After a year filled with robotic process automation pilots, agencies are focused on truly operationalizing in 2019.
According to the Trump administration's management team, about five percent of federal occupations could be entirely automated, while 60 percent of existing jobs could be at least partly automated. That focus on automation is giving cover to agencies looking to move their robotics process automation from the drawing board or the pilot stage into production.
The General Services Administration is working on an internal playbook for developing robotic process automation bots within GSA and plans to begin work in 2019 to create a similar, more general, document for governmentwide use, according to GSA's RPA Program Director Ed Burrows.
RPA is a technology used for automating digital tasks like data manipulation, data entry, email handling, data reconciliation, and even optical character recognition to extract data from documents. Conceptually, the technology is much like a macro or automation script that can be created in a word processing or email progam, but RPA can work across multiple applications and even have bot-specific access privileges.
The federal government's first bot went live in 2017, and 2018 saw dozens of pilot projects -- with several moving into full production. There a number of companies offering projects in this space: Automation Anywhere, UiPath, NCI and Blue Prism are a few of the specialists.
"As a company we've gone from the one or two bots in one or two agencies that they had last January, to we have 25-27 different agencies now that are all using UiPath," said Jim Walker, UiPath's director of public sector marketing. "We're working with them now to try to find out where their roadblocks are."
This quick growth has resulted in agencies looking for resources to learn how to implement these automation technologies within government. When Burrows started speaking about RPA at industry and government events, attendees began reaching out for advice on how to bring the tech to their agency.
"In the course of doing that I realized there was a real need across the government for information sharing in particular and developing best practices and having some type of written implementation and operations guide that would really help agencies," he said. "So we're going to try to address that need in a couple of ways."
One way is by creating a community of practice through digital.gov to help connect the people in the federal government who are interested in and focused on RPA. It would be through this community that GSA would create the government-wide playbook on RPA. There are tentative plans to formally announce the community -- which will be similar to, but separate from, GSA's Emerging Citizen Technology program -- in January 2019.
GSA, meanwhile, has been working on automating some of its own internal processes with RPA. One application enters transaction information into GSA's financial system for purchase card holders, another sends notifications when invoices are nearing their due date to help comply with the Prompt Payment Act.
But GSA plans to use 2019 to begin implementing RPA on even larger "higher value" applications. One area where they see a lot of potential, Burrows said, is with data entry.
"We're finding higher-value processes in the future to automate, meaning large labor hour savings," he said.
Right now, GSA has six active bots with four in the CFO office and two in the Public Building Service. Pre-automation, those processes together were taking 12,000 hours per year for humans to complete. GSA expects to be able to automate hundreds of thousands of hours over the next year of two, Burrows said.
"We're not looking at personnel impact other than so far limited changes in role," he said. "But most of what we've automated hasn't been full time jobs for people, so they've been able to spend more time on their main jobs.
"I think the personnel impacts are further down the road, they will happen at some point."
John Bergin, the executive director for the Information Sharing and Services Office in the Department of Defense, said earlier this year it is all but inevitable that RPA will replace jobs, allowing agencies to spend tax dollars elsewhere.
"Look, I have to buy F-35s, I have to buy Columbia-class replacements." Bergin said. "I need to protect the country, I don't need to protect people's jobs. As a steward of the taxpayer, I must protect the taxpayer base."
National Science Foundation CIO Dorothy Aronson told GCN earlier this year that her agency is considering giving employees the ability to create their own RPA bots. NSF wants to give employees this ability within the next year, she said in October.
"People who have the business knowledge and understand what repetitive tasks they have should be able to use the software … to simplify their own daily work, to move away from those repetitive tasks and actually do more strategic thinking," she said.
The Defense Logistics Agency is the first agency working with UiPath to deploy RPA straight to the cloud, Walker said. Most other agencies are going with an on-premise option, but DLA decided to use a non-public internal cloud to host its RPA solution.
John Lockwood, RPA program manager in DLA Information Operations, outlined a number of business process that could benefit from RPA in a public statement.
"From imaging laptops to configuring servers, from onboarding employees to password resets, any activity that is structured work with well-defined business rules is a potential candidate for RPA," Lockwood said.
More agencies are likely to deploy their first bot in the new year, Burrows predicts. The number of agencies to deploy a bot into full production is still relatively small -- seven or eight -- and most of those only have one, though some leaders in the field have more. NASA and the Food and Drug Administration both have more than a dozen, Burrows said.
"You're going to see an increase in the number of agencies with live bots and you're also going to see those that are furthest ahead ramping up more significantly," he said.
NEXT STORY: The challenge of IA implementation