Description: UiPath’s Robotic Process Automation (RPA) tools let software perform repetitive on-screen tasks that otherwise occupy huge time for workers. The company says it is the fastest-growing enterprise software company ever. Daniel Dines believes RPA will enable people to do more creative, more human work.
The following transcript has been lightly edited and condensed for ease of reading.
Will Software Robots Make Work More Creative?
(Transcription by RA Fisher Ink)
Kirkpatrick: Daniel Dines, please come and join me onstage. Daniel is the CEO of what is probably the fastest-growing enterprise company in history. UiPath does what’s called robotic process automation or RPA. Thank you for coming.
Dines: Thank you for inviting me.
Kirkpatrick: It’s so good to have you. And so, you put out your product less than two years ago, right?
Dines: Well, it’s all of three years since we—
Kirkpatrick: Almost three years, and you’re roughly at $200 million a year annual run rate. You have 1,700 employees. But get how fast this company is growing. We first met them in the next WeWork office from us and that wasn’t the whole company but that was their New York office. Now they have an entire floor of a giant building on Park Avenue and that was like a year ago. So what is it that you are selling that people want so much?
Dines: Well, we have prepared a video to show a bit more. I think it’s better to play a
Kirkpatrick: You want to start the video? No, let’s wait for the video first because I think you should say what is robotic process automation. I think the video’s a good illustration but before we get to it, tell us what it is. It’s hard to kind of grasp.
Dines: Well, the best way I think is to think of self-driving cars. What are self-driving cars? It’s a software robot, in a way. It’s inside the car and it’s able to follow a route on a map, all right?
But in following the route, the robot gets smarter because it understands the traffic it captured using computer vision, all the other cars by GPS, the acceleration and everything, and it’s able to make decisions related to the traffic. And now, in the context of a business process, like paying an invoice or booking a flight on a website, you can define the process. You say, you should follow these steps, these steps, these steps, right? And our software is able to emulate people exactly. So instead of driving a car, we emulate people using web applications, desktop applications, all sorts of applications. So our technology really looks at the computer screen and simulates clicks, types; everything like a human would do.
Kirkpatrick: And the purpose that it will be put to is in situations where people do repetitive tasks that aren’t any fun but that they do a lot and that sometimes they’re quite involved. And we all do this, even signing up for things or changing from one thing to another. But is it your belief that pretty much all those things can be automated using software robots over time?
Dines: Not all of them. At least with the current technology, I think we can definitely automate what is rote, repetitive. It’s mundane, it’s clicking, you know, check boxes, filling forms, all this stuff that I think all of us, regardless of our job, we face daily. It would be great, it’s like having an assistant delegating all these things and then you have a bit more time to think, to do more creative stuff.
Kirkpatrick: And the place it’s taking off, one of the places it’s taking off a lot, is in call centers, for example, right?
Dines: It’s one of the places but back office is equally important in our—
Kirkpatrick: So give an example of a back office application and then maybe we’ll show the video, which is a little more of a consumer application.
Dines: Look, we currently go into finance and accounting departments and look at invoice processing, for example. It’s a pretty complex process from when you receive a purchase order until you issue an invoice, until you validate the invoice and you pay it in the end. There are multiple people involved. But it’s pretty repetitive in its nature and involves different business applications like an ERP and maybe email and different other applications.
Kirkpatrick: Salesforce, possibly.
Dines: Salesforce, possibly, yes. And with our technology, we look at human activities and we really automate human activities. It’s not really end-to-end, the process. We don’t have a view end-to-end. This is invoice processing but what does Joe do in order to validate this invoice? Joe goes into SAP, searches the invoice number, and matches the purchase order. This is something that we can very easily automate and it brings instant value to the businesses. We can make a business case instantly. And you get return on your investment in like two months.
Kirkpatrick: Wow, that’s very unusual. Maybe we should show the video just to give a sense—but I want to just caution, this video shows a consumer application but it just goes to show how completely a process that used to be tedious can get automated using RPA from UiPath. So let’s roll that video quickly.
