Cognitive Agency

The premise

Cognitive agency is the human capacity to direct one's own thinking — to choose what to attend to, decide what counts as a good answer, and own the reasoning behind a conclusion. As intelligent systems absorb more of the work, the risk is not that machines think for us, but that we slowly stop noticing when they do.

This framework treats agency as a measurable, trainable capability rather than an abstract virtue. It asks: at each step of a workflow, who is actually doing the thinking — and does it matter?

Where agency leaks

Agency rarely disappears in one decision. It erodes through a thousand small defaults: accepting the first generated draft, trusting a summary without reading the source, letting a tool set the frame of a problem before you have formed your own.

The framework maps these leak points across a typical knowledge-work loop and offers concrete interventions to restore deliberate human control where it matters most.

Designing for agency

Preserving agency is a design problem as much as a discipline problem. Systems can be built to prompt judgment, surface uncertainty, and require a human decision at high-stakes junctions — without adding pointless friction everywhere else.