Whirlpool Corporation (WHR) Earnings Call Transcript & Summary

Hello and welcome to SSON's Invoice and Payment Automation Digital Summit. It's great to be here with you. Naomi Secor, Global Managing Director of the Shared Services & Outsourcing Network. And I'm absolutely so excited for this next session, which is a panel on how AI is revolutionizing touchless invoice processing, and this is sponsored by our good friends from Kofax. So just before we get started, a couple of housekeeping items. The console before you is fully customizable. I have placed the speaker bios and their contact details right before you. So I do encourage you to connect with them. There is a Q&A tab at the bottom of the console. This is a panel discussion. We want it to be engaging. So please do put your questions in the Q&A tab, and they're going to address them throughout the discussion. We have put a number of wonderful digital assets into our gift bag for you. Kofax has provided a number of great content resources. So do make sure that you pick those up along the way. One is a great white paper on the benefits of accounts payable. These are facts and best practices from leading analysts. So I want to get right in because we've got a whole lot of goodness to get through. So I'm really pleased to introduce our moderator for this panel and my good friend, Paul Bartley, who is currently the Chief Operating Officer of Odfjell Drilling Philippines, which offers GBS to all business units of the Norway-based Odfjell Drilling group. And so in his role, he oversees all staff and operations. He's focused on increasing the number of services offered. He was formerly Director of Global Shared Services for Becton Dickinson, which is a Fortune 500 medical company. And prior to that, where I really got to know Paul was when he led the Program Support Center, the PSC, a $700 million operation with 3,000 staff that provides more than 60 shared services to agencies across the U.S. federal government. But Paul is -- with this extensive background has also been an incredible contributor to Shared Services & Outsourcing Network, speaking in our events, advising as such. So Paul, really, really delighted to have you here, to all of our panelists, to our sponsor, Kofax, and I'm going to turn it over to you to get it started.

Great. Thank you so much, Naomi. It's really an honor to be a member of this panel and moderating it. We've got some fantastic speakers for you today. Let me get right into the introductions. First, we have Hyusein Arabadzhiev. Hyusein is currently a project management consultant with a special focus on SSC and GBS transformations. He supports the entire journey from the initial strategy through process automation, which is obviously today's topic. He has experience in transforming many functions, including order to cash, procure to pay, record to report, VAT tax, HR. And he's also supported SAP implementations and set up service management frameworks. Next, we have Carlo Peretti. Carlo is VP of Finance and Global Director for the business shared services program at LafargeHolcim. Carlo has 30-plus years of professional experience in multiple executive assignments. He's worked in Europe, the U.S. and Asia. He's worked in automotive and construction materials, manufacturing and has extensive experience in setting up multidisciplinary GBS centers in Eastern Europe and India. And he's dialing in from Zurich today. By the way, Hyusein is calling in from Poland, so we've got quite a global reach today. And finally, from Arlington, Virginia, we have David Sentongo, and he's representing the vendor community. In his role as the product strategist at Kofax, David spends his time focused on what is happening within the accounts payable automation market. So he should have a great overview of what many clients are doing. He keeps an eye on what his customers are trying to accomplish and how Kofax prioritizes their investments and innovation to evolve their AP automation product to meet customer needs.

So let's jump right into the content. I know everybody is waiting for that. Let's start out with the basics. Advancements in technology have created a lot of opportunities for streamlining invoice processing. So let's look at what are the expected outcomes, what are people looking for. And Hyusein, you've had experience in a number of different settings. Please tell us about the outcomes that you've sought in the past and what you'd like to see in the future.

Sure. Thank you, Paul. Yes. So in terms of the outcomes, right, that I've sought to achieve from automation perspective in P2P, say, that they were predominantly related to reducing the time for invoice indexing, right, automatically posting invoices with the help of various technologies. Also, I've spent a good amount of time on integrating the warehouse and finance processes, the -- in a completely, say, automatic fashion, also reducing the effort for payment verification, especially in a multi-ERP environment. The invoice adoption, I think that's also very important automation piece, which has to be considered always, implementing supplier portals to centralize the invoice reception, automating the invoice prioritization to avoid the backlog and optimize the discount collection to have any discount agreements. In terms of the future, right, because that was, I think the -- I believe, the second part of the question, right? What the future is holding? I think the smart or the intelligent OCR technologies are going to improve even further, right, comparing to today. Integration between various portals and the ease of the user experience is going to become a higher priority. But really, also, I think what is going to become a priority is the automatic GL account determination with artificial intelligence, the cost center identification with artificial intelligence. Also, I mean, regardless, always the goal will be to reduce, right, the processing times and the invoice processing and storage costs.

Great. Thanks. So what I heard is really a focus on reducing time, automating and also looking at the user experience, especially in the future. Great. Thank you. Let's turn to Carlo next. Carlo, what are the outcomes you're seeking in streamlining invoice processing?

Yes. I'm referring, of course, mainly to P2P. But in streamlining the processing, we aim at the business service center where we consolidated this process mainly to 3 main goals, which are performance, control and cost. In terms of performance, of course, if we consider the P2P process, we -- the accountability of the service center is more and more an end-to-end accountability, even if there's still some responsibility are left to the retained organization and the business unit. And in this end-to-end accountability, of course, the main KPI are usually the paid on time of the vendors, the DPO for our cash flow. So what we -- the technology that we would like to more and more having in place in our process should, let's say, give the streamline, driving this KPI into the world-class performance. In terms of controls, and now having more and more machine learning and cognitive tools in place who are, let's say, catching for potential to pick a payment or helping document recognition or helping vendor statement reconciliation. In this way, of course, the level of control and the compliance that we can insure is probably over 10 or 20x what we could have even only 3, 4 years ago. So this is -- of course, also the challenge on cloud, on cybersecurity is also bigger. So it's not that we are on our last mode, but I would say that these technologies can, for sure, support the accomplishment also of the control goal. And the third one, of course, is efficiencies cost because the streamline of the organization means to be more productive, to be more automated with RPA, with cognitive mode with, let's say, not spending time in manual document handling or paper management. So these are the 3 main typical -- classical driver of the streamline. So performance, control and cost, and I believe we have the technology framework to move forward in these 3 drivers pretty fast.

Very good. Thank you. I like your construct with performance, control and cost. David, you're next. What -- would we not heard -- or what do you want to add to what we've heard so far in terms of the outcomes that people are looking for?

No, thanks for the question, Paul. I mean I absolutely love the points on increasing performance, efficiency and accountability. We're certainly seeing those focus areas as we engage with our customers. But quite frankly, I would say it's just incredible to see the amount of attention just overall that the AP function is getting. So the fact that we've focused on these things is a great thing. And when I speak to customers more broadly, it's evident that like all of these things, just they boil up to a higher point on shifting from being a cost center to a profit-driving organization. So just relooking at the accounts payable function in that regard, which is the expected outcome, how can we drive more profitability versus just traditionally focus on cutting costs. And we see this being the case, right? 52% of organizations see the AP function as critical, superimportant to the business operations moving forward. A big thing that I would also add to the list is a focus -- especially in this broader era of digital transformation, there's a focus on harnessing more data to drive intelligence. Data is a new currency. 80% of executives are just saying that intelligence are -- and data are superimportant elements of the AP function. And about 50% -- sorry, 56% believe that having access to more analytics, having access to more insightful information, all these are very critical to the AP function operating moving forward. And I think that's where we begin to transcend beyond being looked at primarily as a transaction processing function to more of a data and intelligence provider. And some of the capabilities we're going to discuss today around automation, around more intelligent OCR, some of the other things that Hyusein and Carlo [indiscernible] earlier are all critical pieces and levers to getting to the state of being more of a data provider and being less as a transaction processor.

No, it's really interesting. I know that's what most GBS organizations aspire to is to be that value-added partner. And it's interesting that even within the P2P process, you can wear that hat. You can be that answer to your business. So David, let me come back to you again with the next question. We've talked a lot about the outcomes, what people are looking for in terms of tools and process improvements. Let's talk about how to get there. I know you've probably been involved with many implementations. What does it take -- talk to us about how do you reach these outcomes.

Yes. Absolutely. A few ideas, and I'll just talk to some of them today. One is just establishing a clear vision and goals upfront. And I know this is an obvious point, but as I have more conversations across the customers that we interact with here at Kofax, and even in the past, I've spent some time with Deloitte on the service delivery side, some of these basic questions upfront around like what are the goals of the program in terms of long-term and also quick wins, defining those really does set an appropriate frame for how you interact with a vendor, for how you just begin to establish your phase and show you the scale and also engaging others in the organization, which leads me to my second point. I think building strong relationships internally in the organization is just supercritical to success. And my experience, I would say, in a lot of cases, the procurement function and, for example, an AP are many times not as connected, right, as we would like them to be, and something has to be done about that because whatever is done within accounts payable is ultimately impacted by procurement. So our rationalization and efficiency, improvement initiatives that we do at that procurement level certainly do transcend into benefits at the AP level. So we can't look at these things in silos. Looking at the problem from both sides and both organizations extending a hand across each other, I think, really is one of those overlooked steps that takes -- is critical to taking an initiative to the next level. I think the human cost here is also very important. So beginning to sit down with those folks within the AP team and understand their challenges and problems is also an important factor here and mitigate some of the issues around burnout, around loss of interest that can overall affect performance. And to that -- along the lines of better engaging and getting folks within the AP team, excited about what's going on, I think changing that identity also is an important factor to realize its success. As if the people within AP view themselves as transaction processors, and they see their value in terms of the work they're doing around processing the invoices, then any sort of digital transformation initiative is threatening because it's the value that they provide today. But if we begin to help them view themselves as something beyond that and something more strategic than that, then that aligns the automation initiative with helping them get to where they should be. So changing that identity at the AP function is supercritical. Then you ultimately get to choosing the right software partner. And I know we'll begin to talk about what some of those capabilities look like in a vendor, but really encourage you, the delegates here, to press their vendors hard on their ability to deliver to the critical capabilities along this AP process. ERP integration, as Hyusein positioned earlier in some of the intelligent OCR capabilities and workflow. Don't be shy about holding the software partner accountable to its ability to deliver along those lines. The last thing I would position is just setting realistic expectations. So it's not just about bringing technology in, but it's about evolving processes. It's a change management piece here and also level setting of leadership on a scorecard such that leadership is evaluating the program in a fair manner in terms of the quick wins versus items that will take a longer period of time to achieve. So hopefully, these are some points to just begin to spur some thought and help those who are listening to navigate and avoid some of the challenges and lessons learned that many can experience.

No, I really appreciate it. Thank you, David. I like how you said about changing the identity of the people. It goes back to what we were talking about earlier in terms of this function being a value add as opposed to just processing. And it flows right into what you said in terms of the threat of automation. They shouldn't see themselves as that. They should see themselves as a value add and more of an end-to-end look than just the one process that they're taking care of.

Absolutely.

So Hyusein, let's turn to you as a practitioner. What kind of the solutions have you implemented? And how did they go? And what have you learned that you can share with us?

