You might have heard this from others, or experienced it first hand. RPA bots keep breaking every few weeks. While this may be the experience of many, I had a different experience with what I built with Autosmic. I think this problem people see is a function of a few different factors:
- The problem you are solving may involve applications that change a lot
- The way the bots were built, made them prone to the application changes
- The RPA technology you use uses object detection technology that relies on underlying application page changes
Naturally, you wonder, how does one make the bots resilient?
- First of all, you got to understand what kind of changes you are likely to see. For example, a web application from a cloud provider you have no control over, can change in whatever way they want it to (of course most of them are not going to do a UI re-design every month!). Based on that, there are a variety of techniques you can use to build your bot
- RPA is a skillset that needs to be acquired over some time. Similar to programming, if the RPA developer uses correct techniques and relies on more robust unchanging entities, bots won’t break often.
- With Autosmic, we use computer vision to eliminate page/html changes beneath the hood. We don’t want those changes to inadvertently break our bots. The result is that if the pages are visually unchanged, mere html changes don’t affect the bots. In addition, if the objects do shift around, that can be accommodated as well.
How Do You Fix The Broken Bot?
Assuming this was a change in the web application that caused the visual navigation to break, you may end up re-recording the bot and building and testing it all over again.
With Autosmic, the bot is able to recognize the change on the fly, allow you to provide corrective guidance and it fixes itself while in the flow of execution, and continues as if nothing went wrong. You avoid the entire re-building and testing process. This feature can be enabled or disabled, of course as needed.