The US government knows how to improve processes. It also knows how to reorganize programs and how to modernize systems. But with today’s rapid pace of technological breakthroughs, federal leaders run the risk of using AI to make yesterday’s operating model more efficient instead of designing a new, improved, AI-enabled model for tomorrow.
McKinsey’s Rewiring the public sector framework argues that organizations capture AI’s full value by redesigning how work gets done, not simply by deploying new technology. For the federal government, where I served for more than 11 years at the Office of Management and Budget (OMB), the opportunity AI presents isn’t merely to automate existing processes, reorganize agencies, or modernize systems. It’s to rethink—and improve—how government realizes its core mission of delivering to the American people.
For decades, federal transformations have focused on agencies, programs, and technology systems. Success has often been defined by launching a new platform, consolidating an organization, or improving a single process. These are worthwhile achievements, but they do not advance the mission.
If tomorrow’s government is going to organize itself around delivering outcomes, agency leaders could shift their focus from improving individual organizations or programs to redesigning the end-to-end delivery of a mission.
Government’s unique challenges
Making the shift from isolated pilot projects to end-to-end transformation is an especially difficult one for the US government. Unlike the private sector, federal agencies rarely have complete freedom to redesign how they operate. They must align with government-wide policies established or overseen by entities such as the White House, OMB, and the US Government Accountability Office (GAO), even as those policies continue to evolve alongside the technology itself, albeit at a slower and more deliberate pace.
I remember this dynamic well from the early days of the pandemic. Agencies were standing up new programs at extraordinary speed, sometimes serving populations they hadn’t served before. OMB leaders, including me, couldn’t possibly anticipate every question about implementation or issue related guidance before other agencies had to act. So, we worked together in real time. Agencies moved within the guardrails we established at OMB, and as their experience surfaced new questions, the policies evolved alongside implementation. The best path forward wasn’t to wait for perfect guidance; it was to learn by doing and work together.
The task for today’s federal leaders is the same. They cannot wait for every policy question to be answered before beginning to rewire how they deliver their missions. They should move thoughtfully within existing guardrails and authorities, demonstrate to others what’s working, and help inform the next generation of government-wide policy.
Start by stepping back
Rewiring the public sector means stepping back and asking: If we were designing this mission from scratch today—with AI tools, modern data capabilities, and today’s workforce—how would we do so to deliver the best outcome? That’s the difference between automating yesterday’s government and building the government of tomorrow.
Take veterans’ benefits. An individual veteran doesn’t experience a program office or an organizational chart. They experience a journey: applying for benefits, verifying their eligibility, providing documentation, asking questions, appealing decisions when necessary, and ultimately receiving the benefits they have earned. That journey spans multiple organizations, systems, data sets, and personnel. If the government simply inserts AI into each step, it could improve specific parts of that journey. But the veteran’s experience or outcomes won’t fundamentally change.
Having spent years leading federal financial management, I’ve seen firsthand how good the federal government has become at building systems of accountability. Indeed, financial management is an area where the federal government has made substantial improvements. We established common standards, strengthened internal controls, improved auditability, and built a culture where compliance matters. But AI requires the government to build upon those achievements rather than focus on compliance—as well as security, privacy, and responsible governance—as the destination, when it should be the starting point. We celebrate completing the review, satisfying the requirement, or checking the box.
Meanwhile, we spend less time asking the most important question: Did we improve the delivery of the mission? The mission is to disburse veterans’ benefits faster; to help disaster survivors recover sooner; to get taxpayers answers more quickly; and to empower public servants to exercise their judgment more often.
Evolve the entire operating system
To truly transform government outcomes, our Rewiring framework offers important lessons about the levers that enable success. Federal agencies can start by defining the value the transformation is intended to create—that is, defining the mission outcome—which becomes the North Star for the road map, the metrics, and the decisions that leaders make along the way. Then agencies align technology, data, governance, workforce, and funding accordingly to deliver that outcome—not optimize each element independently. Ultimately, that’s the real opportunity AI creates: evolving the government’s operating system so it delivers both accountability and better outcomes for the American people.
I’ve seen this with improper payments. Many agencies got much better at identifying risks, understanding root causes, and reporting where improper payments were occurring. That was important progress, and the teams responsible for that work were doing the jobs that Congress and OMB had asked them to do. Still, the people identifying the risk and the people with the ability to redesign the program weren’t always handing things off or solving the problem together, and the program did not fundamentally improve. That’s what it looks like when we optimize functions instead of outcomes.
A beautifully documented root cause doesn’t prevent the next improper payment. That only happens when the insight travels back into the program and changes how the program operates.
Own the outcome
Rewiring the US public sector requires a different model of leadership to deliver mission outcomes that depend on technology, funding, policy, data, procurement, legal authorities, and cross-agency collaboration. Historically, government has distributed ownership in a way that makes an end-to-end transformation nearly impossible. The CFO owns financial management. The chief information officer (CIO) owns technology. Acquisition leaders own procurement. Program leaders own operations. Each improves the piece they control, but no one is accountable for the end-to-end redesign or the outcome.
Therefore, the most important question a federal leader should ask is, “Who else owns this outcome with me?” rather than “What part of this process do I own?” That’s the leadership model AI demands: co-ownership of mission outcomes.
How to get started
Over the next 12 months, every deputy secretary and assistant secretary can commit to one bold experiment to improve a mission outcome by following the steps below.
- At your next leadership meeting, don’t ask each executive for a functional update; ask them which mission they are co-owning.
- Choose a mission that matters deeply to the public—not a technology project, system modernization, or organizational initiative.
- Define vision and success by outcomes: faster service, greater accuracy, better employee experience, and increased public trust.
- Bring together every executive who owns a part of delivering that mission: the CFO, CIO, acquisition leaders, privacy and legal leaders, HR, and program leadership.
- Give that leadership team one shared responsibility: improve the mission outcome together.
- Don’t wait for perfect guidance before getting started. Work within existing policy and governance, document the lessons learned, and help inform the next generation of government-wide standards.
- Redesign the end-to-end mission and its delivery processes with citizens’ needs and perspectives in mind.
- Challenge every handoff, question every approval, eliminate unnecessary work, share data responsibly, and apply AI where it improves judgment and reduces friction.
- Learn from your experiment and build the foundation to evolve the operating system across the agency.
The leaders who define this next chapter of US government won’t be remembered for deploying the largest number of AI use cases. They’ll be remembered for having the courage to organize around missions, redesign how those missions are delivered, and deliver better outcomes for the American people.