Delegating to AI agents is a management problem, not a technical one

Everything I know about delegating to people turned out to apply to AI agents: clear success criteria, the right-sized task, trust but verify. The failures I’ve had were almost never model failures. They were briefing failures — mine.