Finance Forward: How AI is Shaping the Future of Financial Professionals

In the hushed corridors of financial institutions and the bustling offices of corporate finance departments, a quiet revolution is underway. Artificial intelligence, once the stuff of science fiction, is now reshaping the very fabric of how financial professionals work, analyze, and make decisions.

This transformative wave is the focus of this episode of the SAP Concur Conversations podcast, where host Jeanne Dion sits down with Nathanael L'Heureux, the Chief Solutions Ambassador at Oversight. Their conversation isn't just another tech talk; it's a deep dive into the future of finance itself.

As AI systems crunch numbers at lightning speed and machine learning algorithms spot patterns invisible to the human eye, financial professionals find themselves at a crossroads. The question on everyone's minds: Will AI be a tool that elevates their work, or a technology that renders traditional roles obsolete?

L'Heureux, with his finger on the pulse of this technological sea change, offers a perspective that's both exciting and reassuring. From the automation of mundane tasks to the enhancement of fraud detection, he paints a picture of a future where AI and human expertise work in tandem, creating a synergy that could redefine financial excellence.

But this isn't just crystal ball gazing. The conversation gets practical, offering actionable strategies for professionals looking to ride this wave rather than be submerged by it. How can one become an "AI champion" in their organization? What skills will be crucial in this new landscape? These are just some of the pressing questions addressed in this illuminating discussion.

As we stand on the brink of this new era, this podcast episode serves as both a guide and a call to action. It challenges listeners to embrace the change, to see AI not as a threat, but as a powerful ally in the quest for financial innovation and growth.

You can listen to this episode on Apple, Spotify, or your favorite place to find podcasts.

Finance Forward: How AI is Shaping the Future of Financial Professionals

Find out more about the different AI technologies and how you can leverage them to elevate your business in this episode of the SAP Concur Conversations podcast.

Listen Here

Jeanne Dion:

Hello, and welcome to the SAP Concur Conversations Podcast. Each episode we sit down with industry experts, visionaries, and leaders as they share what it takes to build forward-thinking spend and travel programs. Our goal is to get you thinking differently about how your organization spends money. I'm your host Jeanne Dion. I'm the Vice President of the Value Experience Team here at SAP Concur, and my team works with our customers to drive positive business outcomes based on data-driven insights.

Today I'm joined by Nathanael L’Heureux of Oversight, and we're going to explore a topic that has really dominated much of 2023 and shows no signs of slowing down for 2024, and that's AI, Artificial Intelligence. Specifically, we're going to talk about how AI is transforming the financial functions and elevating the work of financial professionals, how financial professionals can become AI champions within their organizations, and how organizations can embrace AI and the innovation it inspires to bring out the best in their businesses. So Nathanael, I'm really excited to have you here. You're a true AI thought leader, and so I'm glad you're our guest today. To start, would you take a moment to introduce yourself to our listeners?

Nathanael L’Heureux:

Yeah, thanks a lot, Jeanne. It's great to be here. I'm excited as well. My name's Nathanael L’Heureux. I'm the Chief Client Officer at Oversight, and my background is that of computer science. I studied computer science and artificial intelligence at UC Berkeley back in the day, and I worked for a company who's a partner with Concur, it's Oversight and we provide an AI based solution to identify risk, error misuse and fraud within T&E and accounts payable and other programs. So I've been working both on the technical side of AI for a lot of years and also on the value side, bringing that solution to customers. So I've been working with it a lot of years, so I'm very excited to be talking with you today.

Jeanne Dion:

Oh, well, you can't be as excited as me. This is a topic that I really love. And so I want to set a little bit of context for our listeners, at this point in time when you think about finance, so much of the work is manual, almost up to 60%, some people estimate as much as 80% of that work is manual. And then you have a lot of people who are entering into the "great retirement" we've had, and there's a limited influx of new talent into finance and accounting, specifically, fewer CPAs, fewer people taking accounting classes. And at the same time we have shorter term CFOs because more is being demanded of them. The CFO role has really shifted from a historian, which is typically, accounting is kind of history, to being a prognosticator. They really need to understand the past and be able to control the present and predict the future.

But there's also this other side of the coin, which is the fear of AI, that idea that, the robots are coming to take our jobs. When the reality is that, and I say this to my team a lot, I had one of them remind me of it the other day, AI will not replace you, but a person using AI will. And what I mean to tell my team is simply, it's time to upscale. It's time to elevate you as a professional and to figure out how you can increase and better use the data that you have to have a higher level of thinking. So I just wanted to ask you, Nathanael, is there something that you could compare this shift, this AI change to?

