strava? I’ve got like 14 years of stuff in there.

Joseph Drambarean:
Unfortunately, that’s the thing with Strava. They create that walled garden. And that’s why Apple Health, for example, is exciting for me because it’s my data, and I want to control it. If I want to export it all at once, I can. If I want to put it in another app, I can. It’s fully in my control. And that human-centered design principle at Trovata, it goes across the whole platform. It’s not just the user interface, right? It’s our APIs. If you want to build your own app, if you think that the Trovata app is cute but not what you need, you could rebuild Trovata if you wanted to from the ground up with our APIs. We’ve actually externalized our search APIs, our analytics APIs, our forecasting APIs. And it’s there because we have this foundational principle of, let’s say you had Tableau.

And you’re really comfortable in Tableau. And you think that Trovata is great for certain things, but Tableau is what your team uses. Great. Use Tableau. Use Tableau and get the performance of Trovata foundationally, right? All the search capabilities, all the analytics getting rendered. And I think it’s an important shift that has to happen in enterprise. A lot of the team at Trovata, believe it or not, is not from enterprise.

Almost all of us come from the consumer side of the world. Myself, I was at Capital One. A lot of the data transformation and redesigning of Capital One’s banking products came through myself and my team. And it came around this principle of, “Why does it have to look like this? Why does everything have to be so complicated?” I just need to make a payment. Why did it have to be five different tabs that open, then authenticate me, then I jump back into this tab, then I select the thing that I want to pre-fill. But if I don’t have it pre-filled, I have to open another tab to add it to my list of pre-filled contacts. Then I save it, then I go back to this tab, then I send my payment. It’s like what?

Bruce Edlund:
So this is a really interesting conversation because I didn’t realize it was so foundational with Trovata, but I love it because I’ve been giving this kind of feedback to TMS providers. Or like, you know, there would be these consultants or research people that want to hire you and get your opinion on stuff. But, like, one thing that I’ve said over the years is it’s almost like we need a Steve Jobs to come together. The technology’s out there. Like, someone just needs to come together and just make it simpler. Like you said, we don’t need six steps. We want security and all that, but just a couple.

Yeah, things that… let’s just make it as easy as possible. I log in. I love just going into Trovata. I mean, we still have our bank portals and everything, but it’s just so easy for me to just log in to Trovata and search for a transaction there or even, I mean, it could be a balance or whatever. Like, I don’t necessarily remember off the top of my head which bank I’m using in Japan or whatever. I just go in and look for Japan, and these are my accounts. You know, like things like that. Just the ease of use. It’s just really…

Joseph Drambarean:
You know, what’s also funny about what you just brought up is that, from a product perspective, your founding principle on that piece, right? “Hey, I just want to find something,” right? I don’t even know what I’m looking for. Let me just start typing, and I’ll figure it out as I’m going. Foundationally, the software…That’s one approach, but then the infrastructure that says, “I’m going to let you be really loose with what it is that you’re trying to find and not rigid at all,” right? Fuzzy searching, Google search principle, right? One that is so bedrock at this point in our society that you could just go to Google or you can go to ChatGPT or any other service, type whatever you’re looking for in natural language, and expect a result, right?

It turns out that if you want to build things in that way, you have to respect that as an end-user goal. You had mentioned that a lot of these older solutions, right? They’re kind of just databases, right? Databases that were designed a long time ago and that do what they do, right? I actually have a ton of respect, believe it or not. When I go around this hall and I look at some of the service providers, I have respect because I am not a practitioner expert in what they do, and there are decades of accumulated expertise that they bring to the table. What’s missing from the equation is using the most modern approaches that we have, right? Let me ask you this. If you go to Twitter or if you go to Instagram or any of your favorite apps on social media, you type something, do you expect it to show immediately anything that you’re looking for?

Bruce Edlund:
Yeah. I mean, it’s like… almost like it fills in the rest of your search.

Joseph Drambarean:
If you type in a hashtag or if you type in somebody’s name. There are literally billions of users of Instagram. Billions with a B, right? We’re not talking about hundreds of thousands of records, like in the case of a TMS; we’re talking about billions. And that same level of performance of “I think it, I see it, I type it, immediate interaction, I go to the next thing.” The reason it’s that way is because they designed the system with that assumption in mind—that you don’t want to wait. Because you could wait, right? They could have a system similar to the TMSs where there’s data over here in this database, another one over here, another one over here. “We’ll get what you want and just give us a second. We’ll get it for you, we’ll give it back, and you’ll be fine.” That would be a thing if it was in the context of Instagram. But it comes from this philosophy of, “Well, why would you wait? We have cloud computing. We have effectively infinite capacity when it comes to servers at this point. Why would you have to wait?

