Early days were characterized by excitement over the dramatic productivity and cost-saving benefits enabled by RPA. Over time, however, the limitations of rules-based bots have emerged. For one thing, basic RPA tools can’t adjust to new conditions or changes in their environment. Even the slightest deviation from the process they’re trained to follow triggers an exception that requires a human to step in, thereby sapping the solution’s productivity.
Another issue is the complexity surrounding deployment of RPA bots. While instructing a bot to perform a task is relatively easy, it does involve a level of programming expertise. Most end users of RPA are on the business side and lack the requisite technical knowledge. That means that setting up a bot requires an RPA programmer. Demand for RPA skills, meanwhile, is through the roof. (Witness the volume of urgent “we’re hiring” notices on LinkedIn pleading for people with Automation Anywhere, Blue Prism and UiPath certifications.) As a result, because the intervention of scarce technical resources is required, bottlenecks often occur when deploying a bot for a business user.
Today, these limitations are being addressed, as advanced cognitive, machine learning and natural language processing capabilities are being integrated to enhance basic RPA functionality. The 1+1=3 synergy of RPA and intelligent tools has myriad potential applications, such as processing unstructured data and targeting customers for specific product offerings. The combination is also enabling RPA bots to “learn” from mistakes and to understand how people communicate.
Machine learning capabilities allow a bot to identify a mistake and apply a fix. An advanced bot of this type flags an exception, and when that exception is fixed by a human administrator, the bot captures the fix and applies it going forward without requiring a rewrite of the bot’s original instructions. This results in streamlining the task optimizing RPA tools. Natural language processing tools, meanwhile, analyze the context of words and phrases, enabling business users with minimal technical training to interact with RPA systems by using plain English commands and different phrases that mean the same thing.
In addition to extending the cost-saving and productivity benefits of RPA, these enhancements allow enterprises to focus more energy on strategic aspects of shared services and business process outsourcing, rather than continual adjustments to bots.
Daniela Henriques, a Director at Softtek specializing in cognitive-enable RPA, has a new article in Outsource Magazine that discusses the impact of machine learning and natural language capabilities on RPA in greater detail.
Alex Kozlov, Director of Content for Softtek US & Canada
Alex Kozlov is Director of Content for Softtek US & Canada, a global IT services provider. He has more than 25 years of communications and media relations experience in technology and outsourcing, and writes extensively about Artificial Intelligence, robotics, cognitive computing and related topics.
As the Robotic Process Automation (RPA) market matures, enterprises are taking stock of lessons learned and exploring ways to take existing RPA capabilities to the next level.
Early days were characterized by excitement over the dramatic productivity and cost-saving benefits enabled by RPA. Over time, however, the limitations of rules-based bots have emerged. For one thing, basic RPA tools can’t adjust to new conditions or changes in their environment. Even the slightest deviation from the process they’re trained to follow triggers an exception that requires a human to step in, thereby sapping the solution’s productivity.
Another issue is the complexity surrounding deployment of RPA bots. While instructing a bot to perform a task is relatively easy, it does involve a level of programming expertise. Most end users of RPA are on the business side and lack the requisite technical knowledge. That means that setting up a bot requires an RPA programmer. Demand for RPA skills, meanwhile, is through the roof. (Witness the volume of urgent “we’re hiring” notices on LinkedIn pleading for people with Automation Anywhere, Blue Prism and UiPath certifications.) As a result, because the intervention of scarce technical resources is required, bottlenecks often occur when deploying a bot for a business user.
Today, these limitations are being addressed, as advanced cognitive, machine learning and natural language processing capabilities are being integrated to enhance basic RPA functionality. The 1+1=3 synergy of RPA and intelligent tools has myriad potential applications, such as processing unstructured data and targeting customers for specific product offerings. The combination is also enabling RPA bots to “learn” from mistakes and to understand how people communicate.
Machine learning capabilities allow a bot to identify a mistake and apply a fix. An advanced bot of this type flags an exception, and when that exception is fixed by a human administrator, the bot captures the fix and applies it going forward without requiring a rewrite of the bot’s original instructions. This results in streamlining the task optimizing RPA tools. Natural language processing tools, meanwhile, analyze the context of words and phrases, enabling business users with minimal technical training to interact with RPA systems by using plain English commands and different phrases that mean the same thing.
In addition to extending the cost-saving and productivity benefits of RPA, these enhancements allow enterprises to focus more energy on strategic aspects of shared services and business process outsourcing, rather than continual adjustments to bots.
Daniela Henriques, a Director at Softtek specializing in cognitive-enable RPA, has a new article in Outsource Magazine that discusses the impact of machine learning and natural language capabilities on RPA in greater detail.
Alex Kozlov is Director of Content for Softtek US & Canada, a global IT services provider. He has more than 25 years of communications and media relations experience in technology and outsourcing, and writes extensively about Artificial Intelligence, robotics, cognitive computing and related topics.