>> Announcer: Live from Las Vegas, it's theCUBE, covering UiPath Forward Americas 2019. Brought to you by UiPath. >> Welcome back, everyone, to theCUBE's live coverage of UiPath Forward here at the Bellagio. I'm your host, Rebecca Knight, co-hosting alongside of Dave Vellante. We have two guests for this segment. We have Sanjay Sadasivan, he is attended automation process lead at EY. Thank you so much for coming on the show. >> Sure, thank you for having us. >> Rebecca: And we have Goutham Goudgere, he is the attended automation lead at EY. >> Great to be here. >> Thank you so much for coming on the show. >> Thank you. >> So, about a year ago, you embarked on an attended automation project within EY. EY, of course, is a company that helps other companies with their RPA transformations, but this is one you did on your own. I want to hear about the impetus for this project. Why did you start it? What was going on? >> Sure. >> Yeah, so I can take that one. So we started this project, like you mentioned, about a year ago. We both are from EY's SAP practice, and EY has been undergoing SAP transformation for the past few years, and so, we're kind of replacing a whole bunch of, almost 1,400 systems and moving to a single-instance SAP project, which covers everything that our client servers do in the market, entering client information, all the things that CRM does, project management, engagement economics, as well as the whole finance and procurement work. So I think we've got, right now, 100,000 users on the SAP single instance. And SAP is great at what it does, which is essentially entering transactions into the system efficiently at scale, but the feedback that we were getting from our end users, especially those end users that end up using the system maybe once a week to maybe once a quarter, was that it was sometimes too difficult for them to navigate into the system, try to remember all the things that they had to do in the SAP system. So we were at the Miami event last year, and so, we heard about attended automation from UiPath, and we kind of went back and did a POC to see, can we use attended bots on top of SAP system and help the user go through some of those usability challenges? So we started last year, and we are currently live as a pilot. EY is 250,000 users, so our pilot is huge, so it's got about 20,000 users as of this week. >> Why attended bots? What are attended bots? What's the motivation for attended bots? >> So, in fact, when we started the process, we had two options, to go through kind of the traditional unattended bot, which was have the user enter data in an Excel sheet or some sort of screen capture, email that information, and have the bot then enter it. We initially started with that, but that ran into several problems, like the data was already too old by the time it got into the transaction system. Then we had to rebuild the full front end of SAP, which has taken years to build. So that's why we started using attended bots, which was, can we just put a bot on the user's machine, so that when they want to enter a transaction, they just call the bot, and the bot does all the hard work for them? So that's how we've been-- >> Your question as to why attended bots, there's a certain level of intelligence, actually, the user puts in when entering these transactions, so an attended bot actually just takes some information and then plugs it in. But that's not the way EY works. For example, if you're entering an opportunity, there's a lot of thought processes the clients are actually going through. "What's the pursuit going to be?" "Who are my likely sales leads?" "What's the percentage of what I'm "winning this transaction?" So we needed the user to actually enter that information in the system. But using our normal SAP system, especially when you're on a sales lead and you're meeting with a client, entering all that information was getting a bit cumbersome. The attended automation process actually cuts down those steps. For example, if SAP requires you to enter, say, 20 fields, this actually cuts it down to five fields or five screens, and they can enter that information to the attended bot, it guides them through each process, it's a streamlined process, and they can exit out of it, that's it, they're done with the system, focus on the lead, actually. >> So you came to the conference last year, had this eureka moment, "Hey, could we do this "and help our people who are suffocating "under these dreary, tedious tasks?" So, was it a hard sell? I mean, were they easily brought along, of, "Yeah, we want to try this"? Or was there any anxiety on their part, of, "Ooh,"-- >> Sure. >> "What are we doing here?" >> We had that. >> So I think we had a great sponsor within the firm, and we are trying this out for the first time. >> Sanjay: Yeah. >> In fact, from talking with UiPath, the experts here, not a lot of other companies have tried this. So we did go through a step-by-step approach to kind of de-risk as we went through this. We started with a small POC, learned from that. We then put those bots in front of real users, got feedback, kind of Agile approach and built it over time. >> Yeah, I think one of the key points was really doing a business "let's go" definition, so we went to our partner community and asked them, you know, "What are the most frequently used processes "that we should automate with an attended "automation process? "What are the pain points? "What are the current challenges? "We want you to alleviate us." So, basically, we actually used the feedback from the partnership to focus on those particular steps to automate. So, and then change management, obviously, you have to engage with the user community all the time to make sure that, you know, getting the right feedback, maybe adjust our process or how we're building the bots accordingly as you're going through the process, I think that is key as well. >> So, the user experience now, so walk us through what it's like now for the human worker who had these tedious tasks, and now, what's it like with this attended bot? >> So, the attended bot is an application which is on the end user's laptop. So as soon as they open the laptop, it's right there as an icon on their bottom right corner. So they go there, click on that, and it lists about 15 SAP processes that they can run. So they know what they want to do in the system, they want to create a new client, they want to create a new engagement, or a project, and so, they would go there, call that, you know, just click play, and essentially sit back and watch the bot then take over their screen, navigate to the right SAP transaction. And, you know, navigating to this transaction seems easy, but when you have 15 processes that 200,000 users need to know, and it's not straightforward sometimes, the bot does three, four clicks, before they know it, brings them to the right screen, and then it also adds a message on top of every screen that says, "This is what you are supposed "to be doing on this screen. "So, before creating a new client, "first search within our MBM system, "first search within this DNB system, "to make sure you're not creating a duplicate." So we've got help messages added on top of the screens as well, so it kind of takes you through the process. >> So I think one of the main points, yeah. Yeah, I think (clears throat) a normal SAP system in a particular screen, you could always go to a help, maybe a portal, and get some help. But with attended automation, it gives an opportunity for each of those screens as well to give specific help, contextual help. So basically, if you're on a particular screen and you're having an issue with this particular screen, you don't have to pick up the phone and call, maybe, a user desk, or close the screen and look through some manuals. Right there, through the attended automation process, we gave a link where they can get actual information on that particular screen, so they can finish that step without actually closing out of the process itself. So that is one of the big-- >> So the way I explained this to internal audiences is, we have built the bot to be our best-trained employee in SAP. So instead of them calling a human to go through the SAP transactions, the bot is right there, guiding. And the bot is watching what they're doing, so if it gets a SAP error, then it can suggest to them, "Here's two ways that you can get around "this particular error." So it's doing things like it's having your friend sit next to you and tell you how to go through the process. >> Your smartest friend. >> You guys are in the SAP, sorry, you guys are in the SAP practice. >> We are, yeah. >> That's right. >> And, so it's been quite a run, the last 10 years, for Bill McDermott, and I believe they acquired an RPA company, a small, little tuck-in. >> Goutham: That's right. >> But you guys chose UiPath. (laughs) You know, I don't know what that says. >> Goutham: Yeah. >> But what are your thoughts on that? I mean, in terms of, we've been asking practitioners, best of breed, or full suite? Obviously, you went with best of breed. >> That's right, actually, we spoke with SAP before we started our journey. We actually did a POC with one of the smaller firms, an RPA firm, and we spoke with SAP as to what their capabilities were. But just looking at what's out there, in terms of the product suite and how it fits our processes, we just felt like UiPath went that step further, and really met our needs in terms of attended automation. And then, obviously, we looked at the Gartner surveys it's on, and it was right up there in the right-hand quadrant, so we felt like UiPath was the right answer for us. >> Rebecca: Great, well-- >> But we are working with SAP to actually help through this process. So, the bot has to watch the screen, like I mentioned, and kind of understand the screen layout, the screen fields, so, instead of that watching words and names on the field, it's actually using this thing called automation ID. So SAP is adding these IDs on every screen, which actually helps any RPA vendor sit on top of it. >> And that's one of the key learnings that we've been finding out, that SAP, as Goutham was mentioning, we have these automation IDs that the attended bot actually uses to transact with the system. And then we see that SAP CRM as this cloud release every quarter, and they basically push out some code to our systems, which is every quarter, and we see that more and more automation IDs are coming through the SAP systems as well. So we have our own challenges in terms of managing those quarterly releases and make sure that it doesn't break our attended bot. But UiPath so far has been good. >> Excellent. >> Yeah. >> Rebecca: Well, thank you both so much for coming on theCUBE. It was a great conversation. >> Great, thank you. >> Thank you for having us. >> I'm Rebecca Knight, for Dave Vellante. Stay tuned for more of theCUBE's live coverage of UiPath Forward. (upbeat music)

