Leveraging Generative AI for Revolutionising Travel & Expense Management: The SAP Concur Approach

Wondering how SAP and SAP Concur are leveraging generative AI to revolutionise spend management? In this video, Rahul Vig, Senior Director of Product Management at SAP Concur, demystifies the concept of AI and discusses its role within today's organisations.

In the video above, and the transcript below, we answer four core questions:

What is Artificial Intelligence (AI)?

What is SAP doing in the AI space? 

How do SAP Concur solutions leverage AI?

What's next?

Alistair Kent: My name's Alistair Kent and I look after public relations for the UK here at SAP Concur. We've got a really interesting chat today. The topic is artificial intelligence, what it means, how we as a business, SAP Concur, are using it, but how the wider ecosystem uses AI and what it actually means.

Really lucky today to have Rahul Vig with us who is the expert on AI, I think you'd say. Can you tell us a bit about yourself, and what you do here at SAP Concur?

Rahul Vig: Sure, Alistair, and thank you for having me here. My name is Rahul Vig and I've been working here at SAP Concur for over 16 years. And my role here right now is the senior director for Concur product management. I look after the strategy for our Concur Invoice and Concur Expense product suites.

What is Artificial Intelligence (AI)?

AK: Okay, brilliant. So, you're well placed to talk about artificial intelligence and what it all means.

I know it's a massive topic and we're all reading it in the news. Big companies are all saying, "Yeah, we've got AI this, AI that." But can you explain what it actually means? What does it mean to the general person on the street or in a finance department, for example?

RV: Yeah, it's, you know, if you go out today, artificial intelligence, or commonly AI, it's the buzzword everywhere. Everyone's talking about it, you know, but really, if you go down to what is AI, when I talk to a number of people, they often mistake automation with AI.

And I think if you think about automation, it's actually following a series of steps. So, it aims to reduce or eliminate tedious tasks that machines can do. And that's not what artificial intelligence is.

Part of automation can be AI, but not in general. But when we talk about the ability to think like humans or mimicking the thinking, that's where artificial intelligence kicks in. So back in the day, we were doing things like trends analysis or predictive behavior.

That was the earliest forms of AI. And then of late, we started looking into machine learning, which actually became the bigger form of artificial intelligence. And think about it this way, humans cannot actually code for everything, right? We need to have machines who can build the algorithms and then see the data coming through to them and optimise accordingly.

That optimisation, that's what typically humans do. And when machines start doing it, that's what we call machine learning.

So, of late, that's what we've been doing. But then the advancements in AI have taken a further step with generative AI, and that's absolutely new and huge and presents a whole set of opportunities for us.

AK: And that's what everyone's talking about in the media with ChatGPT exactly?

RV: That's right. That's right. You know, people look at that and say, "Oh, you know, I've got this great way of communicating. And if you talk about like ChatGPT or similar tools, all of these are things that process large language models, it's important to think that just that conversational aspect is not the only thing that these technologies bring to the table.

It's a lot more. It's any kind of medium that can be generated, whether it's text, whether it's pictures or so on. That's using the input medium and then presenting the output in a similar way. That's what the generative AI brings to the table. And I think the potential here is absolutely phenomenal.

What is SAP doing in the AI space? 

AK: Yeah, it's a very exciting time. Thank you for that really detailed overview of AI.

I mean, I think everyone can attest to that it's such a complicated subject. Before we get onto us as a business, SAP Concur, can you give an overview of what SAP is doing in the AI space?

RV: Yeah, so within SAP, AI is one of the biggest areas of investment because SAP firmly believes that by having an opportunity with AI, we can have the most desired outcomes for our clients.

Now, all of this is based on three core principles within SAP: relevance, reliable, and responsible.

As far as relevancy goes, we want to make sure that all of the capability is embedded within our technology solutions so that when we talk about future proofing for our clients, it's all there.

When we talk about reliability, your AI is only as good as the data that you have. And we have huge amounts of data within SAP. We work with data all the time. Just combining that data with business processes, that's what's going to make this reliable for our clients.

And then finally, responsibility. Now, you see this that a lot of people talk about AI all the time, but we do it bearing in mind security, privacy, compliance, ethics, and accuracy. And all of that really is what makes it responsible for us.

And you know, we want to make sure that we are at the forefront of this. So, whether you're talking about SAP Concur as a solution with travelling expenses or contingent workforce or procurement, AI's going to be embedded in everything that SAP does.

How do SAP Concur solutions leverage AI? 

AK: I think that's a really important point that people don't really think about is the responsibility side and the ethics side of things cause obviously, we don't want to go down the "Terminator" route with this whole thing.

But it all comes down to data as well because AI is only as good as the data that comes along with it, which actually leads on to my next point. I mean, I know you said you've been here 16 years, which beats my 10 years, but I remember launching ExpenseIt about six years ago and we were so excited about it and we're still using it today and it's one of our key product solutions.

But when we launched it, we were talking about automation and machine learning. Is there an AI element to that now?

RV: That's a great question and I think I'm going to answer to that automation and machine learning piece as well, but ExpenseIt, it was launched even earlier than that, or we've been working on that for a longer time to be honest. But it was a simple concept, right?

You take a receipt, you take the information from the receipt, create the expense for the user. It's great. I mean, I use it all the time. I hope you do too.

But for us, how we went about building that, we embedded what you said, machine learning into it. So, we'd create a model, think about it, just take a simple thing like an amount, right? What we did was create a model to say where would we find the amount on the receipt and then train and train and train that machine so it gets it right.

And you can get it wrong so many ways. I mean, think about it. It could be all handwritten, various kinds of receipts, could have different currencies and so on. Just trying to put all of that together. That's where we brought in that concept of machine learning and used it within ExpenseIt.

Think about hotel itemisations. I mean, you've been to the US. How hard it is and everyone knows just putting in, keying in all of those things and ExpenseIt does a fantastic job of that.

But should we stop at that? No, and to answer your question, we started with machine learning and now we're moving into the world of generative AI with that capability. And we looked at hotel itemisations and whilst we were doing a great job with the machine learning, gen AI actually takes it one step further. It's not there right now in ExpenseIt but it will be there soon.

So, we're not stopping and saying, "That's it, we're done." We are trying to see where gen AI is going to play a part for us, and ExpenseIt is going to be one of those.

AK: I think that's the key thing to actually continuously evolve the solution with AI as well. Because yeah, you mentioned the hotel invoice, which is the bug bear of everyone doing expenses, but what it should be is that it's automatic. The user, like when I'm doing an expense report, I shouldn't need to worry about it, and it should just happen in the background.

I think that's where AI is going to. Is that fair to say?

RV: Absolutely, and one of the other things I want to say is ExpenseIt, today we've got that on mobile, but lots of people don't know. We have also got it on emails. We send your receipts to concursolutions.com, and that's it. It's all ExpenseIt powered, and soon, it's going to be on the web. You drag and drop your receipt on the expense page, ExpenseIt is going to play a part, and all of that is down to AI.

AK: So basically, the expense report will actually end up writing itself, which I think is really cool.

RV: Absolutely. That's where we want to go to.

AK: Brilliant. So aside from ExpenseIt, done there, covered that, what other solutions do we have that are AI-powered or will be coming out?

RV: Yeah, and there's a lot of advancements in that stage, but I'll talk about a couple. And those are not the only ones that we have within SAP Concur. But one which is primarily AI-powered is Verify, and this is a tool that automates the audit solution.

Now, we've got two flavors of audit within SAP Concur. I'll keep it simple.

One is the one like Intelligent Audit where SAP Concur is the ones that get outsourced the audit process. So, we have our auditors running and doing the audit for the clients. And the second is Verify in which the client actually performs the audit. But before that, everything goes through SAP Concur's machine learning process or SAP Concur's AI process to do that kind of audit.

Now, think about it this way, if SAP Concur auditors are the ones performing the audit, what happens is a lot of clients turn around and say, "Let me do some sampling. I only want to do 10% of my reports, 30% of my reports." But with Verify, 100% of all receipts get scanned, and we go through and make sure that they're all accurate, they're all compliant.

I've actually got a stat from one of our clients who automated their audit process using Verify, and they said that they saved 8,000 hours by doing that. 8,000.

AK: Wow.

RV: That's a huge number. It's almost like the number of hours in a year, I think, or close to it. So, think about the value that it brings. It's absolutely amazing. In the past 18 months, I think we've gone through 60 million expense reports or 60 million expense transactions that have gone through Verify, and only 3% went up to the client auditors.

That's huge value. That means that we're giving them time to do something that's more important, pertinent to their business than look at just expenses.

AK: Well, yeah, like what we were talking about earlier, that it just happens in the background.

RV: Yeah.

AK: And that's the really important part.

RV: Yeah. The other thing along with the Verify is Invoice Capture. And again, think of that as ExpenseIt for invoice.

You know, we're basically looking at invoices and saying we can capture the information and create the invoice for you. So it works on similar principles, though slightly different technology at the back end. And that's yet another product that you can get from SAP Concur, which is AI-powered.

So, I just wanted to talk about these two in particular.

AK: Well, exciting times.

RV: Absolutely.

What's next?

AK: Well, we're nearly running out of time, I'm afraid. I'm sorry, guys. So, one last question. Get your crystal ball out. What does the future hold? What is coming next?

I know you spoke about Verify and the audit tools that are kind of live now or coming soon, but what's the next big thing?

RV: I was almost waiting for this question. There's so much buzz and excitement in all of SAP Concur to do with artificial intelligence, especially with the gen AI.

You know, we get asked all the time, "What are you working on?" And think about this as what is our goal? Our goal at SAP Concur is that travel and expenses practically manage themselves, and that we minimally interact with the user. You know, it's all instinctive.

If you think about the words that I used, you know, why do we talk about practically managing and minimally interacting with the user? And the reality is we've been in this business for a long time. We know through our experience that ideally, we would want to automate everything but sometimes, there is a reason that you want a little bit of touch with the user.

You want that additional information from the user. And that's where that whole minimal interaction comes in. Our job is to reduce it as much as possible. Nil ideally, but then not nil. Just get the bare minimum information that you need so that your expense report is completed.

Now, think of a trip, you know? I remember, one of my bosses used to say to me, "You're travelling to Paris. How much is it going to cost me?" It's a hard one, you know? How do I figure out how much is it going to cost me?

But now if you think about that kind of trip where you talk about the concept of planning, booking, expense reporting, if you were to start this concept to say just from a planning perspective, can I actually get those costs easy enough?

And one of the things that we are looking at Concur Request is the cost estimation for various components. And you using generative AI, we can go in and say for my air, car, hotel, build out the estimate. And whilst it's doing that, it can also look at my travel policy and say, "You know what? Your travel policy allows you for this."

AK: And also, it's got your preferences as well.

RV: Absolutely.

AK: So you can actually just ask the question and it does a lot of the legwork in the background.

RV: Yeah, the asking the question may not be here now, but yeah, but with Request, what it's basically saying is, "Hey, this is what you can have from your travel perspective and I'm going to build this thing out. I'm going to create that request for you. I'm going to create the expenses

for you, expected expenses." It can go out to your manager who then goes ahead and approves that.

Now think about it, that's half the job done from an expense work perspective. If it's tied to travel, what I can then do is say, "Okay, now take me through, get that booking done." And then once you're doing your booking, with our evolution of travel, can actually be smarter and say, "I want a hotel that's got a nice coffee shop in there in the lobby and it's close to a metro station." It'll filter all of that.

AK: Yeah.

RV: And then sometimes, you know, I know this, people come in and say, "How do I know what my policy is?" It's a very common thing. You go to a new place, you just don't know. But if you have an in-product assistant that can be of conversational nature and you can actually say, "Well, how much is it that I can spend on my meals?" It'll say, "It's this much." It's great. Or do I need a receipt? Yes. If you can get that kind of answers, having that conversational capability powered by gen AI, I mean, that's where we are trying to get to.

And then when your trip is near done, you've checked out from your hotel, now having your hotel itemisations done by gen AI, I mean that's just a part that I would love to tread on all the way through.

AK: It's like next level stuff.

RV: Yeah. And if you've got all of that, your expense report gets done, I mean, what else would you want? Ideally, that's my trip.

AK: Well, it's been so great chatting to you. And I think I've learned so much and I think we'll have to do another session in about six months' time because it changes so much. But thank you so much for your time, Rahul.

RV: Yeah. Thank you, Alistair.