Bruce Edlund:
Well, and it might not even be the wait. It might be you have to… you’ve got to write like the Sequel Query or you’ve got to put it together. I mean, I remember the TMS we were using at Walmart, and the best feature that I liked about it was that I could just connect to the data tables. Be in an Access database.

Bruce Edlund:
That, to me, was like…

Joseph Drambarean:
You could hardwire.

Bruce Edlund:
I could basically do what I want. It wasn’t the reporting, it was just giving you access to the data. But then, I’m in Access data… I’m writing queries and building reports, and that’s not easy. So everything now is just so much easier. I consider myself a techie person. I can code a little bit, I can do databases and all that, but you don’t have to be that person today. I think that came out in our session yesterday. You don’t have to be that techie person with AI. You can ask natural questions, and you can get to answers. And I think that’s really promising for the future.

Joseph Drambarean:
Beautiful segue, because I was going to pick your brain about this. There is a big debate happening internally at Trovata, and I’m curious to get your take.

Bruce Edlund:
Whose side am I going to be on?

Joseph Drambarean:
AI, when it came out in the form of this open-ended, large language model-driven approach where there’s effectively global knowledge of a variety of topics, finance, treasury, etc. We talked about this in the AI panel. What we’re struggling with, and we continue to struggle with, I think together with our peers in the technology space is, what does this do for user experience altogether? We’ve spent a lot of time crafting the search experience that you’re talking about or the reporting experience that you’re talking about. But does that kind of just evaporate if you become very familiar with the possibility that anything that you’re thinking of, you could just type it or speak it even into an interface, and it will adapt to whatever it is that you want out of it?

And present to you a report or a table or an answer or a strategy or a resource that you could go to to find out more. And we struggle with that because, on the one hand, it’s very breakthrough, right? It really reduces the cognitive load of someone that maybe doesn’t have the expertise. There are so many folks that we work with at Trovata that are part-time treasurers. They’re not full-time, right? They’re in finance roles, and someone has asked them, “Hey, can you manage our accounts and actually just get us a little bit of advantage and yield out of them?” And they’re like, “Quick Google search—how to be a treasurer,” you know, “what do I even do?” So for folks like that, there’s a lot of up-leveling that you get naturally by having access to tools like an AI-generated system that gives you the advantage of all that knowledge.

But for folks like you, where you already have that expertise, you already know what you want, you already have the reporting in place, the infrastructure is there. How do you see the interfaces of AI specifically? Because we all know we love AI. You talked about it in the panel. I love it, you love it. But what about the user experience approach of it? Because this is something that is a huge tension point, I think, in the industry around the world. And I’d just love to get your take.

Bruce Edlund:
I’m not sure if I understood exactly…

Joseph Drambarean:
I could try to rephrase it.

Bruce Edlund:
Yeah.

Joseph Drambarean:
Here’s the problem, basically, is that we keep building features, and it’s hard to build features. Months, not years of investment, design, coding the features, testing them, getting feedback from customers. In an AI prompt environment that’s chat-based those features kind of just come to life through the conversation.

Bruce Edlund:
Sort of like you’re building out the features versus AI kind of just figuring it out?

Joseph Drambarean:
Exactly.

Bruce Edlund:
That’s tough. And I guess we’re kind of thinking about what the future could be. I don’t really know. I mean, I’m assuming we’re going to get to a point where AI could probably figure a lot of these things out. But at the same time, I think as treasury people, we’re pretty conservative. We’re managing the company’s cash. So I think we’re always a little bit hesitant to just take what we get. Even when we run reports in a TMS, I think we’re always kind of checking, “Does that make sense?”

So I don’t know the answer. I do feel like over the next few years still, I think we sort of want the TMS to do some of that work, because I think we would trust it a little bit. If AI just goes out and creates something with my data, then I’m going to be thinking, “Is that really…?” Then I kind of have to take it apart and make sure it’s really what I wanted.

Joseph Drambarean:
This is a tough question for everyone, I think, because it forces you to think about what does the 10 years from now or 15 years from now practitioner think about or do in their day-to-day. Because I’m assuming by then connectivity is not something that we’re going to talk about anymore. I really hope that connectivity is something that we’re not going to be talking about.

