Code For America partners with Anthropic on AI tools for SNAP caseworkers

Code for America, the civic tech nonprofit, is partnering with Anthropic to build new AI-powered tools aimed at helping government caseworkers navigate increasingly complex public benefits policies.

Announced Thursday at Code for America’s annual summit in Chicago, the partnership will initially focus on the Supplemental Nutrition Assistance Program, through a new tool called the SNAP Policy Navigator. Using Anthropic’s AI assistant, Claude, the tool aims to give caseworkers real-time access to federal, state and county SNAP guidance while helping reduce administrative burdens tied to benefits eligibility and policy interpretation.

Beyond the initial pilot, the organizations said, they plan to integrate Claude into additional caseworker tasks, including reviewing eligibility documents, answering policy questions and drafting plain-language communications for benefit recipients.

“This partnership with Anthropic underscores our shared belief that responsible AI is a transformative tool, capable of easing caseworker burden, streamlining processing, and, ultimately, delivering benefits more quickly and accurately,” Amanda Renteria, Code for America’s chief executive, said in the announcement.

The launch comes as states race to upgrade their IT systems and data infrastructure to reduce payment error rates and comply with new program changes by next year. Under H.R.1, the budget reconciliation legislation Congress passed last year, states are required to share in the cost of administering SNAP based on payment error rates, how accurately they judge eligibility and benefit amounts. High error rates, above 6%, trigger corrective actions and potential financial penalties.

State IT systems that process benefit cases, like SNAP or Medicaid, are often programmed with federal rules in mind, prompting caseworkers to follow mandatory steps for certification, verification and budgeting. These systems, known as integrated eligibility and enrollment, or IEE, systems, translate federal, state and local policies into code.

A February report from the Digital Benefits Network, part of the Beeck Center for Social Impact and Innovation at Georgetown University, identified several common challenges with state IEE systems, including technological complexity, aging infrastructure and cross-agency coordination hurdles. Rather than building systems program by program, the report encouraged states to coordinate benefit laws across agencies, create legislative timelines with realistic development cycles and pursue more flexible federal and state funding models.

The network also found that artificial intelligence tools can expedite the translation of policies into software code for implementation in public benefits eligibility and enrollment systems under a “rules as code” scheme. By allowing computer systems to interpret and apply rules directly, the network found that AI tools can increase transparency, reduce ambiguity and enable automated compliance for governments.

According to the announcement, Code for America’s SNAP Policy Navigator will build on Anthropic’s Model Context Protocol, an open standard designed to ensure responses are grounded in verified policy information, rather than generalized AI outputs.

“Technology is most impactful when it reaches the people with the deepest needs. SNAP caseworkers carry an enormous load, interpreting complex rules under tight timelines for the families who depend on the safety net,” Elizabeth Kelly, Anthropic’s head of beneficial deployments, said in the press release.

A recent report published by Code for America assessing government’s use of AI found that states are rapidly transitioning away from experimentation toward broader implementations across operations, including for workforce services, unemployment systems and benefits administration. The Michigan Department of Health and Human Services last March deployed an AI tool to increase the number of cases employees can accurately review. And late last year, Maryland secured grants for AI projects to help connect residents to SNAP and Medicaid services.