 Live, from Miami Beach, Florida, it's theCUBE, covering UiPath Forward Americas, brought to you by UiPath. Welcome back to Miami, everybody. You're watching theCUBE, the leader in live tech coverage. You go out to events, we extract the signal from the noise, a lot of noise here, but the signal's all around automation and robotic process automation. I'm Dave Vellante, he's Stu Miniman, my co-host. Guy Kirkwood's here, he's the UiPath chief evangelist, otherwise known as the chief injector of Kool-Aid. Welcome to Craig Leclerc, the vice president of Forester. Covers this market, wrote the seminal document on this space, knows it inside out, great to see you again. Yeah, nice to see you again, it's great to be back to CUBE. So let's start with the analyst perspective. Take us back to when you first sort of discovered RPA, why you got excited about it and sort of what Forester's research is all about in that space. Yeah, it's been very interesting ride. Most of these companies, at least that are the higher value ones in the category, they've been around for a long time. They've been around for over a decade and no one ever heard of them three years ago. So I had covered at Forester business process management and some of the business rules engines and some of the, I've always been in process. And I just got this sense that there was a way that companies could make progress in digital transformation and overcome the technical debt that they had. A lot of the progress has been tepid in digital transformation because it takes tremendous amount of time and tons of consultants to modernize that core system that really runs the company. So along comes this RPA technology that allows you to build human equivalents that patch up the inefficiencies without touching. I came in on American Airlines and the system that cut my ticket was designed in 1960. It's the same Saber reservation system. So that's the big obstacle that a lot of companies have been struggling to really take advantage of AI in general. A lot of the more moonshot and more sophisticated promises haven't been realized. RPA is a very practical form of automation that companies can get a handle on right now and move the dial for digital transformation. So Guy, we heard a vision set forth by Daniel this morning. Basically a chicken in every pot, I call it. Robot for every person. Now, what Craig was just saying about essentially cutting the line on technical debt. Do you have clear evidence of that in your customer base and maybe you could give some examples? What we're really seeing is that as organizations have to deal with the stresses, what Leslie Wilcox, professor at LSE, describes as the stresses within organizations and particularly in environments where the demographics are changing. What we're seeing is that organizations have to automate. So the best example of that is in Japan because the Japanese population peaked in 2010. It's now falling as a whole. Plus all the baby boomers, people of Craig's and my age are now retiring. So we're now in a position where they measure levels of dangerous overwork as being more than 106 hours a week. That isn't 106 hours a week in total. That's 106 hours a week in addition to the 60 hours a week the Japanese people normally work. And there is a word in Japanese, which is koashi, which means to work once after death. So there really is no choice. So what we're seeing happening in Japan will be replicated in Western Europe and all and certainly in the US over the next few years. So what's driving that is the rise of the ecosystems of technologies of which RPA and AI are part. And that's really what we're seeing within the market. Sometimes these big waves, particularly in infrastructure, like you saw, you kind of saw it with virtualization and some other wonky techs like data reduction. They could be a one time step function and not an ongoing sort of business value creator. Where does RPA fit in here? How can organizations make sure that this is a continuous of business value generator as opposed to a one time hit question? Yeah, well I like the concept of RPA as a platform that can lead to more intelligence and more integration with AI components. It allows companies to build an automation center or a center of excellence focused on automation. But the next thing they're going to do after building some simple robots that are doing repetitive tasks, is they're going to say, oh, well wouldn't it be better if my employee could have a textual chat with a chat bot that then was interacting with the digital worker that I built with the bot? Or they're going to say, you know what, I really want to use that machine learning algorithm for my underwriting process, but I can use these bots to go out and collect all the data from the core systems and elsewhere and from the web and feed the algorithms so that I can make a better decision. So again, it goes back to that backing off the moonshot approach that we've been talking about with that AI has been taken because of the tremendous amount of money spent by the major players to lay out the promise of AI has really been a little dysfunctional in getting organizations eye off the ball in terms of what could be done with slightly more intelligent automation. So RPA will be a flash in the pan unless it starts to embed these more learning capable AI modules, but I think it has a very good chance of doing that, particularly now with so much investment coming into the category, right? Craig, it's really interesting when I heard you describe that it reminds me of kind of the home automation, the Cortana's and Alexa's and consumer side where you're seeing this, you've got the consumer side where you can build skills yourself, teenagers, people can do that. One of the challenges always on the business side is how do you get the momentum when you don't have the consumer side? How do those interact? It's the technical debt issue and it's just like the mobile peak in 2011. Consumers in their hands had much better mobility right away than businesses. It took businesses five, they're still not there and building a great mobile environment. So these, Alexa and our kitchen snooping on our conversation and to some extent Netflix that observes our behavior, that's a light form of AI. You know, there is a learning from that behavior that's updating an algorithm autonomously in Netflix to understand what we want to watch. There's no one with a spreadsheet back there, right? So this has given us in a sense, a false sense of progress with all of AI. The reality is business is just getting started. Business is nowhere with AI. RPA is an initial foray on that path. We're in Miami, so I'll call it a gateway drug to. Yeah, well in fact, there's also an element that the series, the Cortana's, the Electors are very poor at understanding specific ontologies that are required for industry. And that's where the limitation is right now. So we're working on an organization called Humly that focused on those ontologies for specific industries. So if the robot doesn't understand something, then you could say to the robot, okay, sit that in the Wells account if you're in a bank and it understands that Wells in that case means Wells Fargo. It doesn't mean a hole in the ground with water at the bottom or a town in Somerset in the UK, because they're all Wells. So it's getting that understanding correct. You know, I wonder if you guys could comment on this. Stu and I were at Splunk earlier this week and they were talking up NLP and, you know, we were saying, you know, but one of the problems is that NLP sometimes not that great. And they made a comment that I thought was very interesting. They said, you know, frankly, a lot of the stuff that we're ingesting is text. And it's actually pretty good. I would imagine the same is true for RPA. Is that what you see? You were talking about that on stage with across the text analytics. Yes, so, you know, RPA doesn't handle unstructured content the way that NLP does. So NLP can handle voice, it can handle text. For the bots to work in RPA today, you have to have a layer of analytics that understands those documents, understands those emails and creates a nice, clean file that the bots can then work with. But what's happening is the text analytics layer is slowly merging with the RPA bots, platforms, so it's going to be viewed as kind of one solution. But it's more about categories of use cases that deal with forms and documents and emails rather than natural language, which is where chat bots are grounded. So known business processes, really. Known business processes. One example we've got live is an insurance company in South Africa called Hallard. And they've used a combination of Microsoft cognitive toolkit, plus IBM Watson, and it's orchestrated doing NLP and orchestrated by UIPALT, so that's dealing with utterly unstructured data. That's the 1.5 million emails that that organization gets in a year. They've managed to automate 98% of that, so it never sees a human. And the reduction in cost is 91% cost and reduction per transaction. And that's done by one of our implementation partners, a company called LARC AI down there, it's superb. Yeah, so okay, so text analytics is hard. You get, last several years, we have that sentiment out of it, but if I understand it correctly, Craig, you're saying if you apply it to a known process, it actually can have outcomes that can save money. Absolutely, yes. As Guy was just saying. I think it's moving from that rules-based activity to more experience-based activity as more of these technologies become merged. Well, the technology in your view advanced to the point, because the known processes, okay, there's probably a lot of work to be done there, but today, there's so many unknown processes. It's like this messy, unpredictable thing. Will machine intelligence, combined with robotic process automation, get to the point, and if so, when, that we can actually be more flexible and adapt to some of these unknown processes? Or is that just decades on? Oh no, I think we talk at Farster about the concept of convergence, meaning the convergence of the physical world and the digital world. So essentially, digital's getting embedded in everything physical that we have, right? This is think of IoT applications and so forth. But essentially, that data coming from those physical devices is unstructured data that the machine learning algorithms are going to make sense of and make decisions about. So we're very close to seeing that in factory environments. We're seeing that in self-driving cars that fleet managers that are now understanding where things are based on the signals coming from them. So there's a lot of opportunity that's right here on the horizon. Craig, a lot of the technologies you mentioned, we may have a lot of the technical issues sorted out, but it's the people and the interactions. Some things like autonomous vehicles, there's government policies, we're going to be one of the biggest inhibitors out there. When you look at the RPA space, what should workers, how do they prepare for this? How do companies make sure that they can embrace this and be better for it? Yeah, that's a really tough and thoughtful question. The RPA category really attacks what we call the cubicle population. And we're estimating four million cubicles will be emptied out in five years by RPA technology specifically, which that's how we built the market forecast because each one of the