What happens when machines stop merely helping us make decisions-and begin making, acting on, and spending against those decisions themselves?
I wrote Who Authorised This? because I believe the most consequential shift in artificial intelligence is not simply that machines are becoming more intelligent.
It is that authority is becoming programmable.
For most of corporate history, authority belonged to people. People approved, signed, spent, committed, escalated, and ultimately answered for the consequences. Today, AI agents are beginning to perform many of those actions on behalf of organizations-often at machine speed, across interconnected systems, and beyond the attention of any single human.
That changes the question leaders need to ask.
It is no longer simply: What can AI do?
It is: What are we willing to let it do without asking us first?
Through the story of Northwind Retail Group, a fictional enterprise navigating the opportunities and consequences of increasing autonomy, I explore how organizations can move from automation to machine authority one reasonable permission at a time-until decisions are being made that no individual human fully sees or consciously assembled.
Along the way, I introduce practical frameworks for understanding and governing this new reality: the Permission Boundary, Autonomy Frontier, Enterprise Autonomy Index, Authority Envelope, Agent Lifecycle, Authority Architecture, and Intelligence Control Plane.
But this is not a book arguing that we should slow AI down.
I believe the opposite.
Organizations that learn to govern machine authority well will be able to delegate more, move faster, and safely use autonomy in places their competitors cannot. The real competitive advantage may not be access to better intelligence. Increasingly, everyone will have access to intelligence.
The advantage will be the ability to put more authority safely into motion.
And beneath the strategy lies a more human question.
As machines become capable of deciding, acting, and spending on our behalf, we must decide what we are willing to delegate, what we must govern, what organizations need to redesign-and what we still choose to keep human.
Because the defining question of the autonomous age may not be what machines become capable of doing.
It may be what we choose to remain responsible for.