And thanks, Paul. So first, I wanted to say that also completely agree with this, what David has mentioned that indeed, before really looking in the solutions, the important questions to ask -- in terms of the, say, invoice indexing, for example, so let's say, we decided to implement intelligent OCR technology. So the questions to ask here really are like what -- how do I receive my invoices today? For example. That's a fundamental question. Do I receive the invoices on paper? Do I receive them on a PDF? Or do I receive them via vendor portal or any other channel, right? That's maybe the first question to ask. Then in terms of -- and actually, this is going to be the first example, which I will articulate with in terms of the previous experiences. So I can say that this can categorize as the low-hanging fruit, right, based on my previous experiences that was the implementation of the OCR technologies to accelerate the capturing of the invoices and the indexing of them. And here, a key tip, which I can give to the audiences. So to plan in advance, right, for a mechanism to track and measure the volume of the manual paper invoices received from suppliers and to be able to measure by channel, right, how do I receive my invoices today, and to have a strategy to move as much as possible, of course, to receiving the data in a structured way. And here, the aim is, of course, to reduce, of course, as much as possible, if we don't have any legal restrictions, the paper invoices and to switch to PDF or preferably to invoicing, if that's possible. And here, I think the other questions to ask for the OCR or the intelligent OCR, what are my invoice volumes? How do I, let's say, in terms of -- which are my strategic suppliers, both from a business perspective and from a transactional volume perspective, right? And what are the tailored strategies we will follow for them? So this is something, I think, would also will be worth looking at. In terms of the challenges, so I would like to highlight that we have -- because I believe this was the second part of the question, I would like to highlight the challenges related to increasing the automatic posting rate of the invoices, especially due to the very dynamic tax and the VAT regulations in Europe, in particular, right, because in -- whenever there is, let's say, a VAT regulation change, right, or the, say, capturing rules have to change, the posting rules have to change and, say, that it requires some time to adjust the environment for that. So in brief, I think the low-hanging fruit will be the OCR space, the intelligent OCR space and the, say, the portals or the [ AIB ] solutions that require more time to realize the value. Again, we'll refer back to this for David. What David has mentioned that initially, we need to spend time, let's say, on standardizing the processes, aligning the organization, bringing the vision, articulating with the future state in order to be prepared for that technology and to realize the value from it.

Great. Thanks, Hyusein. I'm glad you're touching on the AI aspects since that is our -- the title of our session today, talking about intelligent OCR and the portals. Let's turn to Carlo. Carlo, can you give us some examples of AI and what you've done or what you've seen?

Yes. It's -- in my group, we deliberately decided to not to have an integrated platform for invoice to pay, but rather than go, let's say, piece by piece of the different component to address them with advanced technology. So this, on one side, of course, make a little bit more difficult to our approach, more, let's say, articulated. But on the other side, we can leverage on high technology in a specific process segment. Let me just give you a few example. Now, of course, the OCR and the document recognition is one of the main area of utilization of machine learning and artificial intelligence because we move from the traditional all the OCR, optical character recognition, to the new machine learning system, integrating this technology with our existing workflow. So -- because one thing is the technology that has to be available and effective in the process. Another thing is the change required to implement this technology in our workflow because, of course, having an integrated workflow that is interacting with all the user, you need to make sure that it's working properly. So that is the invoice data capturing. I would say this is the initial artificial intelligence application, which is giving an amazing improvement in our recognition rate. Another area where we combine artificial intelligence and machine learning with our RPA deployment as being in the auto matching and auto posting capability of our process. So to recognize the right GR or the proper PO in order to increase the standard auto posting capability we have from our ERP like SAP. Then moving forward in, let's say, the invoice to pay process, we can identify other areas that are really the ideal field for artificial intelligence deployment. One area is, let's say, to make -- we try to concentrate our service center in a few low-cost country location like India, like Colombia. So we need to become more and more -- to utilize more this center to become more and more language neutral. So we need to neutralize the language to make sure that we can -- we are not language dependent and we are not forced to create centers in some -- in each country or in some specific countries. So to do that, the NLP and the artificial intelligence technology can help a lot, especially in the service desk area, so where we received the query from our vendors to, let's say, to clear this query in a much faster and efficient manner than the old, let's say, help desk approach with the operator who is picking up the phone call or the e-mail, parking it and try to find the right guy who is able to respond, to fix on the queries. So that chatbot application with the usage of machine learning technology and is now -- we are able to combine more than 50% of our query through chatbot with -- until the, let's say, the end of the cycle, so until the resolution. So not only to screen them, which is now 90-plus percent, but also to fix them with a direct connection -- connector to our ERP data, to our flow -- workflow data so we can -- the chatbot can learn from the data and reply to the query about the status of an invoice, about why the payment was not executed and so forth. And then the last, let's say, interesting application I would mention in this list for artificial intelligence is, for sure, the integration with the -- of these tools like the chatbot or the document recognition with the issue ticketing or the service performance. So if we have a query handled by robots, query handled by machine learning chatbot, then this query can be -- when the query cannot be resolved immediately but meets some, let's say, work to be completed, we can link this technology with the service performance, ticketing tools that are in the market. And so I would say that the point is that we have multiple form and fields where we can have a very effective artificial intelligence application like the one that I mentioned. And then the key is to integrate these tools with the overall, let's say, ticketing system, with the overall workflow tool, with the underlying ERP. So these are the one that I would mention. That, of course -- then, of course, there are the others, let's say, not AI-driven, but more traditional like portals, RPA and so forth. But I would say that the injection of the cognitive element of the machine learning element has been essential to the last topic in the capability of these tools.

Very good. Thanks a lot for those great examples. I'm going to turn to some of the audience questions. They're starting to roll in, and I really appreciate everyone's engagement. The first question is about exception processing. And the question is, "How is automation specifically helped with exception processing? Is it more about preventing exceptions or freeing up more time for knowledge workers to handle those exceptions?" And David, I'm going to turn to you first. But if anybody else wants to jump in, we'll take the rest of the panel afterward.

Yes. Appreciate the question. That really resonates. I mean we were doing around a research, I believe it was with The Hackett Group, some months back. And we found that, yes, almost 25% of AP [ stock's ] time is spent on managing exceptions. So we see the pain with our customers. We feel it as well, too. I would say it's on both sides. So there certainly are very robust capabilities that automation can bring in to prevent exceptions in the first place, but the fact of the matter is that they never 100% go away. So that's where some of the workflow capabilities with the solution, that's where they could become -- they begin to come into focus to help manage that exception handling process, getting exceptions to the right people at the right time. If somebody is not in the office, rerouting them to keep the flow of things moving forward. And that's where the benefits of a robust workflow can come into play to overall reduce the amount of time that's spent on processing invoices. We're seeing that -- and many will say it's taking sometimes 20-plus minutes per invoice or almost even days sometimes. Our best-in-class performance gets you down to 2 minutes on average, and the only way to do that is by managing the exceptions more effectively. In terms of preventing exceptions, I think this plays into some of the AI capabilities that we were just talking about earlier in the national language processing. At Kofax, we have invested substantially along these lines, and we get away from relying -- and a lot of the AI we've invested in is around how we understand and process data within an invoice, no matter it's format layout language. I love the language mutual statement earlier. That's really where we're trying to get our customers to be especially given that they're global organizations, they're doing business with suppliers around the world. And so we've built a database of millions of different invoices. And that knowledge base positions our customers to be able to immediately begin the process a wide array of invoices from a lot of different suppliers no matter what the layout is. So there's that immediate value that they can get out of the box. But even then when there is an exception and something is not quite understood, there are mechanisms within the software to learn directly from operator input in a visual fashion. And that's where you get to the higher levels of straight-through or touchless processing over time.

Great. Anybody else want to jump in on that question?

Yes. So Paul, so in terms of exceptions, I think definitely here also, the starting point is to look at the process first on the -- and more specifically, the upstream process or all the upstream processes that generate those exceptions, downstream and accounts payable and to prioritize those and to address them first. But also, let's say, what I think could help in parallel to that or on top of the upstream analysis, and I think this will be related also to this what David mentioned, is to consider an automatic escalation process, that we have an automatic escalation process because if the exceptions also don't get resolved for a very long time, and they are stuck in people's workflows and fail to receive responses and by leveraging either RPA or any other workflow solution, we can automatically, let's say, escalate the invoices up the ladder. And this, I think, also created a culture in the organization to kind of the upstream function, let's say, will become more accountable for resolving those exceptions, right, whenever they get escalated and when they don't respond. So I think this also is another alternative approach to be considered prior to putting any, let's say, more sophisticated technology around that.

Great. Thanks. We have...

In addition to what Hyusein was saying, I would put all this question about how, let's say, technology and artificial intelligence can help in driving exceptions that is, of course, one of the main process challenge that we have. I would put that in the context in my, let's say, repetition of the streamline drivers into the performance context. And we clearly have a KPI in this performance context that is the first pass yield, first pass touch that, of course, we would like to have as higher as possible. But of course, we have to deal with this exception that generally a workflow tool is handling, dispatching to the different, let's say, contact that can drive and resolve this exception. Now as Hyusein said, the automatic escalation and this tool can, let's say, make it more efficient. And on the root cost base to prevent these exceptions, of course, the integration of the artificial intelligence with the existing workflow coding could be really the key enabler. I mean the workflow coding is generally classifying this exception by nature. And of course, for each of them, we find a solution that is given generally by the contact. If we look at the data, the analytics of the history of this query and how they can be -- they were resolved in the past, the machine can learn or predict in a sort of predicting analytics how to resolve the next exception without sending it out to the point of contact. So this is the form of integration between artificial intelligence and workflow or existing, let's say, system that I stressed also before in the -- being the key element for a successful implementation of artificial intelligence, too, in account payable process.

Great. Carlo, we've got a follow-up question from your earlier comments. The question is, "What is the level of automation at your company with intelligent OCR solutions? What percent do you think you're processing with those solutions?"

Now let me give a structured answer on that because there are a couple of metrics to measure that. When -- especially when you compare just high level the different tools, you say 99%, 99.5% or whatever. So first of all, you need to understand how many fields in your invoice the tool is addressing. So if it is 5, it is 8, if it is at line level or if it is just at title or amount level. So this is the first important thing. The second one is, of course, is charactered by -- field by field, which is the recognition rate, which should be 95 plus. And then you need to see on the -- if you have -- of course, more fields you have, less is the likelihood that you have not -- even one miss in your chain to see which is your pass-through. So in that case, you need to do a combination because if you have only one character, one field, just [indiscernible], of course, you need -- you should have more than 95%. But if you have 30 fields, the likelihood that in one of them you miss is possible. So even if at the end, you have only a 30% or 40% pass without touching the invoice from your operator could be also okay. So it's a combination between the number of fields and the recognition rate you have for each fields. So you need to do a simple math. If you want a 98% and you have 10 fields, you have to do, let's say, 0.98 elevated to 10, and you will get your rate of success pass-through that you have. Not sure if I've answered to the question.

No. I think it's not a one word answer, that's for sure. There's a process and a lot of steps here. Let's move on to some more questions. I want to get to as many as I can. The next question has to do with the marketing around these solutions. The question is, "Why do people associate RPA with AI since RPA is programmable macros?" Maybe we're giving the other solutions to too much credit or maybe not enough to the RPA. What do you guys think? David, I don't know if you want to take that as the vendor here, but we can take...

Absolutely. I mean I would attribute that to just broad misinformation that's in the market, which then translates into challenges when organizations are beginning to roll out these things. RPA, I think, is a very robust -- it provides a very robust set of capabilities around automating manual repetitive tasks that are very rules-based, and rules-based is a very important quality of those processes. When we start getting into situations where there is more fuzzy logic or you cannot do the if-then-else sort of statements, I think those RPA processes begin to struggle to scale. And so that's where you get into the delineation between RPA and artificial intelligence. But even artificial intelligence itself is not a black box. It's meant to be complemented and to work alongside people. You still have your subject matter experts in your SMEs who should be working in tandem along with these RPA and AI capabilities to get to that more productive and state. So -- and that ties into the change management, helping people not to feel as intimidated. Artificial intelligence not [indiscernible], and that will completely replace people by any means. I think it does help to alleviate some of the situations where a human being is having to check in every single time. In the future state, we shouldn't be doing those things, and more and more of it is automated. Yes, if there's a simple line in the sand, I think if there's any element of the process that is rules-based, that is programmatic in nature, you can follow specific steps. RPA can come in to do those things on a repetitive basis. And the more transactions you have, the more valuable it becomes. Whereas artificial intelligence will begin to tackle a percentage of those items in a not so rules-based. I'm tying that back to the invoice process or the AP conversation that we're having. A lot of that AI is about like how do I -- if I have this invoice that has certain information in the header and the footer and in the line item details, and I need to do some levels of auto matching and so forth, and every invoice is not quite the same-looking. We see invoices that are slightly different in their layouts. How do I bring in a sort of capabilities that bring some consistency to my ability to pull out the information I need and reduces the amount of time that my staff are spending? How many [indiscernible] look at it? And that's where artificial intelligence can begin to alleviate that burden.

Thanks. Let me take a tangent off of this. You've mentioned several times on the people side, and I think we've talked a lot about the technology in this last 40 minutes. Let's talk about the people aspect. How is it going in terms of convincing your staff that this is a good thing and that it's not a threat to them? I'm going to turn to the practitioners on this one, Carlo and Hyusein. Has it been difficult? Are people ready to accept these solutions? Are they happy? Or has it been a struggle?

If I can start on my side. It's not been a big challenge, this one, because, of course, the -- now the global business service platform is growing quite significantly because -- especially with COVID companies have discovered that this could be an easy way to get value for the company. So we are in a growing path. So being a growing path, the fact that we are, let's say, removing a lot of transactional work, so even some cognitive work, but in sort of -- in basics kind of things is an opportunity for some people to -- they'll get to promote it to more, let's say, interesting kind of jobs. Also, this way is for us a way to replace attrition with experienced people because if you can, let's say, drag out from your current processes some people because they are replaced by AI or by an RPA solution, it's much better to use these people who are already familiar with the organization, with your process for replacing some other people in different job rather than to hire and retrain people from outside. So I believe that automation and artificial intelligence is really fitting well with global business services environment, especially in a period where the environment is significantly growing.

Yes. That was my experience, too, at Becton Dickinson. When we implemented the RPA solutions, I remember this one case. We went from 10 people doing the process down to 2.5. And we just redeployed those people because they were needed. They -- we didn't -- we never let anybody go because of these kind of solutions. So being in a growth environment really helps. I think we've got about a minute left, Hyusein. So I'm going to turn it over to you and get your comments on those, too.

Quickly then because it's a very interesting and a very broad topic, that one for the people. But I think the first reactions, right, of the people are that they're very abstract, right, around -- it's very abstract, right? So people can't imagine, okay, how -- now the machine will come. Do this, what I'm doing, even do better. That's, let's say, the first reactions. But really, the most important message here is to say, okay, today, let's say, the work is executed only by humans. Tomorrow will be machines plus humans. So we need to find a way to partner and kind of to work together those new machines. And then I think people kind of -- because mainly the environment in the shared service and GBS are kind of millennials, so they like technologies. They like gadgets. They like using, let's say, different tools. So then they are more open, let's say, to those type of technologies after this message is conveyed. [indiscernible]

You're right. It's a big topic. Naomi, I'm going to turn it back over to you. It's been a pleasure. Thank you.

Well, thank you, guys. Thank you for the terrific job moderating, Paul, and to all the panelists. That was really, really a great discussion. And to the audience, the questions are coming in fast. We feel really sorry, we couldn't get to all of them, but we will follow up with anyone's question that we could not answer. So -- and again, thank you very much to Kofax for sponsoring this wonderful discussion. So next up, we have a case study from Cargill on how to leverage hyper-automation capabilities to digitize AP transaction processing. This begins in just a few minutes at 10:40 a.m. Eastern Time. So to get there, simply advance to the green button at the bottom of the screen. That will take you right into the next session, and we will see you soon. Thanks, everybody, and have a wonderful rest of the day.