Code for America, a nonprofit dedicated to civic technology, has teamed up with Anthropic to create innovative AI-driven tools designed to assist government caseworkers in navigating the increasingly complicated landscape of public benefits policies.
Announced on Thursday during Code for America’s annual summit in Chicago, this collaboration will initially concentrate on the Supplemental Nutrition Assistance Program (SNAP) through a new tool named the SNAP Policy Navigator. By leveraging Anthropic’s AI assistant, Claude, this tool aims to provide caseworkers with immediate access to federal, state, and county SNAP guidance, while also alleviating the administrative burdens associated with determining eligibility and interpreting policies.
Looking beyond the initial implementation, the organizations plan to expand Claude’s capabilities to enhance various caseworker functions, such as reviewing eligibility documentation, responding to policy inquiries, and crafting clear communications for benefit recipients.
“Our collaboration with Anthropic highlights our mutual belief that responsible AI can transform government operations, making it easier for caseworkers to perform their duties, streamlining processes, and ultimately delivering benefits more quickly and accurately,” stated Amanda Renteria, CEO of Code for America, in the announcement.
This launch arrives at a critical time as states work to upgrade their IT systems and data frameworks, aiming to reduce payment error rates and comply with upcoming program changes set for next year. Under H.R.1, the budget reconciliation bill passed by Congress the previous year, states are obligated to share in the costs of administering SNAP based on their payment error rates and accuracy in assessing eligibility and benefits. Should error rates exceed 6%, corrective measures and potential financial penalties may be enforced.
State IT systems responsible for processing benefit cases, such as SNAP or Medicaid, are typically designed with federal regulations in mind, compelling caseworkers to follow structured protocols for certification, verification, and budgeting. These systems, known as integrated eligibility and enrollment (IEE), translate federal, state, and local policies into coded instructions.
A report released in February by the Digital Benefits Network, affiliated with the Beeck Center for Social Impact and Innovation at Georgetown University, identified common challenges faced by state IEE systems, including technological complexities, aging infrastructures, and issues with cross-agency coordination. Rather than developing systems on a program-by-program basis, the report urged states to align benefit regulations across agencies, establish legislative timelines that incorporate realistic development cycles, and seek more adaptable funding models at both federal and state levels.
The network also discovered that AI tools can streamline the process of transforming policies into software code for public benefits eligibility and enrollment systems using a “rules as code” approach. By enabling computer systems to interpret and enforce rules directly, AI can enhance transparency, minimize ambiguity, and facilitate automated compliance for governmental agencies.
According to the announcement, Code for America’s SNAP Policy Navigator will build upon Anthropic’s Model Context Protocol, an open standard that ensures responses are anchored in verified policy information, rather than generalized outputs generated by AI.
“Technology achieves its highest potential when it serves those with the greatest needs. SNAP caseworkers bear a tremendous responsibility, interpreting complex regulations under tight deadlines for families reliant on the safety net,” remarked Elizabeth Kelly, Anthropic’s head of beneficial deployments, in the press release.
A recent report by Code for America assessing the use of AI in government revealed that states are swiftly moving from testing phases toward broader implementations across various operations, including workforce services, unemployment systems, and benefits administration. For instance, the Michigan Department of Health and Human Services deployed an AI tool in March to enhance the number of cases staff can review accurately. Furthermore, late last year, Maryland secured grants for AI initiatives aimed at improving access to SNAP and Medicaid services.
