A General Services Administration official said (GSA) plans to onboard 60 additional federal agencies to the agency’s USAi artificial intelligence (AI) evaluation platform by the end of 2026, significantly expanding a governmentwide effort to help agencies test, evaluate, and deploy AI tools in a secure environment.
Speaking on June 11 at the Government Service Delivery conference in Washington, D.C., GSA Deputy Administrator Michael Lynch said more than 25 federal agencies are already using the USAi platform.
USAi, which GSA unveiled last year, provides agencies with a shared environment to evaluate generative AI tools without building and maintaining separate testing platforms. The platform enables agencies to assess AI technologies at scale while reducing duplication of infrastructure and resources.
While agencies currently use USAi at no cost, GSA is preparing to transition the platform to a cost-recovery model beginning in fiscal year (FY) 2027, according to GSA’s budget document.
Lynch said the platform’s growth has been fueled by agencies’ need for a secure environment to test AI technologies before moving them into production environments.
“People need a safe, secure sandbox to try out these projects, and again, you’re never going to get it to scale if you can’t effectively determine safety and get a look at it as a tool to put in place,” he said.
The USAi environment allows agencies to evaluate AI models and applications in a GSA-managed setting, assess security and performance, and conduct risk reviews before deploying technologies more broadly.
Lynch said GSA designed the platform to address a common challenge facing emerging technology initiatives: pilot projects that fail to progress beyond initial testing.
“It’s easy to just put stuff on the back burner and pilots just kind of go nowhere,” Lynch said.