Robotic Process Automation with Allen Chan

Speaker 1: From IBM Cloud and Cognitive Software, you're listening to The Art of Automation, with host, Jerry Cuomo.

Jerry: Hey! Hey! Thanks, Ethan! Ladies and gentlemen, welcome to The Art of Automation, a podcast that explores the application of AI- powered automation in the enterprise. I recently read some things, interesting, from a study by the IBM Institute for Business Value. Analysts estimate that enterprises spend billions of hours a year on low- value work that strips people from focusing on things that really matter the most. In this study, they further predict that AI- powered automation will release 134 billion in labor value in 2022 alone. In this episode, we're going to look closer at robotic process automation, or RPA, as one example of a technology that businesses are using to shift those billions by automating low- value work, and with AI, helping your employees become superhuman workers as they tackle the tasks that matter most to your business. For this conversation, we are lucky to have a real, subject- matter expert in all things surrounding business process automation. IBM Distinguished Engineer and CTO of Digital Business Automation at IBM, Allen Chan. Allen has spent over a decade on progressing the technologies behind business process, including workflow, low- code application development, and case management. Recently, Allen was part of the team at IBM that worked on the acquisition that led us to join forces with WDG Automation, which is all about robotic process automation, getting it to work in the enterprise, but this is just the starting point. Allen is also responsible for our technical strategy and the delivery of our next- generation RPA technology. Oh! Did I say that I have the privilege of working closely with Allen, as we are amongst the co- creators of the IBM Automation platforms? With that, I'd like to welcome Allen to The Art of Automation. Welcome, Allen!

Allen: Thank you, Jerry and Ethan, and thanks for your glowing introductions. I'm happy to be here to answer your questions, or maybe have a discussion around RPA-

Jerry: As usual, Allen, let's go right at it. What exactly is robotic process automation, and how does it differ from things like writing a script, and of course, can you give a simple example?

Allen: Okay. Cool. That's a good question. Robotic process automation, or RPA for short, is really a program. It's kind of like a script, but other than just running a script on the background or command line, it's a script that mimic human interactions with their desktop to perform tasks. Some examples are copying information from an Excel spreadsheet to a form, or performing Excel tasks on a website. While we assume, in today's digital world, a lot of human tasks have been automated already, reality is there's still a large portion of our daily work that require manual labors, and many of those are repetitive. Now, imagine if you are a data clerk, responsible for processing incoming invoices sent to you. You will be reading invoices from a email or a fax. It could be a PDF document. It could be a scanned image. You'll be reading that invoice, and then, transcribing the information from that virtual paper document into, maybe, a web- based application. That's exactly where RPA can come to help. If we have RPA, we leverage various OCR technology. OCR stands for optical character recognitions, or the more advanced version could be leveraging intelligent document processing technique to read the invoice from your images, and then, you manipulate your mouse and keyboard on the screen to enter the information on behalf of you.

Jerry: Hmm. That's pretty straightforward. Thanks for that. Allen, this is all about AI- powered automation, so how and where will AI help? Of course, could you give an example of RPA in action with AI?

Allen: Okay. Sure. There are actually several area where AI is widely used in RPA. If you think about the premise of RPA, it's trying to mimic the human manipulations of keyboard and mouse, and how human interact with your desktop. There's couple area. One is reading unstructured and semi- structured documents. Take example of the invoice. You need to be able to fully understand the invoice coming in, and be able extract the correct information. More recently, we have made advancements, like the RPA of automation document processing capability, where you leverage deep- learning technique to extract the relevant information from invoices. Another example where RPA leverage AI is around chatbot or voice. RPA can integrate with system, like Watson Assistants, where user can use a chat interface to interact with the robot in the background to drive work. A third area would be around screen automation, using computer vision, helping the robot to actually understand the different construct within the UI interface, that this is robot, or button, or whatnot-

Jerry: I think that's key, because if you're scripting, it seems to me like errors could occur, maybe, if the button is a few pixels off. It sounds like what you just said, that if AI can actually recognize that a button is a button, it really doesn't matter where it is.

Allen: Exactly. You can ask a bot to say, " Look for the okay button," or, " Look for the reject button," or, "Scroll down." Yeah.

Jerry: Yeah. Good. Good. Good. Allen, anytime this conversation comes up, I always get the question, " Are the bots going to take over the humans," or said another way? Do we need to worry about our jobs? What's the relationship between bots and humans?

Allen: I see bot as a tools that human can use to make their life better. Right? If you inaudible our civilization or history, you see human inventing new tools. Every time a new tools come about, our human intuitions and inaudible help us to propel our next step forward. I see RPA as a tools that will help us to take care of some of the work that we don't want to do anymore, but that also represent opportunity for human to actually look for the next big things to work on.

Jerry: So, different jobs, maybe?

Allen: Yeah. Maybe, yes-

Jerry: Yeah! Great! Allen, who is RPA for, based on this? Are there any boundaries? What is it good for, and maybe, what it's not good for?

Allen: Okay. There's typically several area that people would use RPA on. One is, if you look around your organization, you see a large population of task worker that doing a lot of many work. That's the opportunity for automation, such as processing from email or record reconciliation between systems. We are also seeing area where using RPA to automate not the entire task, but just part of the task. I generally call those micro tasks. An example of that would be a sales executive trying to create report, maybe around sales strategy. As part of the report generation, the exec might need to copy a large set of information from different system, while the actual strategy would still require intuition experience from a sales executive. Some of these many work can be automated using RPA bots. What are the limitations? In some sense, robotic process automation is a misnomer. It's actually not automating the entire process, it's just automating a task. If you attempt to use RPA to automate the entire processes, that's usually where we see people breaking down, in terms of the expectation. Any task that require large amount of human cognition, intuition, is probably a good candidate, as well. The last, but not least, is RPA does not require API integration. If there is already a system with API, usually, in all cases, is more robust to go for API- based integration, especially in high inaudible, large- scale scenarios.

Jerry: Then, Allen, with those limitations, where are we going next with RPA?

Allen: Well, part of it is looking at where RPA can do best, is excel in human task automation, right? There are generally four area. One is the robot will get smarter and try to understand the world better by leveraging AI techniques, such as reading document, computer vision, document extractions. I also see where everyone would have a bot on their desktop. This get into the attendant bot scenario, where the bot are automating part of the workflow on a daily basis, but not everything on the inaudible. The third area is creation of bot will get much easier. There will be better recording capability. There will be more intrinsic use of computer vision, and the leveraging of task- mining technique will be more pervasive to create bots.

Jerry: Okay. That's great, Allen. Both impressive and inspiring, we really see, now, the power of RPA coming to bear. Ladies and gentlemen, you've been listening to The Art of Automation with Allen Chan, IBM Distinguished Engineer and CTO of Digital Business Automation at IBM. Okay. We started this episode talking about enterprises spending billions of hours a year on low- value work, diverting focus from things that really matter most. We saw and heard great examples from Allen on how RPA and AI can use advanced automation techniques that will release significant dollars, perhaps billions in labor, in 2022 alone. I think it's certainly plausible! Oh! Thank you, again, Allen! That's it! Once again, want to thank all of you for listening, and this is Jerry Cuomo, IBM Fellow and Chief Technology Officer of Automation at IBM. See you, again, on an upcoming episode.