Nathanael L’Heureux:

Yeah, absolutely. Change is always a little bit uncomfortable for many of us, as new things have come onto the scene over time, I think that shift, it creates a little uncertainty. So we're used to doing things in a given way. Back in the early days of finance, things was all recorded on paper in big ledgers and the ability to write those things down, it takes a lot of time, it's slow, error-prone. And then as modern technology, especially computers came onto the scene, solutions such as Excel came on board. And back in the day when that first arrived, people were a little bit nervous. They were used to adding machines and they were used to the paper and, "What happens if my computer crashes and will I lose all of this information?" But then certainly today, tools such as Excel and financial systems, ERPs, they're the defacto standard, you have to use it.

And think about the shift in the capability of a single individual in using paper compared to going and using a tool like Excel where they could calculate some so quickly and even estimations, predictions for the future with graphs and charting and seeing trends that, man on paper, that would be really hard. But by using these tools, it really elevated the role of a finance individual going from paper to Excel. And really AI is the next generation of that. It's taking what was powerful tools like Excel and ERP financial systems and now giving you the tools to move that even further. Again, like you were talking about, predicting more accurately the future, whether it's cash balances or even detecting risk. And there are types of AI that have been around and we've been seeing and interacting with them in various ways from typing emails and who get suggestions, but now it's moving into the business sector and how can we take and empower the financial processes to move ahead much further.

Jeanne Dion:

I love how you talk about it as a tool because the reality is that frames it in exactly the way, in a much more elegant way than I did with computers won't come to take your jobs. It's that idea that a tool needs a person to help guide and direct it. There needs to be some sort of human check. So as we think about a lot of these, we're thinking about these new AI based jobs are really going to focus on the AI and the input and then the analysis of what it kicks out. And I know we always talk about AI and ML and LLM and all these different pieces, but would you mind defining, when we are talking about AI and ML, Artificial Intelligence and Machine Learning, what exactly do we mean?

Nathanael L’Heureux:

Yeah, good question. It's important to understand what those terms mean and where they're useful. So LLMs, Large Language Models, are tools that have been built on an artificial intelligence technique known as neural networks. So they're trying to model how the brain works and have all of these levels of analytics that happen internally, but ultimately they're trained ultimately to guess next word is how the large language models work. And most of us interacted with them first in doing probably Google search. You would type in a word and it would give you a bunch of other options out here about what you might be looking for. Or then in your emails, you're typing along and it will suggest the next word. And that's really how they were born. It's built on a technology known as a transformer technology where it was predicting the next word.

And so they found that if they made those models really big and just started feeding it millions and millions of documents, they became good at not just predicting the next word but could then predict whole paragraphs one word at a time and then full documents and in such a way that it sounded very much like an intelligent human. And so all of the biggest buzz in 2023 is certainly around OpenAI and ChatGPT. And those were, again, it's a large language model and it's based on text. So it's very good at understanding texts that you describe or documents, you can put in a whole document and it will understand what you're talking about.

But what they're not good at, and this is important to recognize, is it's not very good at dealing with financial numbers. If you put in a spreadsheet of data, it won't really know what to do with it, it's not built to project or analyze that, but it's really good at analyzing documents like contracts or policies. So that's one thing to be aware of. Different artificial intelligence, which is a very broad category of technologies where computers are emulating the intelligence of a human. So you have something like these LLMs dealing with texts. There's other technologies out there that are working with images, for instance, tightly paired together. DALL·E is one of the big ones out there you could describe to it an image and it will create a picture. Those are two different technologies that are related, but separate AI capabilities. And then there's broader categories of machine learning.

Now these large language models are using a portion of what would be in the larger grouping of machine learning, but there is another area of machine learning that is really good at focusing on numbers, for instance. And will make predictions and will see patterns that individuals, humans can't see. And so when we're talking about a given business process, especially in the world of finances, it's important to understand what AI tools there are and where you can use them and for what. So LLMs are no good at analyzing financial data, but there are a lot of machine learning tools that are excellent at both predicting the future, giving estimations of the future, also identifying abnormalities, fraud detection, very strong at that. And so those are some of the capabilities that's good to understand.

Jeanne Dion:

So if I hear you talking about this correctly, there is a place for artificial intelligence and machine learning in the finance space. It's still going to take humans to interact, because what it does is it basically elevates us, it turns our problem solving efficiency onto hyperdrive. I like to always say, it brings us the needles from the haystack instead of us looking through the haystack for the needles, it's going to present us with the needles and then we figure out what to do with the needles, how do we want to sort them? And so if I think about it that way, from an organizational perspective and organizations making AI part of a strategic focus, is there something that financial professionals can do to really integrate AI into their work? Are there key considerations for the organization looking to incorporate AI into their finance functions, not just in 2024, but building it for the future, building it out for a scalable process?

Nathanael L’Heureux:

Yeah, that's a great question, and absolutely. As an organization and then as an individual, I think there's some good steps that people can take. So within, because the question you asked is really big one here, let me try to break it down a little bit.

Jeanne Dion:

I'm a big question person all the time.

Nathanael L’Heureux:

So just to break it down, I think it's the value of what you were talking about with artificial intelligence is a tool. And the first step that I would recommend for finance professionals is to begin to learn the various AI tools that are there and play around with them so that they can have some conversational capability and understanding.

But then when it comes to, how could a finance professional become a champion or an expert or a hero within their company, and I've seen this happen in a number of different times, and here's a rough outline. So what I would say is, think about some financial processes that you are working with and ones that have either a lot of manual steps or maybe it's just a process that you have responsibility for, just since we're talking about in the context of Concur, you have your expense reporting, you also have invoice processing. So both of those are areas where spend money is going out the door and there's risks involved. And so as an individual, let's decide to focus on a particular area. You could focus on travel and expense or on invoice or both. Let's say you want to make some improvements, there's manual processes there, and how can you improve them?

Well then the first step is, figure out where you want to attack. And the next step would be, what would be some of your goals? Would it be to eliminate waste, identify risk? Would it be more for projecting future outcomes? Would it be controlling out of policy spend in travel and expense or identifying abnormal expenses in accounts payable or external fraud or making predictions?

So as you think about where you would want to choose the domain of one of the business processes and then figure out what would be your goals in this process, then look at the various AI technologies. So if you're going to be looking at documents, look at some of the large language models. If you're going to be working with finance numbers, then some of the machine learning capabilities. There's other things like image processing, computer vision, that's part of artificial intelligence. So there's a number of these capabilities. And then think about, okay, so which of these capabilities would fit to what my goal is? And then there's looking at those tools that are out there, you can kind of say, "Hey, am I going to build this myself?" Well, probably not. There's a, I mean-

Jeanne Dion:

Whoa, yeah.

Nathanael L’Heureux:

Right. So you don't go around and build your own Excel, you want to find a provider. And look at solutions, right? Obviously on the T&E environment, Concur has offerings and Concur Detect by Oversight. And so look at the various vendors out there to help you and understand what their capabilities are and then apply that.

So I've seen some individuals who have done that and who said, "I'm going to try using AI in my organization and say, approach the T&E side, or T&E and accounts payable," and they selected a vendor. And some of our customers I've seen that the individual's, one of them I'm thinking of right now, project was very successful. The CFO was so excited about it. He said, "How can we take this and use it in other places of the company?" And so now this individual who spearheaded this project now is seen as a thought leader across the organization for, how can we deploy this or other things using AI, so he has sought out. And one team that was an accounts payable team, their whole team got recognized by the CEO, the employees of the year, and they were publicly recognized. So championing this kind of a thing can be quite beneficial, like you were talking about, elevating your role as an individual, but then bringing the benefit to the entire company.

Jeanne Dion:

So I have one other big question to ask you, so hang onto your hat. As I think about AI, is this a chance for our finance teams to actually democratize finance data and really put it in the hands of an end user? It allows them to make better decisions in the moment. Because too often, I alluded to it earlier, we're typically historians, right? In finance, we're reporting on things after they happen, not while they're happening, but after they're happening. And sometimes when you get that information, it's too late to make the change. I think about budget and actuals, it's too late for me to put the toothpaste back in the tube after I've spent it. So is this a way for us to get some more democratization that could have an impact on everything from say employee empowerment to spend governance to even bottom line profitability, cash management strategies. These are all things that I think about, my mind gets all excited about. Is this something that you see as well?

Nathanael L’Heureux:

Absolutely, yes. So several areas. I think if you've played around with ChatGPT at all, one of the things about it is that it's very friendly, it's easy to use. I would say the learning curve is much easier than even Excel. You don't have to learn formulas. You're communicating in plain English. And this is really where a huge transformation is happening with individuals who are not experts being able to be given information and a starting point by speaking just plain language. And these GPT interfaces will be interfaces that intersect with a lot of backend data.

So just to give a couple quick examples. Right now, let's just say you are an employee traveling, and take a simple use case, a language use case, like a travel policy. Every company has one and they're usually, oh, I don't know, a hundred pages long. Nobody reads the whole thing, and it's like I have no idea what all the details are in there. But let's say they have a question, you know the traditional way, they'd have to either try to read that big old document, which would take forever, or, and let's say I'm traveling, I'm in India and I don't know what my meal limit is in this city, I have no concept. I could call somebody and then they would've to sort through this document or who knows, and maybe it's time zones, I might not be able to get ahold of them.

Well ChatGPT interface, you can feed in that whole policy document, ask it a question, very specific like that, and it will give you back the details of just what you want it to know quickly, automatically. You don't have to program that, it just does it automatically. So that's one basic text document capability, and we're playing with that and have use cases of that right now.

And then another one would be on the finance side. What if you are a buyer and you're wanting to make an understanding of what vendor to pick? I've got a couple vendors, and what about budget here and what about future? So being able to ask a question like, "If I make a purchase for $5,000 of this type of object, what would be my best vendor?" Asking it in real plain terms because now these ChatGPT modules can sit on top of financial data and give back to you very quickly, now you don't have to go looking it up and running some analysis in Excel, all of that-

Jeanne Dion:

I was going to say, using an Excel spreadsheet.

Nathanael L’Heureux:

That's right. That's right. Which they're great and they're powerful, but man, it's like, "Where did that Excel spreadsheet go?" You've got thousands of them. And so the nice thing is that it will bring together a lot of data and put it at your fingertips. A manager could ask a question like, "What's remaining in my travel budget?" And it goes out and looks it up and they don't have to go searching. Or, "What is cash projection?" All of those kinds of questions now can be fed into, the data is provided into a ChatGPT back end, and now users can ask in plain language and have access to that data. That's where a huge transformation is going to be happening. So that you're not Googling, you're not looking for that lost document somewhere that you can ask a simple question and have access to it very quickly.

Jeanne Dion:

Well, I think you've kind of led me exactly to where, I don't know if you know this about me, I try to synthesize our conversation into three takeaway points and it sure would be easy if I was using ChatGPT and just fed this whole thing in and it would spit it out for me, but I have to use my brain. I have to use my human brain and do some manual work here.

So I want to run by the three things that I've picked up from our conversation today and see if you think I'm on target with it or not. And I think the first one ties back to what you were just talking about. It's that idea of using AI to help interpret or pull out really relevant nuggets of information so you're not spending a lot of time trying to find the information. It gets to that question of, so what? And helps people to really understand the why and then understand what they need to focus on to either correct something, to expand the conversation about something, to become a little more predictive about how things are going, to pick a better vendor, in your example, for what I'm trying to buy. Is that kind of where we're going with this?

Nathanael L’Heureux:

Yes. Yes, I do believe so. So both the generation of immediate information that is now synthesized and summarized for consumption to really help an individual make better decisions.

Jeanne Dion:

The other thing that I took away from this was, and this was really fascinating to listen to you say it, so I'm going to distill it for the audience and see if this is the same for you. When you're looking at installing or using AI as a strategy within your organization, you follow the same process you would for any other tool because it is a tool, and you need people to work it. So you're going to look at your process, you're going to look at where this would have the best outcome for your organization. Then once you understand what your process is and what your goal is, then you look at the tool you're going to apply to it. This is no different than any other tool that we have, right? It's not going to solve the world's problems if we don't pick the right tool to solve the problem.

Nathanael L’Heureux:

That is absolutely right, yes. AI is not going to be washing our dishes, so don't... yeah, so exactly right.

Jeanne Dion:

Darn it.

Nathanael L’Heureux:

I know. Hopefully someday.

Jeanne Dion:

Oh, man. And then the final piece is, AI is really still new. We've been talking about it for a long time, and we've actually been using it for a long time, let's be honest. If you're like me, you've been using it a lot on your text corrects. And as you mentioned, we're using it in Google, it finishes your sentence. If you shop online, all those people who know exactly what you're thinking in your head when you want to buy a new pair of shoes and all of a sudden everything pops up. These are all tied back to some of these AI pieces and tools.

So while it's here every day in our lives, it's new in our business lives, we haven't been seeing it much in our business world. So you mentioned the people who were forward thinkers in this, some of your customers who are forward thinkers and the way that it moved them into thought leadership. So I think my final takeaway is, be a thought leader. This is something that you can bring to your organization to make a huge difference in your area, and it can spread. So don't be afraid of the technology. In fact, be an evangelist for it and become that person who becomes expert at it so that you can teach the rest of the world how to fish.

Nathanael L’Heureux:

Absolutely, yes.

Jeanne Dion:

Oh man. Okay, I hit it all. I didn't even need ChatGPT to do it. That's great. I want to thank you again, Nathanael, for coming. I really appreciate you being here. I love this conversation, and hopefully we can talk again another time and expand a little bit more into the AI capabilities that you're seeing.

Nathanael L’Heureux:

Well, thank you, Jeanne. It's been a real pleasure. I look forward to future conversations as well.

Jeanne Dion:

Yeah. And thank you for listening to this episode of the SAP Concur Conversations podcast. To hear more exclusive insights and interviews from the world of business travel expense and spend management, be sure to subscribe and listen wherever you find your podcasts. And please join us again for our next SAP Concur conversation.