Policy & Practice | Fall 2025

or more than a century, pilots have relied on increasingly advanced software to manage the mechanical aspects of flying The Next Frontier of Efficiency for Human Services with Agentic AI Increasing Productivity and Improving Client Experience F By Jamia McDonald, Hari Murthy, Naman Chaurasia, Reid Exley, and Tiffany Dovey Fishman

English to a simple user interface and, within seconds, receive a summarized answer, with exact citations from written policy and links to source materials. Similarly, customers applying for and managing benefits programs can use service assistant chatbots to get navigational support and answers 24/7 in the language of their choice, measurably improving customer satisfaction. While these types of interactions are valuable in improving productivity, workforce experi ence, and citizen engagement, they are largely transactional and limited in scope. Recent advances in AI technology widen the aperture of what is possible. Most work is not a single, simple task. Real change comes when AI augments whole work flows. Agentic AI has emerged as an important new tool in the AI toolkit to do just that. AI agents are reasoning engines that can under stand context, plan workflows, connect to external tools and data, and execute actions to achieve a defined goal. Agentic AI expands the transformative power of AI and GenAI by aug menting not only knowledge but also execution, enabling agencies to execute their missions with greater speed, efficiency, and precision. While a human gives an AI agent a goal or outcome to achieve, it is the agent that deter mines how to fulfill the goal without the constant need for highly specialized prompts or continual human input. The level of agent initiative can be set to different levels of autonomy relative to the extent of human oversight a particular process demands, ranging from more traditional support

a plane. From maintaining level flight and adjusting engine settings, to navigating way points, these systems are integral to modern aviation. Once an aircraft reaches cruising altitude, the pilot delegates routine control to these sophisticated systems, which execute thou sands of micro-adjustments per second, while the pilot charts the aircraft’s route, monitors weather conditions, and communicates with air traffic control. This shift allows pilots to focus on orches trating the flight, intervening when necessary. Human services stands on a similar frontier. While human services agencies nationwide are already using artificial intelligence (AI) tools to ease workloads for overburdened staff and enhance customer experience for constituents, the emergence of AI agents and multi-agent AI systems mark a significant advancement in what’s possible through human–AI partnership. AGENTIC, “MULTI-AGENT” SOLUTIONS SHIFT THE FOCUS OF AI STRATEGY FROM TOOLS TO NEW WAYS OF WORKING Over the past few years, a growing number of human services agencies have adopted large language models (LLMs) and GenAI-powered tools. These tools serve as helpful assistants, interacting intuitively with humans, synthesizing complex information, and generating content. A caseworker can submit a policy question in plain

Policy & Practice Fall 2025 22