7. Leadership
The skills and behaviours of outcome-driven government leaders
who will build agentic states.
How It (Doesn’t) Work Today
For three decades, digital transformation has reshaped leadership expectations and operational norms in high-performing enterprises. But a governance gap persists: Many governments are still a generation behind with their leadership practices. Civil service and political leadership alike struggle to adapt to the digital world around them.
Two intertwined challenges stand out in the current situation:
First, political and administrative leaders continue to treat technology and digital systems as siloed support functions. Data and even AI are seen as technical or back-office concerns, disconnected from the core public service mission and strategic objectives. In some cases, digital illiteracy is not just tolerated but effectively rewarded, both in senior civil service appointments and at the ballot box.
Second, governments have been slow to adopt broader leadership and management best practices that have become standard in high-performing organisations, a trend often highlighted in analyses by organisations like the OECD and public sector consultancies. These best practices include rigorous data-informed decision-making, disciplined performance management based on meaningful metrics for both process and outcome, and agile methodologies focused on iterative value delivery, and a relentless focus on clearly defined outcomes.
This foundational leadership and management deficit contributes to a misalignment between technological capabilities and government priorities, and perpetuates organisational dysfunctionalities.
A Vision for Outcome-Driven, Agentic-Powered Leadership
The surge in generative and agentic AI means that expectations of leadership change and become more demanding: Enterprise leaders are increasingly expected to navigate AI's potential to reshape entire business models, manage hybrid human-AI teams at scale, govern systems with emergent properties, and drive innovation at an accelerated pace. We should expect no less of government leaders. This fundamental shift does not exempt government leaders.
Leaders understand that agentic AI reframes how the state must conceive of itself. As systems that reason, act, and learn autonomously begin to take on core functions, public sector leaders must confront foundational choices about where the state should act, and how.
In this paradigm, leadership is no longer just about delivery. It is about designing the posture of the state. In this regards, leaders must be conscious of the new key questions they need to address:
Investing or divesting: Scarce resources, including compute, talent, and institutional focus, must be intentionally directed toward public goals where automation can deliver the greatest value. Not all functions should be scaled equally; prioritisation becomes a strategic act.
Regulator or deregulator: Rules are no longer only enforced, they are encoded. Leaders must determine how policies are embedded into systems, and under what conditions those systems can adapt dynamically based on performance, confidence levels, or local variation
Central orchestrator or federated enabler: Leaders must decide which powers remain centrally coordinated and where to enable autonomy — whether through local institutions, regulated non-state actors, or even citizen-controlled agents. These structural decisions will define how flexible, resilient, and trusted the Agentic State becomes.
These are not merely technical or organisational decisions. They are acts of public judgment and political design. The challenge is how to lead like an Agentic State: to govern with clarity, constraint, and intentional absence. This means designing not only for effectiveness, but for discretion, knowing when not to act, when to yield, and how to create space for autonomy without losing legitimacy.
In this era, key strategic imperatives for government leaders will include:
Defining public value and strategic intent: Leaders should define clear societal outcomes and public value in a way that can be handed over to AI agents, while also recognising when automation has failed. This process begins with a personal commitment to learning about AI.
Ensuring ethical, democratic, and accountable governance: Leaders should establish and enforce robust ethical frameworks for all public sector behaviour, both human and AI, rooted in democratic values, human rights, and public trust. They mandate transparency, fairness, human oversight, and clear accountability mechanisms for all AI deployments, while also ‘working in the open’ to build transparency and public trust.
Leading people and organisations through transformation: The human dimension remains ever more critical to leadership. It requires all the leadership traits that bring critical judgment, creativity, empathy, and risk-taking to the forefront in human teams.
Additionally, effective leadership of agentic AI in the public sector will recognise that AI and humans require complementary but distinct leadership skills:
Agentic AI performs optimally when directed by clear, measurable outcome metrics and unambiguous strategic intent provided by leadership. This is because it has the capacity for autonomous execution and optimisation.
Humans, especially when dealing with complicated change and working with AI, need leaders who can provide emotional intelligence, encourage open and clear communication, and leave room for experimentation, learning from mistakes, and quick iterations within the strategy.
Key Questions
What requires political leadership? And where can senior civil servants lead the way?
Which aspects of agentic transformation require elected leaders to set direction and provide democratic legitimacy? Where can senior public executives lead change independently, without waiting for political mandates?
How do we grow a generation of AI-native public leaders? What kind of training, mentorship, and career pathways will equip tomorrow’s leaders to steer agentic government, blending technological fluency with public purpose, systems thinking, and ethical reasoning?
How do leaders define strategic intent in a way that AI can act on and humans can rally around? What does it mean to set a ‘machine-readable mission’? How do leaders craft goals that are precise enough to guide AI agents, yet inspiring and flexible enough to motivate human teams?
Do we need a new leadership archetype for the agentic AI era? Is it time to replace, or evolve, the role of the government CIO? Should governments appoint Chief AI Officers with the authority to integrate AI across operations, ethics, and strategy?