GSA’s USAi: A Short-Term AI Testing Solution for Federal Agencies
The General Services Administration (GSA) recently launched USAi, a governmentwide AI testing tool aimed at helping federal agencies evaluate various artificial intelligence models. Zach Whitman, GSA’s Chief Data Officer and AI Officer, emphasized that USAi is intended as an “immediate term” support service rather than a permanent solution.
Whitman explained that the GSA does not plan to remain the long-term provider of this AI testing infrastructure. Instead, the agency expects market forces to eventually offer more tailored AI solutions to government agencies. For now, USAi serves as a low-barrier platform to help agencies quickly test and access multiple AI models without facing acquisition or technical hurdles.
Addressing Dynamic AI Models and Agency Needs
The AI landscape is highly dynamic, with new models continuously emerging and improving. Whitman noted that “every week you have a new model that is potentially outperforming another,” which makes it challenging for agencies to keep up. USAi’s shared service platform allows agencies to experiment with AI models from major providers like Anthropic, OpenAI, Google, and Meta—all available at launch. More models are expected to join after passing GSA’s safety reviews.
However, Whitman also acknowledged limitations. As a shared service, USAi offers general-purpose AI models, which means agencies with specialized or niche use cases might find it insufficient. The platform does not provide full control or autonomy over which models agencies can use. Still, the focus remains on helping agencies ramp up AI capabilities at little to no cost before transitioning to market-based solutions.
Path from Testing to Procurement
After testing AI models on USAi, agencies can choose to procure these solutions through the standard federal marketplace channels. Some agencies may continue using USAi as market conditions evolve, but the platform primarily serves as a testing ground. This approach prevents agencies from getting locked into specific AI products prematurely and encourages informed procurement decisions.
Building on Internal AI Tools
USAi expands on GSA’s internal AI tool, GSAi, which was developed for agency-specific use earlier this year. Whitman highlighted that adapting GSAi’s infrastructure into a broader framework for other agencies was a relatively straightforward step. This shared foundation helps federal entities adopt AI solutions more efficiently, while the GSA continues to refine the platform based on user feedback and evolving needs.
Collaborations and Future Priorities
As part of its OneGov strategy, GSA has partnered with AI providers like OpenAI, Anthropic, and Box to offer discounted products to federal agencies. The agency is also considering streamlining AI technology reviews within the federal procurement process to accelerate adoption.
For government professionals looking to expand their AI knowledge and skills, training resources are available. Explore latest AI courses designed for public sector applications, or browse AI training by job role to find relevant programs.
Summary
- USAi is a temporary, shared platform enabling federal agencies to test AI models easily and affordably.
- The GSA expects market solutions to eventually replace USAi for long-term AI needs.
- Agencies can trial models from major AI providers and decide on procurement through federal marketplaces.
- USAi builds on GSA’s internal AI tools, adapted for governmentwide use.
- Collaborations with AI companies aim to reduce costs and simplify acquisition processes.
USAi offers a practical entry point for federal agencies to explore AI capabilities while preserving flexibility for future procurement decisions.
Your membership also unlocks:
