GSA steps from automation to AI
Connecting state and local government leaders
Now that the General Services Administration has experience with robotic process automation, it’s looking to artificial intelligence to inject more efficiency into complex processes like contracting.
Now that the General Services Administration has experience with robotic process automation, it's looking to artificial intelligence to inject more efficiency into complex processes like contracting.
While RPA bots can be implemented relatively quickly by automating established processes, AI takes more time and expertise because it forges new paths by finding more-efficient ways to work with both processes and data, said Michelle McNellis, GSA’s director of acquisitions at the Office of Products and Programs, at a June 3 Defense One virtual event on automation in acquisition.
GSA uses an AI-based bot to track, find and change Section 508 disability clauses in contracts to ensure compliance, and that work is more advanced than just rote processing, according to Omid Ghaffari-Tabrizi, director of GSA’s Acquisitions Centers of Excellence. That review takes "some degree of intelligence," he said, but the output is always reviewed by humans to ensure accuracy.
GSA has been at the forefront of implanting bots, with dozens performing repetitive electronic processes. One bot sends notifications about invoices and another automates the work associated with processing offers under the Federal Acquisition Service's Multiple Award Schedules.
It's also using bots for its FASt Lane, eOffer and eMod processes, Ghaffari-Tabrizi said. FASt Lane is the agency's program to accelerate how IT contractors get new products onto its buying schedules, while eOffer/eMod allow vendors to submit modifications to their contracts.
Other federal agencies looking to harness similar RPA capabilities, said McNellis, should move deliberately, getting input from all agency operations, including finance, IT, acquisition and management. Legal issues and IT capabilities need to be addressed before moving ahead with either AI or RPA efforts, she said.
This article was first posted to FCW, a sibling site to GCN.