Fundamentals of Treasury Management with AI

Redefining the Fundamentals of Treasury Management with Artificial Intelligence

Tracey Knight

Tracey Knight

Director - Solution Engineering, Treasury
Company Logo

Connect on Linkedin

Tracey Ferguson Knight , who is the Director - Solution Engineering at HighRadius with over 25 years of treasury experience, she has helped hundreds of companies select and implemented technology to transform their treasury departments.

Session Summary:

Takeaway 1:
The importance of cash forecasting
Key Points
  • Estimate year-end cash-basis income.
  • Forecast the use of credit facility and interest expenses.
  • Maintain credibility among shareholders.
Takeaway 2:
Challenges with the traditional forecasting process.
Key Points
  • Low visibility into the individual forecasts.
  • Too many variables.
  • No drill-down capability at the client level.
Takeaway 3:
How HighRadius solution improved the treasury system for HNTB
Key Points
  • Automatic gathering and consolidation of A/P and A/R data and bank data from the cash forecasting and cash management cloud.
  • Clear visibility into the cash position along with drill-down capability.
  • Increased accuracy into the cash forecasts.

Tracey
All right and welcome, good morning, happy to be here in New York with you all today, even though the title says, what is it, say redefining the fundamentals of Treasury management with artificial intelligence? It’s really a bit of a case study at our Houston radiance. We had one of our clients speak. And so we’re going to actually talk about that presentation and the results that he, the treasurer of this company, spoke about just last month. So the idea is to talk first a little bit about the blueprint that he had in mind when they started the project. The first phase of this project, which was cash forecasting, the second phase cash management, and then ultimately the results that he was able to achieve. So this client is a leading engineering firm, meaning that they design airports and stadiums like this one, although not this one. They actually designed the new stadium, the Oakland Raiders Stadium in Las Vegas. And so Allegiant is one of theirs as well. They’re based in Kansas City, been around for a long time. So a fairly established company, one that you might think would be a bit hard to forecast for given that there are project based companies, which is kind of an interesting model. One of the other things that made them interesting here just kind of skip over that one is why they want to forecast and why he was so interested in improving his forecast greatly. One was estimating year in income from a cash basis.