In the meantime, I will ask you, while we’re waiting for the video, one of the things that I should have said, when introducing you, is that you were at Microsoft for a long time and you are Romanian originally but you had this idea and you moved back to Bucharest and spent a considerable amount of time developing it there. And it really kind of emerged very fully formed when you launched it. Still, you have a lot of backend in Romania but you’re a truly global company. What gave you the insight to start this company?
Dines: Well, definitely not my experience at Microsoft.
Dines: Because it’s—and seriously, at a big software company, you feel that all processes should be already automated. People within Microsoft, Oracle, even SAP, I think they don’t get the reality on the field. And look, we didn’t start with this vision, building robots. We started actually with a consumer application, something as simple as a dictionary. So in order to build a dictionary, we wanted to build something nice as clicking on any word on the screen and then would offer a pop-up to look up a word or something like this or search on Google. So that was the idea of the application that we built. And we thought getting the text and understanding the positions on the screen is an easy task. But it took us almost a year to build something as—
Kirkpatrick: Like a click-box here—
Dines: On your phone, right now if you long-click on a word, you can look up that word in most applications. So this is what we built, I don’t know, 12 years ago for desktops, for browsers, for every application. So we built this computer vision technology that understands text.
Kirkpatrick: It kind of understands whatever’s on the screen.
Dines: Yes. But it was very difficult to market the consumer application from Romania. So guess how many of you know how to show Romania on the map?
Dines: But we’ve been lucky to discover that this technology can be applied into more of a business setup. And we started to sell it to other software companies. So we pivoted from consumer to software to developers, to a small category. And we evolve it to really understand the screen. So it’s a full computer vision state that understands screens from all the applications in this world. And then at some point we wanted to build something with more of a broader audience because we made kind of a lifestyle business from this computer vision. So then we built this process automation technology, where somewhere—and it’s very visual, drag and drop—you show a bot what’s the process and then the bot executes the process. And it’s using computer vision. And it suddenly took off, really, without—we’ve done everything wrong as a startup. Seriously, we are an open book of mistakes. But somehow a very big market found us.
Kirkpatrick: Well, you found a need that was huge because there is so much repetitive work in so much back office activity particularly, right?
Dines: Yes, we discovered the business process outsourcing industry. I didn’t know that such an industry existed. When I worked for Microsoft—I thought that Microsoft really announced that [LAUGHS] business process outsourcing exists but it’s an industry that generates north of $300 billion per year, employs many millions of people and they all do these repetitive tasks. So starting there, we really understood the use case of our application.
Kirkpatrick: So in a way, it’s automated business process outsourcing, in a sense.
Dines: Well, business process outsourcing kind of set the stage for robotics. Because in order to shift a process from the U.S. to India, they had to simplify this process a bit and make it easily replicable, without the business knowledge. And that made it easier to apply robotics, to apply technology to emulate people doing it.
Kirkpatrick: Does it look we have the video to play or not? We’re good. Okay, let’s roll it.
Video: This is Dylan with the UiPath tackler fantasy football bot. Do you play fantasy football? Do you forget to set your lineup every week? Do you lose excruciatingly close battles with your childhood best friend just because you forgot to switch out Tom Brady on his bye week? This tackler bot is here to help. Allow me to introduce you to R2 I Will Beat You. R2 I Will Beat You is a bot in our U.S. UiPath fantasy football league. As you can see, he has some starters in that are either injured, out suspended, or on bye week this week. Let’s run the tackler bot and see how he can help R2 I Will Beat You set his lineup. When the bot kicks off, it will open the ESPN website and ask us what team we want to navigate to in order to set their lineup. In this case, we will type in R2 I Will Beat You. Click OK, it will navigate to his home page and begin setting his lineup. Here, he’s moving a running back into the suspended running back spot. Here, he’s moving a wide receiver with a higher projection. And again, substituting out a wide receiver that is on bye week. Here, he’s substituting a tight end for an injured tight end. And last substituting a defense for one that is on bye week as well. R2’s lineup is now optimized and set and he’s ready to soundly beat my fellow UiPath employees. Let’s hope he doesn’t learn how to talk trash anytime soon.
Kirkpatrick: Well, it must be a good video if Kai-Fu Lee is recording it. But so that’s a consumer use but in reality, there’s a lot of processes that aren’t that different in, say, processing an insurance claim or doing any kind of back office activity. So the real value opportunity here is to help people do less boring stuff, right? And so how is that working in reality? Because of course a lot of people might think this is going to displace jobs. Is that what you’ve discovered?
Dines: No, it’s not about displacing jobs, it’s about eliminating repetitive activities on one’s daily plate. And on the contrary, we discovered this type of technology increases one’s job security. Because once you are no longer required to do this completely automatable job, you know, repetitive tasks which are really low value for an enterprise, you have more time to do the higher type of jobs.
Kirkpatrick: So, more creative work, more empathetic work, something that really involves interacting with another human being, which is what everybody would rather do than just be entering forms.
Kirkpatrick: Right. So the idea is, in effect, to make work more enjoyable, in the end?
Dines: I think that actually technology makes work more enjoyable and this fear about robots, it’s particularly because we fear something that is similar to us. We don’t fear mobile phones or cloud technology or Bitcoin or whatever. And technology itself has the potential to displace and change a lot of jobs. It’s not robotics. But people fear robotics because they are similar to us, they feel they are similar to us.
Kirkpatrick: So I’d like to ask if there’s any comments or questions from the audience. Because I have more but does anybody have any thoughts or questions? Okay.
I know you talk about, I mean, this idea that robots emulate people and that a robot could therefore be software instead, it’s sort of a surprise. Most people don’t think that robots could be in software so there’s a little bit of a surprise right there that I think you get over. So just talk a little bit about the company’s trajectory. You’ve now been growing—where do you think this goes next?
Dines: Well, we feel certainly that automation will create a new mindset in how we work in the future. And it’s not only about automating existing work, but it’s really about thinking of new work through automation. I think it’s really important that the next generation of business analysts building processes will think automation first. So they will design processes to be automated day one. And that’s the type of technology that can facilitate it. Because we are humans so we will design processes around things that we know, like human-like interfaces, conversational interfaces, visual graphic user interfaces. We cannot design processes around APIs. Because it’s not fast enough, really, and this is a technology that allows us to design automated processes day one and this contributes a lot to the scalability and increased productivity.
Kirkpatrick: And in the end, that’s how technology causes economic growth.
Kirkpatrick: Last question, I know you have an idea that really everyone should have these and you even have a kind of goal in your company of a robot for every person. Talk about that.
Dines: Yes, like a PC on every desktop changed completely how an office operates, changed completely the landscape of computing, we feel that we have to bring a robot to everyone. And that robot should not only take the repetitive tasks but should change the mindset of people, should allow people to embrace the idea. Because really, it’s all about changing in the mindset. And we are happy to work with a few of our customers that plan to deliver over the next couple of years, you know, 150,000 robots into—
Kirkpatrick: In one company.
Dines: In one company.
Kirkpatrick: Yes. And I know in Japan it’s being hugely embraced.
Kirkpatrick: So in the future, most of us, if we were to look at the things we do that are really boring but we have to do them anyway, on our phones, on our PCs, we could envision a scenario where that stuff’s all going to get automated in some way. That’s a great thing to look forward to.
Kirkpatrick: Well, I hope you get a robot for every person, including everyone here. Okay, real quick, sure. Okay, that’s spontaneous but can we get the mic over there real fast?
Yemi: Thank you very much, name is Yemi. Question for you, do you see integration with hardware in your future or do you draw a line there? And question number two, does your software, can it overcome the human tests, things like captcha on forms? Do you think you’ll get to a point where you can overcome that? Can you do that already? Thank you.
Dines: We are not looking into hardware right now, not at all. And second, yes, the robots are getting new skills. Our robots in the current form, maybe they are pretty dumb; they follow rules, they are very smart at understanding computer vision, but they cannot understand captcha. But with the progress of machine learning, we will learn new skills—and not only we. Behind the scenes, we have a marketplace where anyone can publish skills for the robots and these skills are based on AI. That will increase and they will be more and more human-like.
Kirkpatrick: So there’s AI in the backend, but there’s a lot of kind of community sourcing of functionality, which you’re encouraging and which will grow over time so the kind of silly things that we do on our phones are just the things that might emerge in that sort of open marketplace.
Kirkpatrick: So, thank you so much. Really good. Great to have you here and congrats on your tremendous success so far.
Dines: Thank you so much, David.