Enterprises are rushing to embrace robotic process automation (RPA) but, as we covered in The Barriers to Robotic Process Automation, Part 1, they’re also hitting considerable challenges along the way. While we cited prominent barriers to the effective implementation of RPA in part one, here, we will provide practical ways organizations can overcome these barriers.
One of the biggest causes of RPA failure is the lack of executive-level sponsorship for such initiatives. When it comes to RPA success, getting management’s attention–and keeping it–is essential. One way to do this is to focus on RPA quick wins and efforts that have a measurable return on investment.
Next, IT leaders need to understand how significant digital transformation is to AI technology.
“An efficient digital transformation strategy will eliminate internal IT silos through integration and automation, leading to opportunities for AI projects,” said Marcel Shaw, AI expert and Federal Systems engineer at Ivanti. Shaw pointed to how organizations that interface with customers will see opportunities for chatbots that leverage natural language processing technology, thereby reducing demand on operators while enhancing the user experience.
“IT leaders will be able to take advantage of machine learning technology that will support and enhance complex automated processes that will fulfill customer requests from beginning to end, without intervention from a human analyst,” said Shaw.
Such benefits could make RPA an easy sale to executive leadership. Other benefits that can be used to sell RPA to leadership include cost savings, more efficient processes, improved process quality and free staff to focus on higher-value tasks.
“The only thing holding AI back at this time is a lack of experience with the technology. AI will take many years to develop. Some estimates point to a 75-year maturity lifecycle for AI. AI technology is designed to learn. As it learns, it will correct mistakes and mature, much like a human has to learn,” said Shaw.
Next up: The high level of investment needed for effective, automated and intelligent robotic automation.
“There’s a great deal of catching-up to be done before most companies can even consider building out their own advanced developments with AI,” said Salvatore Stolfo founder and CTO at Allure Security. “It may be more cost-effective and faster to partner with a software vendor that has already put the time into dedicated research and development to build an AI-powered solution, and then customize that solution for your particular use case.”
Simha Sadasiva, CEO at Ushur might agree. “Many intelligent automation solutions on the market today require a heavy IT investment. This can include hiring an outside third-party just to manage the platform they’ve invested in,” said Sadasiva.
“Those attempting to manage the technology in-house are finding that it requires personnel with sophisticated developer-level knowledge, which is in short supply today. These issues are causing some companies who were initially interested in exploring intelligent automation solutions to abandon their projects altogether,” added Sadasiva.
Sadasiva said one possible way to solve this is to evaluate potential zero code platforms. “These are SaaS-based intelligent automation solutions born in and live in the cloud,” said Sadasiva.
Finally, and while perhaps self-evident is often overlooked: know what needs to be achieved with robotic process automation.
“An organization needs to identify what the problem is that they are trying to solve with AI. This needs to be the most important pain point to them at that point in time. Inherently AI requires data to be able to train the machine to learn,” said Ben Mercer, co-founder at personalized web experience provider Personify XP.
“The challenge for some organizations, when it comes to accessing data, is that valuable data is stored on legacy systems and difficult to extract. In cases like this, it may make more sense to train the AI on new data, than fighting with legacy systems that may delay or reduce the ROI of the deployment,” said Mercer.
Effectively managing data, AI and the resulting automation requires staff with strong skills and experience in these areas.
“Regarding skillset development, most of the skills needed can be learned, but you need to find different types of team members who have a great comprehension of data and database management and secondly a development background,” said Jesse Tutt, CEO at mattress company Gotta Sleep.
Finally, chances are if an enterprise is planning to do something there’s already someone who has tried and has lessons to share. “Every automation project has likely been done by someone else already. Tips and tricks are freely available from others who have more experience so networking on Linkedin, meetups, and conferences are key,” adviced Tutt.