GSA signs its first direct small-business OneGov deal, offering $1 AI pilots

Federal agencies can test an AI platform for $1 before buying discounted software licenses under a new purchasing agreement. The General Services Administration has reached its first direct OneGov deal with an 8(a)-certified small business, (US)AI, bringing the Maryland company’s tools for developing and operating AI applications into the government’s centralized technology buying initiative.

The supplier, based in North Bethesda, Maryland, offers an Intelligent Computing Platform with FedRAMP High authorization. That authorization allows it to support federal cloud workloads requiring a high level of security protection. The agreement gives agencies access to the platform, rather than tying the purchase to one named AI model.

OneGov is GSA’s strategy for consolidating federal technology purchases, standardizing contract terms and reducing duplicative acquisition work. Its previous agreements have largely involved major technology companies, including Microsoft, Google and OpenAI. This deal adds a small-business supplier to that direct purchasing route; it is not GSA’s first direct agreement with an AI company.

A two-week trial, then discounted licenses

Agencies can run designated two-week proof-of-value pilots for $1. These are short trials intended to assess the platform’s usefulness. The software offer is separate: eligible licenses receive discounts of up to 30%.

The offer runs through September 30, 2028, according to GSA’s agreement listing. Nextgov reported the deal on October 23, citing a factsheet on GSA’s OneGov website that sets out the license and pilot terms.

The platform is intended to help agencies move from AI pilots to operational use by securely developing, deploying, managing, and governing AI-enabled applications, agents, workflows, cybersecurity capabilities, and mission processes.

GSA OneGov factsheet, quoted by Nextgov

Tools to run applications, not just try models

GSA describes the platform as covering several stages of an agency’s AI work: developing applications, deploying them, managing their operation and governing their use. Agencies can use approved AI models of their choice.

The listed capabilities extend beyond generating answers or drafting text. GSA describes tools for building software, handling data, detecting fraud and defending systems. The agreement also covers infrastructure management and governance, risk and compliance capabilities. Those are descriptions of the platform’s scope, not measured results from agency deployments.

The names also require a distinction: (US)AI, the company receiving this agreement, is not USAi, GSA’s separate governmentwide AI evaluation service. The latter gives federal employees a shared environment for evaluating and experimenting with models from multiple providers.