Policy & Practice | Fall 2025

eligible population. With many state budgets in crisis, the concern is that some states may be unable to cover these additional costs and opt out of providing a SNAP program altogether. Proactive mitigation is critical for the survival of this crucial program that 12.3 percent of the U.S. popula tion relied on in 2024. By leveraging Agentic AI, agencies can ensure that each application is processed with the same high standards, regardless of the complexity or volume of cases. This minimizes the risk of human error, which can lead to incorrect determina tions, delays, and the need for costly rework. Furthermore, consistent appli cation of rules and procedures helps maintain fairness in service delivery, ensuring that all clients receive the

information. Protecting individual data from misuse and unauthorized access is vital to upholding the integrity and con fidentiality of the public services sector. Trust and explainability are also crucial components—agencies must guarantee that AI-generated deci sions are transparent and subject to audit. This involves furnishing stakeholders with clear explanations behind AI outcomes, thereby fostering understanding and confidence in the underlying processes. Agencies must establish clear gov ernance frameworks for AI adoption, including mechanisms for oversight, auditing, and continuous improvement. They must also engage stakeholders, including clients, advocates, and front-line workers in the design and

n Data Infrastructure: Robust data systems are the foundation of effec tive AI. Agencies must invest in data quality, interoperability, and governance. n Capacity Building: Staff at all levels need training and support to understand and work effectively with AI tools. n Continuous Improvement: Technology is not static. Agencies must establish processes for ongoing evaluation, feedback, and iteration. Seizing the Opportunity The challenges facing state public services agencies are real and pressing, but so too are the opportunities. Step changes in AI, particularly the

Agentic AI can significantly enhance the reliability and integrity of the public service system and allow caseworkers to focus on the most complex and sensitive cases, where empathy and compassion are paramount.

benefits they are entitled to in a timely manner. Consequently, Agentic AI can significantly enhance the reliability and integrity of the public services system and allow caseworkers to focus on the most complex and sensitive cases, where empathy and compassion are paramount. The Imperative of Responsible AI As with any technology, the deploy ment of Agentic AI must be guided by principles of responsibility and ethics. Transparency, fairness, accountability, and privacy are not just buzzwords, they are essential safeguards that ensure AI serves the public good. To mitigate the risk of perpetuating systemic inequities, it is essential to train AI systems on diverse datasets. Promoting fairness and justice within AI applications is fundamental for sustaining public trust and providing effective services to all members of society. Safeguarding data privacy is of particular importance, especially when managing sensitive personal

implementation of AI systems. By doing so, they can build trust, mitigate risks, and ensure that technology enhances rather than undermines the human services mission. A Strategic Approach to Technology Investment To fully realize the benefits of Agentic AI and other technological advance ments, agencies must adopt a strategic, ecosystem-wide approach. This means moving beyond one-off automation projects and investing in a cohesive technology infrastructure that supports integration, scalability, and innovation. Key elements of such a strategy include: n Leadership and Vision: Senior leaders must champion technology as a strategic priority and articulate a clear vision for its role in service delivery. n Cross-Functional Collaboration: Information technology, program, and policy teams must work together to align technology solutions with operational needs and policy goals.

emergence of Agentic AI, offer a path forward, enabling agencies to deliver more efficient and responsive services. By embracing a strategic approach to technology investment, grounded in responsible AI principles, agencies can move closer to the goal of one-touch eligibility processing. The integration of AI can sig nificantly free up human resources, particularly caseworkers, who are often burdened with administrative tasks. In the context of OBBBA, this reallocation of resources can be instru mental. Caseworkers, now relieved from routine data processing and eligibility checks, can devote their expertise to ensuring that the most vulnerable cases and clients receive the human touch and attention needed. As agencies implement AI-driven processes, they build systems that are resilient, adaptable, and future-ready, positioning themselves to better meet the demands of OBBBA and beyond. Now is the time for action. The tools are within reach. The vision is clear. What remains is the leadership and commitment to make it a reality.

Fall 2025 Policy & Practice 33