digital workers replacing a cubicle worker will cost $11,000 or what? That's how we built up the market forecast. And there are going to be automation deficits. It's not all going to be relocating people. We think that there's going to be a lot of disruption in the outsourced community first. So companies are going to look at contractors, they're going to look at that BPO contract. And then they're going to look at their internal staff and our numbers are pretty clear. We think they're going to be four million automation deficits in five years due to RPA technology specifically. Now, there will be better jobs for those that are remaining, but I think it's a big change management issue. When you first talk about robots to employees, you can tell them that their jobs are going to get better. They're going to be more human. They're going to have a much more exhilarating experience. And their response to you is, what they're thinking is, damn, robots going to take my job. That's what they're thinking. So you have to walk them up the mountain and really understand what their career path is and move them into this motion of adaptive and continual learning and what we call constructive ambition, which is another whole subject. But there are employees that have a higher level of curiosity and are more willing to adapt to, to get on the other side of the digital divide. Yeah? You mentioned the market. You guys did a market forecast. I've seen, I've read stats a little over a billion today. I don't know if that's consistent. Yeah, that's not right. Yeah, yeah, yeah. And then, is this a 10X market? I mean, when does it get to 10 billion? Is it five, seven, 10 years? So we go out five years and have it be close to 3 billion. I think the numbers I presented on stage were 3.2 billion in five years. Now that's just software licenses and it's not the services community that's around that. Yeah, you probably triple it if you add in services. I think two to three times service license ratio. And, you know, there's always an issue at this point in emerging markets. Some of the valuations that are there, that market, 3 billion has to be a bit bigger than that in eight or nine years to justify those valuations. So that's always the fascinating capital structure questions we create with these sorts of things. So you described this kind of one for one replacement. I'm presuming there's other potential use cases, or maybe not, that you forecast. Is that right? Oh, no, you mean for the cubicles? Yes, it's just cubicle replacement in that 3 billion, right? It's other uplifts. No, no, there are use cases that help in factory automation and supply chain and guys carrying around clipboards and warehouses. There are a tremendous number of use cases, but the primary focus are back office workers that are tend to be in cubicles and contact center employees who are always in cubicles. And then we'll see if the non-obvious ones emerge. Now I think ultimately what's going to happen is the number of people doing back office corporate functions. So that's both the sort of finance and accounting procurement HR type roles and indeed the industry specific roles. So claims processing insurance will diminish over time. But I think what we're going to see is a concomitant increase in the number of people doing customer experience because it's the customer intimacy that is really going to be differentiated organizations going forward. The market's moving very fast. I mean, in reading your report, it's like you were saying, I think yesterday's features are now table stakes. And so everybody's watching everybody else. You heard Daniel today saying, hey, our competitors are watching, we're open and they're going to steal from us. Okay, so be it. Rising tide lifts all boats. What do you advise clients in terms of where they should start, how they should get started, obviously pick some quick wins, but what do you tell people? Well, I always say pretty much the same advice you give almost on any emerging technology. You know, start with a good solution provider that you trust. Focus on a proof of concept, POC and a pilot. Start small and grow incrementally and walk people up the mountain as you do that. That's the solution. I also have this report I call the rule of fives, that there are certain tasks that are perfect for RPA and they should meet these three rules of five. You know, a relatively small number of decisions, a relatively small number of applications involved and a relatively small number of clicks in the click stream. 500 clicks, five apps, five decisions. Look for those in high volume and they have high transaction volume and you'll hit RPA gold. You'll be able to offset two and a half to four FTEs for one bot. And if you follow those rules, follow the proof of concept, good solution partner, everyone's winning. Yeah, practical advice to get started and actually get to an outcome. Anything you'd add to that? In most organizations, what they're now doing is picking one, two or three different technologies to actually play with to start. And that's a really good way. So we recommend that organizations pick three, four, five processes and they do a hackathon. And very quickly they'd work out which organization they want to work with. And it's not necessarily just the technology and in a lot of cases, UiPath isn't the right answer. But that is a very good way for them to realize what they want to do and the people which they'll have to do it. Great, well guys, thanks for coming on theCUBE, sharing your knowledge. Thank you, pleasure. Appreciate your time. Thanks so much for doing it. All right, keep it right there, everybody, Stu and I will be back from UiPath Forward Americas. This is theCUBE, go right back.