Who actually owns the call when AI is in the workflow?

Why it matters

Insurance, finance, and product leaders ask for AI policy but need decision architecture. A charter connects tier language to named humans who can be promoted, demoted, thanked, or fired, not algorithms holding accountability.

What this is

Decision Rights is the three-tier line for human-owned judgment: Delegate (the machine decides; humans audit—reversible, individually low-stakes, high-volume, with sample audits and an escape hatch), Augment (AI drafts; a named human decides and must defend the call from a blank whiteboard without the model), and Reserve (human-only judgment—AI may retrieve data, research, and model scenarios, but synthesis, weighing, and recommendation remain human). Accountability never belongs to the algorithm. A named human remains accountable. The Human Veto requires named authority, defined grounds, no-fault review, and rehearsal.

What decision failure does this prevent?

Teams adopt AI faster than governance catches up, shadow automation runs where policy exists but no tier labels exist on real workflows, and incidents reveal nobody knew whether the call was Delegate or Reserve.

Who commonly owns this problem

Chief Risk Officer, General Counsel, or AI governance lead rolling out tier labels

Practical example

A claims workflow scores recommendations automatically. Without tier labels, legal assumes Reserve while engineering ships Delegate behavior, until an audit asks who owned the call.

What to do in the next 15 minutes

Open the Decision Rights Charter builder, pick one recurring workflow, and run the three-question tier test, Delegate, Augment, or Reserve, for each decision in that flow.

See where you stand →