This article by Wayne Ariola was originally published on CIO.com
RPA is clearly the IT media darling of the year. Just go to any business or tech publication, and you can’t escape headlines touting RPA as the fastest-growing market in enterprise software. Per Gartner, the RPA market grew over 63% last year. Organizations plan to invest $10 to $20 million annually in RPA solutions to increase productivity, reduce costs, improve customer satisfaction, and help employees focus on creative, higher-value work. “A bot on every desktop” is the new mantra, and companies that promise to deliver this are valued in the billions. Is all this exuberance justified? Or are we looking at another Google Glass?
As a business person, I can clearly see the reason for the demand. Digital transformation and business optimization initiatives want to eliminate latency and waste from the business process, and RPA seems like magic that automates away these inefficiencies. However, as a technologist, I sometimes feel like RPA is the poor man’s API: a tech hack that strings together sequences which should be actually solved by good application architecture or an integration strategy. When you describe RPA’s general approach to a technologist, you get an array of responses ranging from laughter to confusion.
Once RPA advances from initiative to implementation, a more pointed technical limitation emerges: bots break. For better or for worse, bots do as they’re told. Most bots are programmed to use “screen scraping” or visual recognition techniques to automate tasks. This means that if all the expected elements are in the expected places, repetitive, rule-based tasks can be completed much faster and more accurately than they could with humans. However, modern business applications are highly dynamic. New features, bug fixes, changing dependencies (e.g., connected third-party APIs and applications), and even design updates might not phase humans completing a task, but they could make an obedient bot’s head spin. As a result, most bots are as “high-maintenance” as a Hollywood star.
This dichotomy between business vision and technical reality leads to what I have coined “The RPA Death Spiral.” Here’s how it typically plays out:
- With a distinct business benefit in mind, the line of business (LOB) requests an application optimization or integration through the designated development channels.
- The LOB fails to get that request prioritized by development.
- In true “rogue IT” fashion, the LOB bypasses development and tries to achieve the optimization/integration via RPA.
- With the help of a third-party, the LOB automates the process with a bot.
- One of the applications involved in that process changes (e.g., an upgrade, UI change, application logic change, etc.).
- The bot breaks.
- The LOB needs technical assistance to fix the bot.
- The process repeats.
To avoid this death spiral, RPA bots must be resilient to change, and bot maintenance must be simple and straightforward—which leaves script-based approaches out of the equation. Otherwise, RPA is simply short-lived automation that creates technical debt while leaving an organization worse off than it started.
For RPA to deliver the expected business value, it’s absolutely essential that LOB process owners have a fast, reliable way to keep RPA in sync with the evolving process. That’s why Tricentis RPA taps 12+ years of research and development in resilient automation to provide the industry’s most resilient bots. With our model-based automation, bots don’t break easily—they rapidly adapt to application changes. We solve the most significant challenges affecting RPA implementation today: high RPA downtime and lack of skilled resources required for bot creation and maintenance.
Learn more about our unique approach to resilient RPA bots in Resilient RPA with Tricentis Model-based Automation.