Public-sector organizations face growing pressure to improve outcomes for residents while making taxpayer funds go further. When deployed responsibly, artificial intelligence (AI) can help reduce wait times, simplify routine interactions, and free public servants to focus on more complex work. However, realizing that potential requires more than isolated pilots: it calls for redesigning processes, ways of working, and operating models so AI becomes part of everyday service delivery.
In this article Hrishika Vuppala, Tim Fountaine, Tim Ward, and Tony D’Emidio set out a practical, four-part approach that governments can use to move beyond experimentation and scale AI responsibly, delivering measurable gains in efficiency, service quality, and resident experience.
Why the moment is unique — and what stands in the way
Governments worldwide face fiscal constraints, skilled-workforce shortages, rising citizen expectations, and declining public trust. Many AI efforts remain stuck in “pilot purgatory” because organizations struggle with data access, workflow integration, model risk, and high ongoing operating costs.
The authors argue that capturing AI’s value is rarely about tools alone. It requires rewiring operations by reimagining workflows end to end, adopting new ways of working, and engaging the workforce to drive adoption. Even advanced tools will fall short if structural issues — outdated processes, fragmented decision-making, and misaligned workforce models — are not addressed.
A simultaneous, four-part approach
The recommended approach comprises four moves to be undertaken in parallel:
-
Lead with mission outcomes in strategy: prioritize outcomes beyond cost-cutting, set bold ambitions, start from a clean sheet when needed, and design for continuous evolution.
-
Reimagine workflows end to end: enable teams to execute and innovate safely, adopt agile, product-based operating models, deploy scalable technology to reengineer workflows, build a robust data foundation, and maintain financial discipline.
-
Build the operating system around the technology: emphasize organizational change management, human-centered design, real-time data-driven and ethical AI governance, and evolve procurement and contracting for scalable outcomes.
-
Keep humans in the loop for consequential actions: define human sign-offs according to consequence rather than technology category, making human oversight a political and operational safety net.
Public-sector examples showing measurable impact
The article presents several case studies where governments delivered value by pairing AI with process redesign and governance.
-
A national public customer-service organization used a domain-led approach and dedicated cross-functional teams, strengthened data and technology foundations, and embedded risk, assurance, and ethics controls. Over two years the program improved developer productivity by roughly 30 percent and identified more than $1 billion in inaccurate or fraudulent payments, while reshaping nationwide service delivery.
-
A professional licensing department redesigned a physician-licensing process and incorporated agentic AI. Process mapping was prioritized; rules were codified into a structured digital rulebook and AI agents were used to extract application data, validate credentials, flag missing or inconsistent information, and produce structured case summaries with explainable recommendations. Staff kept decision authority for complex judgments. As a result, potential send-backs fell by 40–45 percent, review rounds were halved, and review timelines decreased by about 50 percent to 12 days.
-
A national unemployment agency that handled roughly two million benefit applicants annually redesigned its end-to-end benefits journey. The program set a clear North Star to become customer-centric and a learning organization, integrated AI (mobile-friendly digital channels, processing bots, and personalized communications), measured outcomes across the process, and created clear handover points to ensure operational accountability.
Other examples include a healthcare agency deploying an enterprise-wide dashboard and training over 1,000 leaders on performance management, and a public-sector financial institution that migrated roughly half its applications to public cloud while modernizing infrastructure and adopting low/no-code platforms to speed development and improve resilience.
Strategy, capability building and risk management
Effective rewiring begins by defining mission-critical goals — for example, improved public health outcomes, shorter response times, or higher operational efficiency — so technology serves clear objectives. Scaling digital solutions requires intentional organizational change management and capabilities such as procurement, training, culture change, and oversight. Risk and ethics must be embedded from ideation through post-deployment monitoring.
Agency leaders should design for risk management from the start rather than retrofitting governance. The authors note troubling statistics: more than 70 percent of federal technology programs remain over budget or behind schedule, and over 80 percent of organizations that deployed AI reported no tangible enterprise impact.
The road ahead for governments
Policy signals and demand for AI procurement are fostering innovation; many governments are pursuing national or regional AI initiatives (examples cited include Singapore, the United Arab Emirates, the United Kingdom, and U.S. states such as California and Pennsylvania). The number of countries with national data and AI strategies has risen to 95, compared with fewer than 20 in 2020.
Generative and agentic AI could reshape work by enabling partnerships between people, agents, and robots, improving agility, parallel processing, personalization, elasticity of operations, and resilience. But these technical capabilities will only translate into better public services if governments take a holistic approach that prioritizes strategy and people alongside technology.
Practical first steps
To maximize the odds of successful transformation, the authors recommend immediate actions for leaders: define mission-driven goals, map end-to-end processes, build strong data and technology foundations, embed ethics and risk controls early, run targeted lighthouse initiatives to demonstrate value, and adopt agile, product-based operating models.
Conclusion
AI offers a rare opportunity for governments to rewire how they operate and deliver services. The technology can enable substantial productivity and service-quality gains, but only when paired with mission-focused strategy, end-to-end workflow redesign, organizational capability building, and explicit human oversight. Taken together, these elements increase the likelihood that AI deployments will move beyond pilots and produce measurable benefits for residents.
The article is authored by Hrishika Vuppala and Tim Ward (senior partners, McKinsey Southern California), Tim Fountaine (senior partner, McKinsey Sydney), and Tony D’Emidio (partner, McKinsey Washington, DC). The authors thank Ali Ustun, Anne Neville-Bonilla, Deidre Harrison, and Kelly Ungerman for their contributions.
