Robotic Process Automation (RPA) has gone in just a few short years from a few scattered pilot projects to an essential tool deployed by dozens of agencies. In the process, Federal agency IT leaders have found that automation at scale comes with challenges, and they shared some of their best practices for and lessons learned for integrating RPA into existing systems at an FCW virtual workshop on October 26.

USCIS RPA Program Lessons Learned 

Establishing the RPA Program at United States Citizenship and Immigration Services (USCIS) began as uncharted territory for Meikle Paschal, RPA program manager at USCIS. RPA was a new and somewhat unexplored concept in the government space. And while many are excited about the technology, it’s not always easy to implement.

One of the first challenges the agency encountered was making necessary software available to all users, Paschal said.  To get to that goal, his team realized that they would have to build an enterprise approach.

“We had to integrate the deployment of [this] software with our software center. Through this enterprise approach, [we were able to] deploy the software to our users without having to install it in person individually,” he said.

However, that approach also brought some challenges. The RPA team realized that users began to download bots without oversight or governance, leading to “rogue” bots and automation. The team knew that there needed to be some governance and oversight to combat this challenge, especially if they wanted to program to be sustainable. USCIS laid out rules and processes that users had to follow to build out automation as quickly and efficiently as possible but remain concise with the process.

The rules and processes also ensured that all users developed quality automation for technical excellence within the RPA program. The bots that delivered value were resilient and didn’t fall apart upon deployment, Paschal said.

He also emphasized the need for partners in the process, and said RPA is not a journey that agencies can undertake alone. In its journey, USCIS partnered with a small automation company and shared their findings with other Federal agencies.

Integrating RPA into Existing Systems

Often, agencies will talk about RPA as a singular solution for a host of their automation needs. But according to Frank Wood, supervisory IT program manager for RPA at the Defense Logistics Agency (DLA), RPA lives in multiple capability exchanges, and acts to relieve organizational automation requirements.

“RPA executes pre-programmed steps of a business process using existing IT infrastructure to simulate the actions of a human executing that process,” Wood said.

But there are several workflow characteristics needed for RPA to be feasible and relevant, he added. For example, rules and processes must be implemented that follow a predefined  “decision tree” for RPA to provide valuable data or information.

Additionally, when integrating RPA into existing systems, agencies may encounter unstructured or paper-based data, and that can present problems for RPA tech. Therefore, Wood added that agencies must use cognitive technology to handle unstructured data and have solutions to handle non-digital data.

Wood also emphasized that while RPA is all about integrating systems and providing an enabling technology, rather than being a system on its own. “It glues some systems together in a particular way to imitate and replicate human actions with interfaces, typically web interfaces,” Wood said.

Read More About
More Topics
Lisbeth Perez
Lisbeth Perez
Lisbeth Perez is a MeriTalk Senior Technology Reporter covering the intersection of government and technology.