>> Announcer: Live from New York, it's theCUBE covering Big Data New York City 2016. Brought to you by headline sponsors Cisco, IBM, Nvidia, and our ecosystem sponsors. Now, here are your hosts, Dave Vellante and Peter Burris. >> We're back. Goutham Belliappa is here. He's with Capgemini. He's the Big Data Integration and Analytics Leader at Capgemini. Welcome to theCUBE. >> Thank you. Happy to be here with you. >> So a lot going on this week at Big Data. You guys have one of the top SI's consultants in the world. What are you seeing as far as the transformation of organizations to become data driven? What are some of the drivers that you're seeing out there? >> It's a good question. So a couple of years ago, we started on this journey with Cloudera about four years ago. When we started this journey on LinkedIn, you saw the poster that said, "Big Data is like teenage sex - everybody talks about it, nobody does it." Right? The reality shifted considerably. So while the technology's evolved considerably over the last four years, the most important thing is most of our clients are feeling pressure from the disruptors in Silicon Valley. You see the AirBnb's and the Amazon's and the Google apply pressure's on traditional industries that didn't exist before. For example, a lot of our auto clients don't believe auto clients are the biggest threat. They believe Apple, and Google, and Amazon are the biggest threat. Right? Because what our clients are afraid of, the incumbents, the traditional companies are afraid of, is they don't want to become a commodity manufacturer of components for a software company. They don't want, for example, GM manufacturing a part that Apple is putting the wrapper on, selling and making the margin on. So, more and more tech is driving the industry to where GE made the announcement they no longer want to be known as an engine manufacturer, they want to be an IT company. >> Peter: Or a financial services firm. >> Or a financial services firm. And you see the same thing in pharma as well. We see the pharma companies don't want to be known as manufacturers of med devices, they want to own the service industry. Move up the value chain and secure the revenue stream. So that's what's changing the industry as a whole and then Big Data Central to the strategy of data-enabled transformation. >> So it's like the death, what was the article we saw yesterday? Who wrote that? "The Death of Tech". It was Rob Thomas, right? The death of tech companies is now the rebirth of... all companies are tech companies. >> All companies are tech companies and that's the future of all companies: to be a tech company and move from selling commodities to selling services and having a vested interest in the outcome that the clients receive at the end of the day. >> Yeah, I once wrote a piece many year ago that suggested that we would see more non-tech companies generate SAS and Cloud applications than tech companies themselves. And while it's still hasn't come true there's evidence on the horizon that it very well likely will be a major feature of how companies engage their customers through their own version of SAS or deploying their own Clouds for their own ecosystem. And you can go back, thirty years, thirty-five years and look at MAP/TOP for example and the promise of what it meant to define and deploy standards that could integrate whole industries around data. Hasn't happened, but we can see it actually happening on the horizon. What industry? I mean, you're still looking at things through an industry lenses, right? Where do you see it happening before it's happening elsewhere? >> So, the first place it happens naturally is tech because they're closest to it, right? To give you the classic example, I can go anywhere and buy an Office license today. I have to subscribe to Office, right? So, what it's done to Microsoft, it's changed the fundamentals of the balance sheet from selling perpetual licenses, getting revenue once and then having the prospect of not having a customer later, to selling it over a sustained period of time. So moving from one-time revenue hits to perpetual revenue. So tech is where it's starting off. And even in tech, we're actually pushing the boundaries by working some of our providers like Cloudera and some of the other providers out there to move from a perpetual license model to as-a-service model. So what this enables people like us to do is to offer as-a-service to our customers because our customers need to offer as-a-service to their end users as well, right? I gave you the example of GE because it's public knowledge. They want to move up the spectrum of not selling an engine but leasing an engine to an airplane