This firm is an S-Corp, meaning that they’re a privately held with shareholders but that are closely held in just a few individuals. So that meant they need to accurately forecast their year end cash income that they’re going to be paying out so that the individual shareholders can properly pay their estimated like quarterly income tax. And it’s really more of a planning feature for their business beyond the usual reasons why companies forecast. So a very strong need, and you’ll see it kind of with one in three here is that the Treasurer’s credibility is on the line. You know, if you’re underestimating or even overestimating significantly, then these shareholders are either over or underpaying their quarterly income tax by quite a bit. So it made quite a bit of difference. And of course, they wanted to forecast for the usual reasons. Like most corporations, they still got regular everyday liquidity concerns so so they can plan their borrowing and investing appropriately. Now, one of the things that makes this forecasting for this company so difficult is that they’re usually running about eight 500 different projects at any one time. Now, interestingly enough, this company is U.S. based, so this is all U.S. projects, but quite a few different projects running at any point in time. So for Treasury to be able to properly forecast when cash will be coming in and out of the bank accounts is a pretty significant deal. So the first thing they wanted to start with was forecasting. They went to their bankers when looking for a partner and their banking partners actually suggested that he take a look at how radius and that’s how we became engaged with this company.
Now what we ended up doing was automating the air and AP flow into the HighRadius system so directly from their ERP so that each day or each week, as they were re-forecasting on a weekly cadence, we would get the new open A/R and A/P coming directly into HighRadius, not only open but also closed air and AP. And that would make up the basis of the new data that’s flowing into the system. Originally at the beginning of the project, we start with about two to three years of historical data and use that for our data sciences team to build and train the models, along with bank data all flowing into the system so that we can produce a monthly forecast being re-forecast on a weekly basis. Now they were, as this particular company focused mostly on their quarter ends for the next one and two quarters, so it would be a monthly forecast, but they were highly focused on quarter ends. Now, the implementation methodology is what the treasurer says he was most happy with during the data science phase, the part where we get that two to three years of historical data and build and train the AI models. That’s where our team becomes really intimately familiar with our customers and we do this for each and every different customer that we bring in.
In fact, he said, in some ways, in some ways, the questions that we asked showed that we were understanding parts of their business even better than some of their own employees. So he was very happy with that process that we went through in learning and understanding their business so that we can build these accurate models. Then they go through the user acceptance phase, and that’s where clients get to use and continue their existing process right alongside with the new HighRadius system. So they’re able to see for themselves exactly what the differences are in their forecast and see the results as they begin coming in so that they can see whether or not they’re really happy with the forecast. They can examine and test that accuracy of the forecast for themselves. And then once they accept it, once they feel confident that the process is working and continues, we do the cut over and they go live in the system going forward. So this particular client out of Kansas City has been live now for I think, he said. He went live in January, had been using it for quite a few months prior to then. And so he’s seeing those three, six and nine month results now where he’s able to see how accurate that quarterly forecast is on a live basis. Now, the results are actually startling, startlingly high. And this came exactly directly from his presentation. And so he’s focused on the 90 day accuracy, the three month and the six month those one and two quarter periods.
And you’ll see on the bottom numbers first that the accuracy for the three month forecast is about 94 percent. And then, strangely enough, higher at the six month point, which is 97 percent. Usually for most companies, the closer in you are, the more accurate you are, and then it trails off a little bit as you go out into the future. Their business is highly seasonal, and the historical information is leading to pretty extraordinary high numbers of accuracy, even out six months into the future. And that’s a significant improvement over the manual process they were doing prior. So you’ll see it says 59 percent improvement there over their current or their previous manual process and the three month frame and then an 80 percent improvement over their six month historical forecast. So they’ve seen amazing results with HighRadius. And this, of course, is leading to the credibility that the Treasurer has now with his board, with those closely held ownerships with his own CFO structure. All that is making the difference in him feeling more and more confident about the results that he’s getting from high ratings. So much so that he says, Well, what else can we do? After he saw the results in forecasting, he said, Well, what else do you have? And that’s when we began to look at the cash management, which is more a thinking and focused on that daily cash process as opposed to the longer term forecast.
And so the same thing applies here on the short term basis, the cash management module. We offer two modules here at HighRadius and Treasury cash management and cash forecasting. So then he took a look at the cash management module and said, Well, you know, we have a variety of different problems, most of which come from the fact that we are manual. Everything is being done in Excel. A variety of different data sources one primary bank, but they were logging into different parts and under different user IDs to gather all their information, even from the one bank. And so we’re still a time consuming, very manual process, all being done in Excel. So utilizing our cash management module, he wanted to increase visibility so that any time they could see their cash and cash balances right now and what’s coming up in the near term increased confidence when making investment or debt decisions and then providing that one stop shop for a holistic view of cash. So with the high ratings system now, coupled with the cash forecasting module, we continue to bring in that information and it’s shared between both modules around Air AP, which are the two biggest parts of their cash position and cash forecast, bringing in daily information from their bank both on a previous day and intraday basis where they all flow into the system. And it does two things. One is to create the the cash position so they can clearly see what’s happening at their banks, even on an intraday basis.
And then, of course, it’s also flowing through to the cash forecasting module, so that’s being constantly updated as well so that they can produce their weekly update to the forecast. All of the information is available real time in the system to all interested parties who have who have access to the system, which is particularly interesting because their CFO a rare case where he actually wants to be able to log into the system himself and keeps it up all day long. And so this is just an example of the screen that the CFO, he says keeps up on his screen almost all day, occasionally drilling into more details of where they can look at the Treasury dashboard and get a sense of three primary things where they stand right now. Their daily cash position, both from a previous day and intraday basis, get a sense of the trending which, as you’ll see in the upper right around cash flow trends. What are its total cash looking like? What’s net cash flows looking like over the past six weeks and then seeing the forecast for the next three months? First, at a high level, they’re graphically with call out specifically to the high and the low points in the forecast. Now, at any point, they’re actually able to drill in and get to more details behind that. And so it’s interesting that the CFO, he says, actually keeps it up on his screen all day long and will sometimes call and say, Hey, what’s going on with this? What’s going on with that? They’re not sure if it’s an advantage having this readily available to them all day long or not.
But the CFO loves it. And that’s the idea behind what you’ll hear SaaS you talk about in the next session, which is that autonomous software is this idea that you’ll have easy-to-use software that’s made so that you can create and see business intelligence right from the screen. And that’s what AT&T B talked about. This ability to see the information in a very usable way for executives to be able to drill in and have the information in a way that the actual users can use and most importantly, getting amazing results. Those accuracy numbers, the improvement over their previous process is quite frankly startling. He said, You know, in some ways it seems unbelievable, but it’s true. And so he’s been getting great benefits and it’s showing the advantages of artificial intelligence and the change that it’s brought to cash forecasting that we’re pretty much nothing else had changed for about 20 years. Companies have been doing it the same way, but yet artificial intelligence is bringing a new process, some new changes, and improvements to cash forecasting like never before. And I think this company’s results are a great example of just that. So that’s to give you an example of a real-life live client that we have here at HighRadius, so at this point, I’d be happy to take your questions. Yes, ma’am.

