The General Services Administration has implemented 10 robotic process automation (RPA) bots in the past year and has plans to reach 25 by the end of the fiscal year, following the President’s Management Agenda to shift federal government employees’ work from low-level to high-value tasks.
RPA uses software-based bots to automate mundane, rule-based, high-volume repetitive tasks, like queries, calculations, copying and pasting from system to system and maintaining records.
GSA is already seeing benefits and further-reaching implications with deploying RPA, said Ed Burrows, a senior adviser to GSA’s chief financial officer. In fact, the administration’s current use cases are primarily in RPA, rather than artificial intelligence.
“The President’s Management Agenda has a goal of shifting employees to higher-value work, and RPA is specifically mentioned in that memo from [Office of Management and Budget],” Burrows said at the Jan. 16 ACT-IAC Artificial Intelligence and Intelligent Automation forum in Washington, D.C.
The bots that are deployed and planned for at GSA are primarily for financial operations, Burrows told GovernmentCIO Media & Research. One bot, he said, pulls a file on invoices due for payment and notifies those responsible for accepting the invoices so they can be paid.
This keeps GSA in compliance with the Prompt Payment Act, which requires vendors to be paid within 30 days, according to Burrows.
GSA is also using a bot for its lease process. “We have a bot that goes to our leases, extracts key data and reconciles it with the same data in other systems,” Burrows said. When the bot is running, the lease will appear on the screen, and the bot will begin comparing the data on the lease with other data. “So, it appears to be intelligent,” he added.
And in deploying these bots, Burrows provided some lessons learned, challenges still to overcome and the current and future benefits in RPA and AI investments.
Implementing RPA Processes
It took about a year to deploy the 10 bots GSA currently has, as there are a lot of upfront work in the process, including documentation and analysis. “We found that a majority of our processes had to be changed in some way,” Burrows said. That also takes time as full-time subject matter experts are busy doing their jobs and don’t have much extra time to meet with RPA leads to document tasks. Before an agency can implement RPA and start development, Burrows said these early processes must be done.
In fact, GSA’s capacity right now is about one bot a month, and that’s an area the administration is looking to make more efficient and increase capability.
One way that’s being done is by training staff in bots. Burrows said about a dozen people in the office of the CFO are trained in bots now, but it still seems hard to scale up. “We haven’t gotten there yet,” he said.
What GSA found when it started applying RPA and automating certain tasks was that many of those jobs were being done by mid-level employees. “It’s actually very easy to move them to the more valuable work” because they immediately know what they can do with their extra time, Burrows said.
But this process takes more planning when RPA replaces the tasks of lower-skilled workers. “That’s a more challenging situation,” Burrows said.
RPA's personnel benefits won’t always be seen immediately, as it may take waiting for the wave of retirements in the next 10 to 15 years. Implementing RPA to replace 100 human hours of manual tasks, for example, could help attract and recruit incoming young IT talent capable of managing a suite of bots and monitoring performance — a more attractive job posting than continuing to follow manuals and repetitive work.
Ultimately, Burrows said, agencies should think about hiring employees for different capabilities. “We shouldn’t be hiring for positions that can be automated,” he said. “That becomes a dead-end job. So that’s one thing to keep in mind — we should think about automation first.”
Perhaps investing in RPA begins with a change of mindset within the organization. Even with the 10 bots, Burrows has noticed a change. “People started thinking in terms of how can their jobs be automated,” he said, “and it just sweeps across the organization.” People are realizing when tasks currently being done manually should be automated.
Future of RPA and AI
With RPA on the rise, Burrows doesn’t think agencies are quite there yet in terms of readiness to invest in AI. “Many agencies are in the pilot phase,” he said, “I think we need a number of home runs.”
“In our experience so far, some processes of automation would eliminate thousands of hours [of work],” Burrows said, such as a process that would eliminate 7,000 hours a month of data entry. As soon as agencies start automating processes with that sort of return of investment benefit, that’s when investments and bigger commitments in such technologies will take off, Burrows said.
In turn, RPA can help with data quality and cleansing — manual work that takes time, but is crucial for implementing proper AI.