Agentic artificial intelligence (AI) is driving a fundamental shift in capability, allowing systems to autonomously execute end-to-end, multi-step workflows. This technological progress is poised to transform how governments operate and serve citizens. However, without a strategic, evidence-based grasp of where agentic AI can deliver the greatest public value – balancing high potential with manageable complexity – governments risk investing in the wrong places, undermining confidence in the technology and launching pilots that fail to scale.

This report applies a novel, department‑agnostic perspective on government activities. It provides a clear approach to assess where agentic AI can deliver the greatest public value and what risks must be managed before deployment at scale.

Drawing on a global assessment, it offers initial orientation on starting points for agentic AI as well as corresponding real-world use cases, highlighting current initiatives in the public sector.

The report maps 70 core government workflows against two dimensions – the opportunity for agentic AI to
add public value, and the complexity of deploying it responsibly – showing where governments can act with confidence, where targeted preparation is needed, and where caution is warranted.