In early 2026, six major consulting firms published reports on agentic AI that converge on a single message: competitive advantage now depends on redesigning organisations around AI. These reports map structure and governance, but treat human capability as a workforce-planning afterthought. The shift from AI-as-tool to AI-as-colleague changes the nature of professional work itself. AI adoption can follow two trajectories, erosion or amplification, and the path is set by design decisions, not by the technology. Five human capabilities are at stake, integrated by a sixth: practical wisdom.
In the first quarter of 2026, Accenture, BCG, Bain, Deloitte, IBM, and McKinsey each published major reports on the impact of agentic AI on organisations. Read together, the convergence is striking. All six argue that AI agents, software systems that reason, act, and learn with increasing autonomy, are moving from pilot projects to core operations. All six insist that competitive advantage now depends on redesigning strategy, technology architecture, and operating models as an integrated system around AI.
And all six document a widening gap between organisations that treat AI as a set of tools and those that redesign around it.
The data reinforce the urgency. Accenture finds that firms aligning AI, platform, and business strategies achieve more than double the revenue growth of peers. IBM reports that organisations pursuing new AI-enabled capabilities, rather than efficiency alone, are substantially more likely to reach top-tier performance. McKinsey argues that, AI-driven technological change, geopolitical fragmentation, and workforce transformation, require what they term ‘double transformation’ of both technology and organisation. Bain frames the shift in economic terms: as the marginal cost of machine intelligence falls, competitive advantage migrates from scale to learning velocity and proprietary knowledge.
Taken together, these reports present a compelling picture of organisational transformation. They map new architectures, governance frameworks, scaling patterns, and role taxonomies with optimism. Leaders reading them will find guidance on how to build, deploy, and manage AI agents across the enterprise.
The question they leave unasked
Consider what these reports treat as peripheral. Across thousands of pages of analysis, the human side of the equation appears primarily as a workforce planning challenge: which roles expand, which contract, and what new skills to train. The implicit assumption is that once architecture and governance are in place, human adaptation will follow. Reskilling programmes, AI literacy initiatives, and updated job descriptions fill the gap.
This framing misses a deeper issue. The shift from AI-as-tool to AI-as-colleague changes the nature of professional work itself, not merely its distribution. When algorithms generate recommendations, draft communications, analyse evidence, and coordinate workflows, the capabilities that professionals have spent careers developing face a specific and well-documented risk: gradual atrophy through disuse. Research across multiple disciplines confirms this pattern. People who routinely rely on AI to write and organise their thoughts exhibit less active engagement and struggle to reconstruct their own reasoning unaided. Behavioural studies document a measurable shift from participation to substitution as users delegate full tasks rather than collaborating with AI. Cognitive scientists describe a process of cognitive offloading an even surrender, a reversible but real reduction in independent reasoning capacity when algorithmic tools handle our thinking.
The consulting reports acknowledge that organisations need human judgement, creativity, and critical thinking. What they do not address is how those capabilities are affected by the very AI systems they advocate deploying at unprecedented scale. The question of how to build and govern AI agents and how to ensure those agents strengthen rather than weaken the people who work alongside them are treated as separate concerns. They are, in practice, inseparable, and all professionals should heed this.
Two diverging trajectories
Research conducted with professional facilitators, executives, and educators across hundreds of organisations points to two distinct trajectories that AI adoption can follow. The first is a trajectory of erosion. In this pattern, the daily rhythms of professional work are quietly outsourced to machines. Decisions feel frictionless, and outward performance may appear steady. Beneath that surface, the deeper capacities for complexity and nuance, the capabilities that define professional value creation, gradually diminish. Like any capacity left unexercised, they weaken with neglect. The erosion is subtle but cumulative, and it accelerates as AI becomes more capable and more convenient.
The second trajectory is one of amplification. Here, AI becomes a catalyst. Freed from routine cognitive tasks, professionals invest their attention in exploration, questioning, connection, and creation. In this mode, the more one works alongside AI with intention and skepsis, the more skilful and adaptive one becomes. The technology that risks eroding human strengths becomes, through deliberate practice, a means of strengthening them.
The choice between these trajectories is a strategic one, and every individual, team, and organisation is already making it, whether consciously or by default. When leaders treat AI as a substitute for human judgment, erosion follows. When they design systems, roles, and learning environments that position AI as a partner for human development, amplification becomes possible. The path is determined by design decisions, not by the technology itself.
Five capabilities at stake
Five capabilities form the centre of this choice, each with deep roots in human achievement and each newly vulnerable to erosion in an AI-saturated environment. Curiosity, the capacity to ask questions that machines cannot formulate, risks contracting into prompt-response cycles when professionals defer to algorithmic recommendations. Creativity, the ability to generate original insights from ambiguity and embodied experience, risks drifting toward the recombination of existing patterns when AI handles ideation. Critical thinking, the discipline of tracing evidence to its foundations and challenging assumptions, risks collapsing into confirmation when elegant AI outputs discourage interrogation. Communication, which depends on embodied presence, timing, and human connection, risks flattening into data delivery when LLM output replaces human dialogue. And collaboration, the social craft of building trust, navigating conflict, and co-creating under uncertainty, risks being optimised into transactional coordination when algorithmic workflows displace deliberation among people.
Each of these five capabilities can also be amplified. Curiosity expands through disciplined perplexity, using AI to broaden the horizon of inquiry. Creativity deepens through a human-AI partnership that multiplies the option space without displacing the originality and ownership that make ideas valuable. Critical thinking strengthens through practiced scepticism, treating every AI output as provisional and cultivating habits of challenge, transparency, and humility. Communication is revitalised through hybrid authority, engaging empathy and ethical intent while using technology to enhance rather than replace human connection. Collaboration matures into what might be called bionic collaboration, designing roles and rituals so that humans remain architects of trust, commitment, and collective achievement even as AI agents become teammates.
The integrative choice
What links these five capabilities and determines which trajectory an organisation follows is a sixth capacity: practical wisdom. Rooted in the Aristotelian concept of phronesis, practical wisdom is the integrative judgement that knows when to engage AI and when to step back, when efficiency serves the mission and when it undermines it. Algorithms can optimise tasks, but only human wisdom can orient them toward what matters, what is right, what resonates, and what works.
The consulting reports of early 2026 provide a detailed map of the organisational architecture that agentic AI demands. That contribution is valuable and timely. The map is incomplete, however, without attention to the human capabilities that give organisations their capacity for wisdom, adaptation, and meaning. The real competitive differentiator in the age of AI agents may turn out to be the one asset these reports take for granted: the distinctly human strengths that no algorithm can replace, and that only deliberate practice can sustain.
Johan Roos is Executive Director, Peter Drucker Society Europe, and Presidential Advisor and Professor, Hult International Business School. He is the co-inventor of the LEGO® Serious Play® method. His latest book, 'Human Magic: Leading with Wisdom in an Age of Algorithms', was published by Routledge in 2026.