So if you assume the playing field of everything is connected, it’s easy to do, we’ve flattened the entire onboarding to nothing, it’s all plug-and-play just like your consumer apps, the data is democratized, I can take it wherever I want, I can use it in any way that I want… the tough thing about all of it is will the future practitioner treat a TMS as a workstation or would they rather treat it as a resource? A person, a never-tiring thing that gives them specifically what they’re looking for, so that they can just do their job of being a strategist ultimately that informs the organization.

Bruce Edlund:
Yeah, I think it’s hard because we are dealing with, you know, complex companies, lots of data. I mean, a lot of times the answers that we’re looking for are big answers. It’s not just, “What’s global cash today?” You know, it’s like, “Well, what’s cash like in all these different countries?” And “Why did this and why did that?”… And so while I feel like AI or features could get to the point of answering a lot of the questions, you know, when you have meetings with your treasurer, CFO, or the CEO or the board, it’s always iterative. Right? So you give an answer, somebody asks a question, “Well, why is it this way?” “Why did cash go up like that?” or “Why is this country so high?” Who’s going to answer those questions?

And I think that the AI can help get you to the answers and probably give you sort of like this cheat sheet type thing, like tables of data. When we give global cash, we give it by country, and then we can split it out from there by entity or whatever. You’ve got to have that underlying. Because if you get a question and you can’t answer it, then, like, what are you there for? So I think there can be answers, but the technology has to disaggregate as well and show, “This is how we got to it.” This is the data that we’re using so that we understand all the pieces to the puzzle because these global companies are so complicated.

Joseph Drambarean:
There are complex systems ultimately that AI could probably extract insight, but they are not controlling them and strategizing what to do.

Bruce Edlund:
Yeah. I mean, it’s like the intern analogy, right? I mean, you’ve got this intern, and they work 24/7, whatever, but you’re always checking the work, right?

Joseph Drambarean:
They don’t have the keys.

Bruce Edlund:
They don’t have the keys, and you’re just always checking the work. And not just checking the work, but really trying to understand, “Why is this answer like this?” How could it change if this other thing changed like, you know, rates or FX or whatever?

Joseph Drambarean:
I think that’s actually a perfect segue. Thinking about the future and all of this AI stuff, let’s pretend you’re the czar of AFP for a second. And you have all of these brands, all of these companies at your disposal, and you get to wave your magic wand and make change in this industry right now, starting today. What’s your top item?

Bruce Edlund:
KYC. I mean, honestly, like we were joking at Cloud because I think Rob and our team said our jobs will never be replaced because of KYC. Like these, you know, whoever the regulators, countries, they make it so hard for us. There’s so much paperwork and differences and wet signatures and electronic signatures. Like, that is really the hard part now. A lot of the stuff that we could digitize, you know, we automate that as much as possible, but it’s all this other stuff that’s so hard. And so I don’t think the magic wand will probably ever fix that unless maybe, I don’t know, blockchain…

Joseph Drambarean:
Some sort of digital contract.

Bruce Edlund:
Yeah, there’s got to be, like, why don’t we have digital identities for all these legal entities? And they should have all this KYC information out there, and all the banks could access it. I mean honestly, that’s my biggest nightmare is KYC. I just want to… you know, I think like 10, 15, 20 years ago, we were all about, “Don’t have all your eggs in one basket, diversify a little bit with your banks and stuff.” I’m at the point where I could care less about that. Like, I need to simplify just for fees and for KYC. I don’t want to have all these banks.

Joseph Drambarean:
Again, that’s the human-centered design. When you sign up for a new app, I’m assuming you use “sign in with Google” or “sign in with Apple.”

Bruce Edlund:
Yeah.

Joseph Drambarean:
KYC is handled. These companies know who you are. They have everything there is to know about you.

Bruce Edlund:
Yeah, probably too much.

Joseph Drambarean:
And that is kind of part of this whole philosophy shift, right? We need everybody to think in that way. Right. KYC is almost intentionally designed to be…

Bruce Edlund:
I think it is. I think it is. Yeah. 

Joseph Drambarean:
Well, this was one awesome conversation. I did not expect it to go all the different directions that we took it. But thank you so much for joining us, Bruce, and hopefully you enjoyed the conversation.

Bruce Edlund:
I did.

Joseph Drambarean:
We’ll see you next time on Fintech Corner.