Audience Member:

So cash management tool, you have it coming directly from the bank. But if the bank is already sending it to the company, is there a connection where the company could just send that on to HighRadius rather than have it go directly from the bank?

Tracey:

Certainly, we can gather the data from wherever you have it available. So if you’re already bringing it into the ERP or maybe a TMS, we can extract the data directly from those systems. Or you could just SFTP them to us. So there’s a variety of different options so that you don’t duplicate anything fees.

Audience Member:

Yeah, that’s what I was thinking. Ok? And we do a 20 week outlook. Is there a way to customize the dashboard, the Treasury dashboard to go out more than three months?

Tracey:

Yes, you can definitely go out further. We’ve got a major release coming up in for forecasting in November for cash management in October. We’ll be talking about those and I think two or three sessions from now, and it’s going to give more flexibility to go out further to define your own fiscal calendars, even for like a four or five or some 13 week calendar, if you desire. So lots of

Audience Member:

And I assume that you can drill down to region to entity.

Tracey:

Yes, you’ll see a little bit of that during the keynote address. In the next session, we’re going to give a very, very short demo, but there is also a demo station out here on the concourse, and so I’d invite you to stop in and take a look, and we’d be happy to show that to you.

Audience Member:

One more question. So you mentioned cash forecasting, cash management. Do you guys have like a bank account management tool? Is that coming? Is that existing?

Tracey:

Great questions. Did that slip you the check after the after the session within our cash management module? In the very next release, we are starting to have some bank account administration features like account tracking, where you can track the, you know, open and closed dates, signers, tax ID numbers, basic services and things like that all included in the cash management module. Ok, awesome. Thank you. Hmm.

Audience Member:

Typically what’s the time to implementation from, let’s say, kick off to to go live for a typical project? And when you’re talking cash management, are you also including like the payments module, basically payments tracking all that kind of functionality or that’s on the roadmap as yet?

Tracey:

So when you ask about the implementation timeline where you’re asking specifically about forecasting or cash management,

Audience Member:

Yeah, let’s say a typical client wants to do both. Ok? The completely manual like like the example we had, and now they want to automate. So what’s the typical project timeline like?

Tracey:

Okay, so I’ll give you the usual answer first, which is, you know, it depends. It really does depend on the scope. But the data sciences phase is usually about two months. And so that’s just the start of the project. And that’s just because there’s usually so much data that they have to do some cleaning and normalizing of that data before they can even begin to build and train the models. And so to get a truly great and accurate forecast, it is a bit of a time consuming process where the HighRadius time is what takes up the bulk of the implementation. So on average, I would say roughly four to six months, but that can be longer or shorter depending upon the number of banks that you have, how many RFPs you have, are you implementing in one region or globally? All those things can make a difference. And then you asked about other modules. Right now, we’ve got just cash management and cash forecasting, but we are definitely growing the product. And that’s the direction that we’re growing is to add on additional modules like financial instruments for debt or investment tracking and a payments module as well. In the short term, we’re likely to find and utilize a payments partner. But long term, we would probably, you know, build out the whole thing. But in the short term, just by the very nature of payments, we’ll want someone who’s already extremely experienced with it like a, you know, a swift service bureau partner or something like that. So we are growing and in time will be kind of a full TMS, but always with a heavy emphasis on cash forecasting because of our use of artificial intelligence. It makes us kind of uniquely different than what you’re seeing from most other providers, not using just some kind of profit, profit, or some kind of generic AI model. But truly building custom models for all of our clients that results in the kind of improvement and the accuracy improvement like you saw in this case study.

