There is a quiet conceptual revolution happening in how the most effective leaders understand their own role. It does not announce itself loudly. It shows up in the questions leaders ask rather than the answers they give, in the design choices they make about how their teams are structured, and in the way they talk about their responsibility to the people who work with them. The revolution is a shift from managing to cultivating — from the industrial-era model of the manager as performance monitor and output controller to what might be called the stewardship model: the leader whose primary responsibility is creating the conditions in which people, and their increasingly capable AI partners, do their best work.
This shift is not primarily a response to AI. It has been building for decades in the evidence on what actually drives sustained high performance in knowledge-work organisations. The research on intrinsic motivation, psychological safety, learning culture, and distributed leadership has consistently pointed toward the same conclusion: the management practices that worked for routine, predictable, output-measurable work have diminishing effectiveness as the work becomes more complex, more contextual, and more dependent on the full cognitive and creative engagement of the people doing it. AI is not causing this shift. It is accelerating it to the point where leaders who have not made it will find themselves managing in ways that are not only ineffective but actively counterproductive.
McKinsey's 2026 research on the state of organisations identifies what it calls the inside-out leadership premium: the performance differential between leaders who cultivate internal clarity, purpose, and conditions for human flourishing and those who rely primarily on external controls, performance metrics, and compliance mechanisms. In AI-augmented organisations, that premium is larger than it has ever been.
Decision Elasticity
One of the most significant leadership challenges created by the Agentic Era is the compression of decision timelines. AI systems can analyse, recommend, and — in increasingly autonomous configurations — execute, at a speed that routinely outpaces the rhythms of human deliberation that most governance frameworks are built around. The response of many organisations has been to try to match that speed: to push decision-making down the hierarchy, to create faster escalation paths, and to streamline the human review that sits between AI recommendation and organisational action.
The stewardship mindset treats this challenge differently. Rather than simply accelerating human decision-making to match AI speed, it asks a prior question: which decisions actually need to be faster, and which need to be better? The two are not the same thing, and confusing them is one of the most common and costly mistakes in AI-era governance design.
BCG's 2026 research identifies what it calls decision elasticity — the capacity of a leadership team to move fluidly between fast, AI-assisted decisions and slow, deeply considered human judgments, calibrating decision speed to decision consequence rather than defaulting to one mode or the other. Leaders with high decision elasticity maintain a clear taxonomy of their decision landscape: which decisions benefit from AI speed and scale, which require human judgment that current AI cannot replicate, and which require both in sequence. This taxonomy is not fixed — it changes as AI capabilities evolve — and maintaining it is itself a leadership practice that the stewardship mindset makes central.
The psychological component of decision elasticity is equally important. Leaders who feel compelled to match AI speed in every domain, or who feel that the slowness of human deliberation is a personal or professional liability, are more likely to make decisions that are fast but poor. The stewardship mindset includes a settled confidence in the value of slow, careful human thinking in the domains where it matters most — a confidence that is not defensive about AI capability but is clear-eyed about what human judgment brings that AI does not.
The Agent Manager
Every leader who manages a team that uses AI capabilities is, in a meaningful sense, already an agent manager — responsible not only for the performance and development of human team members but for the configuration, oversight, and improvement of the AI agents operating within the team's workflow. Most leaders have not yet updated their mental model of the management role to reflect this reality, which is one reason why the returns from AI deployment are so unevenly distributed across organisations.
Effective agent management is not a technical function. It does not require leaders to understand the engineering of AI systems. It requires leaders to apply the same clarity about purpose, quality standards, and feedback mechanisms to AI agents that they apply to human team members — and to maintain a clear understanding of where the AI agent's capability ends and human judgment must begin.
This means developing fluency with the specific failure modes of the AI systems operating in the team's workflow: the contexts in which they consistently underperform, the kinds of outputs that require human review before action, and the escalation conditions that should trigger deeper human scrutiny. It means creating the equivalent of team norms for human-AI collaboration — shared standards for how AI outputs are reviewed, how errors are identified and fed back into the system, and how the team's collective experience with AI tools is captured and used to improve the configuration over time.
The stewardship mindset makes this a natural extension of the leader's existing responsibility for team effectiveness. The agent manager is not a new role. It is the existing role of team leader applied with clarity to a team that includes both human and digital members, and with the intellectual honesty to acknowledge that managing digital teammates requires developing new skills and new habits of attention.
From Performance Management to Stewardship
The conventional performance management model — goal-setting, monitoring, review, reward, and correction — was designed for a world where the primary challenge was ensuring that people with defined tasks performed those tasks to defined standards. In the complex, adaptive, cognitively demanding work environments of the AI era, this model is not so much wrong as insufficient. It measures the wrong things, at the wrong frequency, with the wrong feedback loops.
Self-Determination Theory, developed by Deci and Ryan, identifies three fundamental psychological needs whose fulfilment predicts intrinsic motivation, sustained engagement, and learning: autonomy (the experience of choosing one's actions rather than being controlled), competence (the experience of growing in capability and effectiveness), and relatedness (the experience of genuine connection and belonging within a social context). Performance management systems built on external monitoring and correction actively undermine all three of these needs when they are dominant rather than supplementary — signalling that the individual cannot be trusted to self-direct, emphasising performance gaps over capability growth, and substituting formal process for the relational investment that genuine connection requires.
Stewardship-oriented leadership addresses all three needs directly. Autonomy is cultivated by clarifying purpose and boundaries and then genuinely stepping back — trusting people to find their own path within those boundaries rather than monitoring the path itself. Competence is cultivated through regular, forward-looking conversations about growth — not what has gone wrong, but what the person is learning and what they need to go further. Relatedness is cultivated through the sustained, genuine investment in knowing each team member as a person — their motivations, their fears, their sources of meaning — rather than treating one-to-one conversations primarily as performance management occasions.
The First 100 Days
The transition to a stewardship mindset is not primarily a conceptual shift. It is a behavioural one, and it is most effectively made through a set of concrete practices adopted consistently over a defined period. The first 100 days in a new leadership role — or in the deliberate adoption of a new leadership orientation in an existing role — are the period in which new behavioural patterns are most malleable.
The first action is a decision landscape audit: a structured review of the decisions the leader is currently making or reviewing, categorised by consequence and by the degree to which they genuinely require human judgment versus AI-assisted analysis. The purpose is not to delegate more but to understand where the leader's attention is creating value and where it is functioning as a bottleneck or a signal of distrust.
The second action is a team listening exercise — not a survey, but a series of individual conversations in which the leader asks each team member what conditions most enable their best work and what most gets in the way. The insights from this exercise inform the leader's stewardship agenda: the specific environmental and relational conditions that are worth investing in most urgently.
The third action is the establishment of a regular rhythm of forward-looking development conversations — distinct from performance reviews — in which the leader and each team member explore what the person is learning, what challenges are stretching them, and what support would accelerate their growth. These conversations are short but consistent, and their value compounds over time.
The fourth action is the development of a human-AI team charter: a document co-created with the team that articulates the norms of how human and AI members of the team work together — including how AI outputs are reviewed, how errors are handled, and how the team's collective AI fluency is developed over time.
The fifth action is a stewardship commitment review at 30, 60, and 100 days — a structured moment of honest reflection in which the leader asks whether their behaviour over the preceding period has been consistent with the stewardship orientation they are committed to, and what adjustments are needed.
If you would like to assess your current leadership style and the alignment of your capabilities with the stewardship approach, our [Leadership Styles Inventory](/diagnostic/leadership-styles-inventory) and [4C Leadership Audit](/diagnostic/4c-leadership-audit) provide structured frameworks for understanding where you are and what to develop next.