manufacturer and then owning the services revenue on it, right? So when Delta, let's say, that's leasing the engine is no longer owning a commodity, they're becoming asset light, right? The companies like GE and other companies when they become tech, they need to become asset light as well, which means not being burdened by land, labor, and capital but, as they get paid for outcome, they want to pay for outcome as well. >> Somebody's got to own the asset eventually. This is not a game of musical chairs where the asset-owning music keeps playing and then it stops and somebody's got all the assets. >> Ghoutham: Exactly. >> So how do you see... the global sense of how organization, how is this going to get institutionalized? Are we just going to have a few companies with enormous assets and everybody else running software? How do you think it's going to play out? >> Good question. So Jeff Bezos was at a manufacturing company outside of Arland recently and he pointed at and antique generator sitting next to the plane and said, 'Back in the day, everybody had 'a generator sitting next to the 'company producing electricity.' But today we have a big distribution plan and we get it off the grid, right? So to your point, yes, we see the scale and the price reduction coming from a few companies owning those pieces of assets. For example, it's almost impossible to compete with the Amazon's and Google's of the world today because at the scale that they receive. And the customers get the benefit of that. Similarly, you'll see the software, right? So software, you see the software companies owning the assets and title and leasing it back to the customer. So to your point, yes, we're moving to a model where it's more scalable and the price efficiencies of them, they're passed on to the end consumer. >> Peter: So historically, in a more asset-oriented company, historically, if you take a look, for example, at Porter. Porter's competitive strategy. So Porter would say, 'Pick your industry' where an industry is a way of categorizing companies with similarly procured and deployed assets. Automobile had a collection of assets and hotelery had a collection of assets. So pick your industry based on your knowledge and what kind of returns you're likely to get. Pick your position in that industry and then decide what games you're going to play using the five-factor analysis you did. But it was all tied back to assets. So if the world's getting less asset-oriented, hard asset-oriented >> Ghoutham: Hard assets >> What does that do to competitive strategy? >> Good point. So the hard assets are getting commoditized. The value comes in what you can build on top of the hard assets, which is your IP, right? So the soft assets of IP and software is where the value's going to be. So there's a lot of pressure on hard-asset companies. You see many companies getting at the server market because they can't compete with the Amazon's and the Google's. They can wide-label and manufacture all their stuff. The differentiation is going to come in the software. That's the reason companies like GE and the other pharma companies and automobile companies want to become tech companies, because that's where the margin is, that's where the differentiation is. It's no longer in the tangible, hard-assets but it's in what you can do with them. >> Dave: Well, and it says data's going to be one of those differentiators. >> Yeah, yeah. >> And a big asset so what... Everybody in theory has to become data-driven, maybe in fact has to be- >> Data is their asset, is their differentiator. >> You've pointed out many times all this digitization is data. >> Peter: Well, yeah. >> Digital equals data. >> So our basic proposition is that increasingly the whole notion of being a digital business is about how you differentially use data to create and sustain customers. So let me build on that for a second and say that there's this term in economics known as "asset specificity" which essentially is the degree to which an asset is applied to a single or limited numbers of uses. Programmability reduces asset specificity so if we go back to the airline engine example, GE added programmability to an airplane engine and was able to turn it into a service. Uber was able to add programmability to a bunch of consumer cars and was able to turn it into a ride sharing capability. What does that say about the future of an industry-oriented approach to conducting business if I am now able to reconfigure my asset base very quickly and the industry's based on how my assets are reconfigured. What does that say about the future of industry? >> Ghoutham: So, in my opinion, I don't think the future of industry is going to change because you still going to have a specialization based on the domain you're selling to and the expertise that you have. >> Peter: So it's customer-focused industry definitions not asset-based industry definition. >> Ghoutham: The hard assets or going to get commoditized and get moved out to a few specialty players. But the differentiation is going to be on how you serve the customers and the type of customer that you serve. >> Dave: So what are the head winds you're seeing in terms of customers getting to this data nirvana? What are the challenges that they're facing? >> So, Peter Drucker. There's an attribute of Peter Drucker, regardless of who said it, 'Culture eats strategy for breakfast.' We work with retailers all the time who understand that they face an existential threat from Amazon, however their culture prevents them from being like Amazon. It prevents them from experimenting. It prevents them from failing fast. It prevents them from acting together. For example, a lot of customers want to have an OmniChannel strategy. It's a seamless commerce strategy but then they have a silo for the stores they have a silo for the call centers, they have a silo for the web, but they don't act together. So culture is one of the biggest barriers we see in enabling that journey. Tech, we know that tech works. Two years ago we're doing technical POC's. Today, we're not anymore. We know that tech works, right? So get over it. So it's a culture and the attitude and the ability to change how you go to market that's to me the biggest challenge. >> Peter: But isn't there also finance? Because hard assets still are associated with a rate of amortization, depreciation, and utilization. There's expertise and what not built up around that, and this becomes especially critical when you start thinking about the impedance mismatch between agile development and budgeting, for example. So how do you anticipate that not only culture has to change, but also the way we think about finance? Or is financing disciplines end up being a part of the culture? >> Ghoutham: So you're absolutely right. So, financing discipline has to be part of the culture. To give you an abstract example, back in the day when we did a data warehouse or a data project, we'd do a huge, let's say for lack of an argument, 10 million dollar project. Today we're doing 40, 50, 50k, 100k projects. So Agile has gone from fixed scope where you laid out a two-year project with an end in mind and by the time you achieve that end the requirements have changed and the business has moved on, to achieving small objectives. So we're consuming it in chunks. You're going from fixed scope to fixed budget. So I've got a certain allocation that I need to use and I prioritize it on a regular basis on how I want to consume that basis that I have. >> So it's almost a subscription? Are you going in basically almost subscription-basis? Going to a customer and saying, here's the outcome. We will achieve that outcome over a period of time. You'll sign up to achieve that outcome over a 12-month period and will consume that budget in 12-month increments? >> First and second, in any given period, you can re-prioritize the outcome that you want to achieve. During the journey for 12 months, if you realize something new, you have the flexibility to change. Let me take out this chunk of work and do something else so I have the flexibility. >> Peter: So you can redefine the outcomes? >> Yes. >> It's almost like, I don't know if you'd call it this, I'd be interested to know what you guys call it, but it's almost like a subscription-to-outcome business model. >> Ghoutham: Exactly. >> Dave: Service is a service. >> Ghoutham: We call it sprint as a service. >> Service is a service. >> We call it sprint as a service is our defined model of how to go to market around that is we know two sprints ahead what we're going to deliver. Everything else is indicative, right? Because not everything we do has to succeed. That's a mindset change that our customers need to realize. We believe the biggest reason clients fail is because failure is not an option. They put so much behind it, when they fail, it's catastrophic. >> Peter: Because careers fail- >> Yes >> Peter: And not the project fails. >> Exactly. >> Dave: You're not saying "failure equals fire" mentality. If that's the culture, then people refuse to fail and they end up failing. >> Until it's catastrophic. >> (Dave laughing) >> So I was having a conversation last week at Oracle OpenWorld when theCUBE was here, great show, and had a really good conversation with a competitor of yours who talked about how they were going to use machine-learning in the contracting process by sweeping up all kinds of data and that would help them actually define the characteristics of what they were going to deliver. How much work was going to take, how much labor, what other resources? And they were able to get rid of the 500 thousand to five million dollar part of the assessment or the assessment part of a deal, drive it down to 50 thousand dollars or less and in the process come up with contracts who are much more customer-friendly. What other types of changes are happening in the services business as we do a better job of packaging intellectual property whether it's this "service as a service" or "service subscription" or whatever you mentioned or even thinking about machine learning being applied to the contracting process. >> Dave: "Sprint as a service" >> That's correct. Sorry. Thank you. >> You've asked a number of questions so first thing >> I did. >> Let me talk about machine learning and human task automation. So one of the biggest things we're doing today is learning to understand and automate human tasks. One of the biggest things we've seen, supply chain companies for example, is they don't have enough planners, right? So you hire a bunch of planners. You have different variations and skills. So we're taking the top 5% of planners, automating what everybody else does and letting them handle exceptions. And workforce automation, in many of those areas, we're beginning to automate human tasks and letting the human handle exceptions that a machine cannot handle. So machine learning has becoming fundamental in everything, and not just contract negotiation, but actually enabling companies to scale in areas where they could never scale because they never had enough people to do it. We're not just doing it externally to our clients. One of the things we're doing internally is we don't have an Big Data developers so we're beginning to use machine learning to automate a lot of tasks that developers will do. Industrialize a lot of it so we can scale in our delivery approach as well. >> Peter: Excellent. >> Come back to this event. You guys are here, you're on the floor. We've been talking all week about, you know, Hadoop is kind of yesterday's news. >> Ghoutham: Yes, yes. >> What are you guys seeing? You got a big chunk of customers that said alright, we're going to invest in Hadoop. We have the skill sets. And then a big chunk of... I'm not going there. And now they're sort of looking at new ways. Whether it's Cloud, whether it's Spark. >> Peter: And a big chunk of customers will say I do want to go there, but I'm having problems getting there. >> Yeah, right. And I got some serious challenges. So what are you seeing there, and how is CapGemini helping them? >> So we did an analysis with Forrester and one thing we'll say that 100% of our clients are going to Hadoop. It's not 95%. So everybody's going to Hadoop in one way, shape, or form. Whether you go with the traditional distribution, go with an Amazon as your whatever, everybody's going to Hadoop in some way, shape, or form. To address the reluctance, we spoke about the Uberization of the industry, which is you have a contract, which is an outcome-based contract. So we go to our clients who have fears about moving to Hadoop and say, 'We'll take the risk'. Let's write an outcome-based contract to move you guys into the noob because you know you need to go there. You're afraid to go there so we'll take the risk, we'll shift the risk over to us and we'll move you onto Hadoop. The last piece is industrialization. So back two years ago, we designed code for every little thing that we needed to do. Today, we've automated a lot of our code generation from existing systems, from knowledge we've gained, including machine learning to we're able to mechanize a lot of the code. Frankly, we did it because we had a developer shortage. So we started industrializing a lot of our IPN, our assets, and our learnings, but this is also helping our customers move on to the new world. It's improved the quality of a delivery. It's improved the velocity of a delivery. It's reduced the price where we're much more competitive. To give you an example in the BPO space back in the day we did labor arbitrage. But more and more, like with our clients who use manual auditing, we're using machine learning to automate a lot of that. And that more than pays for the cost of Hadoop. So to answer your specific question, gone are the days of 'Hey, I want to get into Hadoop.' The question is what business value can I achieve? How fast can I achieve it, and if you're afraid, can I take the risk for you? >> And that business value, historically, if I can use that term on such a nascent industry, Has been... the ROI's been a Reduction on Investment. >> Ghoutham: Correct. I'm going to lower the cost of my enterprise data warehouse. >> Ghoutham: That was two years ago. >> Okay so what is it today? >> Today, it is 'How can I reduce your marketing span? 'How can I optimize your marketing span? 'How can I improve the accuracy 'of your supply chain planning?' So it's more in terms of directly delivering business value versus the cost reduction. Many of our clients say the cost reduction is irrelevant. Frankly, because the business case is so huge. To give you an example of one of our supply chain clients, their fill-rate for orders is 60% which means they're a big manufacturer, they're only to fill 60% of the orders that come through. That's because they're not able to plan where to deploy product and so on and so forth. So if you increase it by 5%, it's a 300 million dollar annual business case. My two million dollar data warehouse optimization, it's irrelevant. It's peanuts in a 300 million dollar annual business case. It's things like that that's helping machine learning and Hadoop evolve in the ecosystem. The cost-reduction play was just a way to slide the infrastructure in. You can do a lot more with it. >> And when you're selling to the CIO's and business leaders, that resonates. >> Ghoutham: Yeah. Absolutely. >> Great. We'll have to leave it there. Thanks very much for coming to theCUBE, Ghou. >> Ghoutham: My pleasure. My pleasure. >> Alright keep it right there everybody. We'll be back with our next guest. This is theCUBE. We're live at Big Data NYC. Be right back. (techno music)