Audience Member:

Thank you, Tracey, any other questions, we’ve got a few more minutes. So I have a question. So you mentioned that this client was relying on both AP and A/R interfaces to get to the accuracy in the forecast, I think a lot of companies and Treasury folks find that most of the variability is on the A/R side, and you would assume that’s because that’s your customers paying you versus you being able to control the outflows right on the AP side. But it sounds like AP was fairly significant in their forecasting as well. And I’m just curious if we know what was the challenge about that and how I could help. Is it because it’s a project based company and it’s not just predictable like payroll and those types of outflows?

Tracey:

It was a number of things AP seems to come up most for companies, mostly because of their own internal processes and the fact that people are, you know, unreliable. You might have some large invoices come in, but if nobody enters them into the ERP, there’s no visibility into them. And so companies find that even though technically they’re in control of AP if they don’t have visibility into all of the invoices that need to be paid, things often get in at the last moment. And so even though you might have been forecasting for the last month, if something comes in right before the due date, there might be some big swings right before the pay run. And so those make a big difference. Using A.I. were able to forecast not just based on what is open and currently in the ERP, but also augmenting that data with historical trends of what is still likely to happen given the past behavior. And so the AI models are taking into account not just what is already known and open, but also adjusting those based on historical information on what is likely to still come, even though it’s not in the ERP yet. And so that’s one of the things that that lends and adds to that ability to forecast accurately. Did I answer both of your questions?

Audience Member:

Yeah, very well. I mean, so for if the company has a bunch of people like Sean who are late in putting their stuff into the system, it’s actually predicting that as well. That’s pretty cool. I would also assume that in addition to the increases inaccuracy that they gain, that there was probably a fair bit of time savings just in, I think you said, a 20-year-old process they had. Was it all in Excel or?

Tracey:

Yes, it all was in Excel. But you know, there was a little bit of time savings, but not to the point where there were decreasing headcount. You know, like most treasuries, they’re operating very lean. But the idea is that you can shift your time and have more time for the analysis, a little less time on the keying and re-keying and re-keying and re-keyingthat so many companies do. So typically, the ROI, if there is one, is not from headcount reduction, but just more from a reallocation of your time.

Audience Member:

Yeah, that makes sense. Good stuff, so any other questions? Yeah. So from an IT perspective. How much of the work is?

Tracey:

Most of the work is on HighRadius with the current version that you’ll see out on the concourse. We typically take and do all of the static data set up and kind of all of the companies and the banks and the bank accounts and the rules and the variety of different things that have to happen with the new coming version. We still will own the majority of that, but it is exposed to the client in the future, which means that when you have small changes, you’re actually free to be able to make those yourself. You don’t have to come to us for those. So we see that as an improvement from the IT or the company perspective. It really depends on your ERP if you’ve got the big name brand ERP. We maintain and have certified standard connectivity with those ERP, like the SAP, the Oracle net suite, forgetting the other ones. There’s a couple more that we keep standard connectors, too. And so the idea is that we make it as light as possible for your I.T., for those ones that we see the most on a recurring basis where they’ll need about, let’s call it two days or so of their time to do the initial install of the extraction tool where we’re providing all of the code, all of the everything that they need. Now that still utilizing what’s been our historical methodology. We are moving more and more toward APIs as well. And so with our newest extractor tools with some of the the new newer ERP that are all cloud based, we’re moving to API connectors so that the extraction of information can be real time and much easier to set up.

Audience Member:

Coming over here. Sure. Are you using those who are using doing here?

Tracey:

I am not technical enough to say how we do our API calls, so I’m sorry. I’m not the one to ask, but I can check with somebody and get back with you.

Audience Member:

And I would invite you to stop by the demo booth because our head of product management is over there doing demos, and he probably knows the answer

Tracey:

To that one. Right? To be sure, I look for you later on. Can I see what’s your name or your company? Patrick Cramer. Ok, great. I’ll be sure to look for you after I get the answer because I honestly do not know. Ok.

Audience Member:

All right. Anyone else? Ok, well, thank you, guys. Thank you, Tracy. All right, fantastic.

Enjoyed the Session?
Here are some other sessions you might be interested in

Future of Treasury Powered by AI
Tracey Knight

Tracey Knight

Director - Solution Engineering, Treasury
Radiance speaker's company logo
Peter Seward

Peter Seward

AVP, Product Strategy, Treasury
Radiance speaker's company logo
Treasurer’s Guide to Monetizing A/R for Accelerated Revenue Growth
Timothy Batsche

Timothy Batsche

Director – National Client Group, B2B Payments
Radiance speaker's company logo

Experience the Power of AI-Enabled Receivables
& Treasury Applications

Schedule a free demo