Code for America and Anthropic pilot AI tool to help SNAP caseworkers navigate complex benefits rules
Code for America and Anthropic announced Friday a partnership to develop AI tools that help caseworkers administer the Supplemental Nutrition Assistance Program more efficiently. The SNAP Policy Navigator, built on Claude, allows eligibility staff to input policy questions and receive plain-language answers with cited sources.
A caseworker can ask how a client's income change or new federal rule affects their benefits. The tool outputs an answer with suggested next steps. The system provides clarity on policy without making eligibility decisions - those remain with the caseworker.
The tool is built on federal regulations, state manuals, and official policy directives. It uses Anthropic's model context protocol to create a secure connection between data sources and the AI application.
Pressure from recent policy changes
The announcement comes as states and localities scramble to implement rule changes to SNAP that took effect last July. The changes expanded work requirements, shifted administrative costs to states based on payment error rates, and tied benefit calculations to food price changes.
SNAP participation has declined by more than 3 million people across 36 states since January, with further drops expected as new rules fully take hold.
Caseworkers already face overwhelming caseloads, funding uncertainty, and workforce shortages. One former caseworker described the job as "an email inbox that's always full, where each one requires care and attention, but you're constantly getting interrupted as you try to work through the never ending inbox of people to help."
Broader toolkit in development
Beyond the SNAP Policy Navigator, Code for America and Anthropic plan to develop additional Claude-based tools. The suite will help benefit workers answer policy questions, review eligibility documents, and draft communications to recipients.
Michael Lai, who leads state and local AI for Government at Anthropic, said the goal is practical: "AI shouldn't be used for AI's sake. We want it ultimately to be helping in this human way and trying to make benefits administration more efficient, more accurate and more human centered."
Your membership also unlocks:
