The Age of Diffused Agency
In our previous post, we explored how anti-fragile chaos engineering can help organisations build resilience in the face of uncertainty. We saw how injecting controlled volatility can strengthen systems and prepare them for real-world shocks. But what is it about the current environment that makes this so crucial?
Leadership is moving beyond the jagged frontier of what machines can do. Artificial Intelligence is not yet Artificial General Intelligence, yet the boundary of machine capability continues to advance. The space of tasks that require human-only intervention shrinks each quarter as new language models pair with agentic tooling to run longer chains of execution with less supervision. Competence that once demanded firms, teams, or specialist expertise now sits within reach of motivated individuals, sometimes on a single high-end consumer GPU.
Agency is diffusing. Execution power is no longer a privilege reserved for large organisations because it is being unbundled and placed directly in individual hands. With the right orchestration, anyone can behave like a chief executive who directs an army of digital staff. This is not the singularity, yet it is already a strategic revolution.
How this post fits the series
- Resets the cultural context after discussing stress-tested autonomy, showing why agency abundance changes power dynamics.
- Provides the rationale for introducing autonomy gradients so leaders can choreograph decision rights as capability diffuses.
- Reinforces the need for double-loop learning to keep assumptions honest when individuals can act without waiting for hierarchy.
From Firms to Individuals
For centuries, firms were necessary because coordination and execution required scale. Labour, capital, and know-how were bundled together to deliver outcomes that no single person could manage alone. AI-driven agency changes this equation, as a single operator can now perform tasks that previously required dozens of people.
The line between employee and tool is blurring as agentic systems collaborate with people. Institutions are losing their monopoly on competence, and what remains scarce is trust, judgement, and access to distribution. In Wardley terms, the value chain is reorganising: execution is becoming a commodity, while orchestration and positioning are becoming more strategically important.
New Strategic Dynamics
Autonomy length
Autonomy length is a practical measure of leverage. Leaders need to assess how far an agent can run before a human needs to intervene, as every additional step compounds both opportunity and risk.
Copy latency
Copy latency is a live metric. When competitors can reproduce what you build in a matter of days, your competitive advantages will erode unless they are based on brand, proprietary data, or regulatory advantages.
Security asymmetry
Security asymmetry is intensifying. Low-skill actors now have the power that was once reserved for states or major corporations. This means that defence must become agentic to match the speed and creativity of attacks.
Oversight load
The oversight load is threatening to swamp institutions. Contracts, compliance checks, and verification obligations are multiplying as synthetic labour outpaces traditional governance processes.
The Human Scarcity
As the act of doing becomes a commodity, leadership shifts to the scarcities that only humans can provide. Judgement is needed to select the right problems and set the constraints that agents must respect. Taste is needed to determine what good looks like in a world of overwhelming output. Trust is cultivated by building resilient relationships, communities, and brands that can withstand cheap imitation. Negotiation is needed to secure distribution, regulation, and access to scarce resources like compute and differentiated data.
Execution still matters, but it is increasingly automated, orchestrated, and audited. Leaders are now more like editors and conductors, curating leverage rather than being hands-on doers. This is why many organisations are investing in autonomous strategy execution stacks that let agents trigger plays directly from the map. They are also pairing them with anti-fragile chaos engineering drills that keep human judgement sharp by building anti-fragile reflexes.
Societal Shifts
Diffused agency is reshaping more than just firms. The economy is benefiting from productivity gains, but wage pressure is intensifying in mid-skill execution roles. Inequality is widening between those who are agent-rich (with compute, literacy, and networks) and those who are agent-poor. Institutions are struggling because law, procurement, and compliance are lagging behind accelerated execution. This is creating friction, liability debates, and a demand for new governance defaults. Security threats are scaling down to individuals, so countermeasures must scale automatically in response.
Leadership Beyond the Jagged Frontier
Leadership in this environment is not about commanding labour, but about curating leverage. Agentic stacks should be designed as if they were teams, with clear responsibilities, playbooks, and escalation paths. It's important to measure autonomy length, monitor copy latency, and track oversight costs to understand where intervention is essential. When diffused agency becomes overwhelming, it's time to revisit the governance patterns in continuous map governance to keep intent aligned across the swarm.
Invest in assets that compound over time, such as data, distribution, and trust, rather than chasing every new output. Anticipate that governance will drag more than technology will accelerate, because execution has become a commodity. The new leverage is in higher-order strategic choices: where to play, how to constrain, and when to intervene.
References
- Brynjolfsson, E., Raymond, L. R., & Rock, D. (2023). Generative AI at work. National Bureau of Economic Research Working Paper No. 31161. https://www.nber.org/papers/w31161
- Stanford Institute for Human-Centered Artificial Intelligence. (2024). AI Index Report 2024. https://aiindex.stanford.edu/report/
