Purpose-Bound Autonomy: Why Architects Need More Than OAuth Scopes

Agent platforms default to coarse scopes: read mail, call CRM, use payment API. Enterprise architects need finer grain—search flights but do not book, approve up to a limit, act only for this customer case, only until this delegation ends.

Purpose-bound authority in practice

• Bounded work — a named undertaking with owner, scope, and expiry
• Approved purpose — policy evaluated when authority is minted, not inferred later
• Argument binding — tool calls must match what policy approved at mint time
• Revocation that propagates — kill switches that affect the next action, not the next login

Where CDA fits (architect depth)

Continuous Dynamic Authorization (CDA) is EmpowerID's implementation pattern for purpose-bound autonomy: compose authorization at mint time, re-evaluate when signals change, and enforce at the execution boundary. The durable governed-undertaking concept is informed in part by Karl McGuinness's Mission-Bound Authorization Handbook. EmpowerID uses the general term governed undertaking while composing purpose-bound authority with shared signals, action-time policy, Governed Execution, and causal evidence.

Autonomy without purpose binding is just faster privilege escalation.

Pair this model with Governed Execution so agents never hold the keys—and with cryptographic receipts so auditors can trace each action to approved purpose.