让代理人工智能为政府服务：一个准备框架（英）

Making Agentic AIWork for Government:A Readiness Framework I N S I G H TR E P O R TA P R I L2 0 2 6 Contents Foreword3 Executive summary5 1The agentic opportunity6 1.1From process digitization to outcome orchestration71.2Why agentic AI matters for governments91.3The tactical challenge: where to begin9 2An innovative government readiness framework for agentic AI10 2.1A function-based assessment lens112.2The assessment of potential against complexity132.3A topography of government readiness152.4From global topography to regional roadmap22 3Learning from successful deployments26 Conclusion30 Appendices 31 A1Methodology31A2Comprehensive breakdowns of agentic AI assessment33scores for all 70 core government functions Contributors37 Endnotes39 Disclaimer This document is published by theWorld Economic Forum as a contributionto a project, insight area or interaction.The findings, interpretations andconclusions expressed herein are a resultof a collaborative process facilitated andendorsed by the World Economic Forumbut whose results do not necessarilyrepresent the views of the World EconomicForum, nor the entirety of its Members,Partners or other stakeholders. ©2026 World Economic Forum. All rightsreserved. No part of this publication maybe reproduced or transmitted in any formor by any means, including photocopyingand recording, or by any informationstorage and retrieval system. Foreword Mohamed Bin TaliahAssistant Minister, Cabinet Affairs forGovernment Experience ExchangeAffairs, Ministry of Cabinet Affairs,United Arab Emirates Fabian MehringState Minister for Digital Affairsand Chief Information Officer,Free State of Bavaria Martina Klement Permanent Secretary for DigitalTransformation and AdministrativeModernization; Chief Digital Officer,State of Berlin Agentic artificial intelligence (AI) accelerates thisshift – and brings a new quality to it. Where earlierwaves of digitization moved paper processes on toscreens, agentic AI systems can plan, decide andact across entire workflows, coordinating steps thatpreviously required manual handoffs, interpretingcontext and delivering outcomes rather thanoutputs. For governments already under pressurefrom multiple directions, this is not a theoreticalprospect. It is a practical lever. Over the past decade, governments around theworld have invested heavily in digital transformation –and it has paid off. Public services have movedonline, data-driven decision-making has gainedreal traction across many administrations, and thegovtech market has matured from a niche into aglobal force. Technology is no longer a supportfunction for the state. It is becoming central to howgovernments operate, deliver services and earnthe trust of their citizens. The drivers for this development vary. In somecountries, citizen expectations have outpacedwhat traditional administration can deliver,creating urgency to rethink how services aredesigned and provided. In others, demographicdecline means that doing more with fewer peopleis not a choice but a reality. For some, digitalinfrastructure is the backbone of economiccompetitiveness; for others, it is the path torebuilding public trust after years of institutionalunderperformance. The starting points may bedifferent, but the direction is the same: technologyis moving to the core of government. The opportunities are vast. Yet, opportunity has neverbeen the bottleneck for government technology.The real challenge is strategic: understanding whichoperations benefit most, sequencing adoption sothat early efforts build capability rather than drainit and maintaining public accountability throughout.This demands clarity, not haste. Governments that approach agentic AI withdiscipline and strategic intent will not only improvetheir own operations. They will shape the normsand expectations for how this technology isgoverned in the public interest. That makes thisnot just an opportunity, but a shared responsibility. Foreword Ammar AlkassarMember, Board, GovTechDeutschland Roshan Soorunsingh GyaChief Executive Officer, NorthernEurope, Capgemini Stephan MergenthalerManaging Director; Chief TechnologyOfficer, World Economic Forum Governments around the world are exploring thepotential of artificial intelligence (AI), and agentic AI inparticular is attracting growing attention as a way tomove beyond isolated tools towards systems thatcan coordinate, decide and act across complexworkflows. Yet for most, this remains difficult totranslate into practice. What is often missing is acredible basis for deciding where to begin, whatto prioritize and how to move from experimentationto operational reality. That is what this report sets out to do. The strategyand action of public administration are shaped byorganizational boundaries – ministry by ministry,department by department. Agentic AI does notwork that way. It operates across workflowsthat span organizational lines: “cybersecuritymonitoring”, “document life cycle management”,“eli