What is a good use case for a robot? Is there something, what is not to do with other integration tools?
HiQ is strong ing hyperautomation business. Hyperautomation means process automation with integration platform that utilizes also AI and RPA technology seamlessly. This is why also we have RPA professionals at HiQ and we provide UIPath to go with our integration platform frends. Of course frends can orchestrate any RPA tool and not just UIPath. But so what? Who wants to use them anyway?
I am so called RPA-sceptic. But I have my reasons for it. When RPA hype started, many used RPA tools "wrong way" and made complex recordings that run a long time even though most of the tasks could have been executed faster and less error prone with any integration platform. RPA seemed to be interesting to companies for two reasons:
RPA was cheaper to implement than traditional integrations (even though the operations might be more expensive - but that is not always discussed when the tools are purchased)
RPA recording is easy to understand and to specify even for a less technical person because the robot simulates an actual person and one doesn't have to understand the process beneath
Many of the RPA examples I encountered would have been simpler if proper interfaces would have been used instead of the UI and the messages would have been mapped with a mapping tool instead of an Excel sheet. I wouldn't describe myself as a person that opposes change, but at this point I was eager to do facepalms and swear like Captain Haddock.
First RPA offer I was involved in was a horrible experience. The RFI included some use cases and every one of them screamed "Yep, we didn't want to purchase the add on to our expensive ERP and use interfaces. We decided to use the UI and export Excel sheets even though that is quite a kludge." At that point when the use case described that the robot should "wait for a large excel file to open" I wasn't sure if I should laugh or cry. I decided to get more coffee. We didn't win that offer - wonder why.
So I think I can say that my relationship with RPA didn't start smoothly. Because I'd like to consider myself to be an open-minded person, I decided to give RPA a new and fair chance.
One of our customers has purchased licenses to both frends and UIPath and they have had a PoC with RPA. After this one case they haven't utilized RPA, but utilized their integration platform frends or other systems. While working for this customer I encountered a case where there was no interface to query some information but one had to use a website to download the data. Yes, that system is very 2000 but let's not go there now. I started to think if this finally was a case for the damn robot and suggested RPA to the customer. They replied that the use case was way too complex and error prone and they would rather download the data manually to csv and import that to the target system. I believe they were right. Unfortunately this experience didn't improve my relationship with RPA: apparently if integration platform couldn't do it, either couldn't RPA.
I have still not lost my hope with RPA. Since then I have read of few use cases where RPA has been (at least then) a proper choice. RPA products have, among other things, capabilities to read text from pictures. One can purchase many other products that do that too, for example Azure AI tools. I wouldn't necessarily purchase RPA product if I want to read my receipts to text. But if I would happen to have a RPA solution already and need to read pictures to text, why not use it?
This seems to be the case with many RPA solutions: you have other ways to do the same thing. You can fill in the forms and print pdf files with a robot. But there are solutions that take for example xml input and export preformatted pdf files with the given data. In the end the question is, what is the solution landscape in the company. It rarely pays off to purchase a solution for just one need but to use existing ones if that is anyhow possible.
So the answer is no, RPA is not just for the stupid and the cheap. Sometimes it is the best available option. One should keep in mind though, that when hyperautomation is implemented correctly, most of the tasks are run on integration platform that utilizes robot for some single task. Integrations are usually faster and more robust than RPA recordings. And I am still willing to make specifications for a RPA case if that is ever needed. But I'm not holding my breath.
Minttu is always ready to change her opinion after she has heard good justification.