News
On May 18, 2026, the U.S. Digital Corps (USDC) released findings that implementing Model Context Protocol (MCP) servers improved AI systems' data retrieval accuracy by 95%, with potential applications to open data, internal analytics, and customer experience. USDC released Improving LLM Access to Federal Open Data: A Pilot Study of MCP Servers for Federal Data Access detailing these results. The report offers both technical and strategic guidance for federal teams building MCP servers and a roadmap for experimentation and scalable adoption across government.
In one of USDC’s first major cross-program collaborative efforts1, Fellows explored using the Model Context Protocol, an emerging open-source protocol for integrating Large Language Models (LLMs) and external resources, to redefine how LLMs interact with federal data. Instead of relying on scraped or static information, MCP enables direct connections to authoritative federal data APIs, unlocking real-time, structured, and verifiable data access.
This work demonstrates what’s possible when USDC Fellows collaborate across agencies to solve real problems. It highlights the power of designing data-forward approaches for AI implementation, and the impact of making federal data truly accessible, not just available. At its core, this is about transforming how government data is used, shifting from static publication to dynamic, AI-enabled access that better serves the public, policymakers, and agencies alike.
Federal employees or contractors with questions may reach out to mcp@gsa.gov.
-
Launched in early 2025, the AI Readiness Collective brings together U.S. Digital Corps Fellows across agencies to tackle a core challenge: making federal data usable, trustworthy, and actionable in an AI-driven world. This report, authored by Bella Mendoza, Mark Aronson, Haley Johnson, Sam Levy, and Mohammad Arifur Rahman, turns that vision into action aligned with the Foundations for Evidence-Based Policymaking Act of 2018 and America’s AI Action Plan (July 2025).
