Coordinate and act
Challenge-and-response callouts + "question the unusual, listen, don't anticipate."
Say it out loud, confirm it back
A spoken check on critical handoffs, one person states the action and another confirms it, paired with a habit of naming anything odd.
Borrowed from crew resource management, the teamwork training used across commercial aviation. Read the original source.
Ongoing habit · An hour or more to set up · Anyone on the team can start it
Try this first
On important AI handoffs, require one exchange out loud. The person acting states what they are doing and someone else confirms it.
Use it when
Critical handoffs are assumed rather than confirmed, or the team pushes past its own stop-points.
Skip it when
Save it for critical handoffs. Formal callouts on trivial actions become noise and people stop hearing them.
How to introduce it
For critical handoffs, especially where someone is acting on AI output, say it out loud. One person states the action: acting on the model's recommendation to do X. Another confirms it back: confirmed, acting on X. Pair it with a standing habit of naming anything that looks unusual instead of assuming it is fine.
How to show up
Set the spoken confirmation for the handoffs that matter. Thank the person who flags something that turns out to be nothing. Treat one flag as fussing and the flags stop.
How long it takes
A short session to set the practice. It then lives in the work.
What makes it hard
Naming the unusual fights a strong pull to assume the expected and keep moving, so name that pull and let people catch it in themselves. Callouts feel stilted at first and get dropped under time pressure, which is when they matter.
What it looks like when it's working
Handoffs get confirmed out loud and odd signals come up early instead of getting explained away. Silent handoffs and smoothed-over anomalies mean the habit did not hold.
How long until it sticks
Expect a few weeks of use before the spoken confirmations feel natural rather than awkward.
How you know it stuck
Confirming a critical handoff, and speaking up about the unusual, is how the team works.
The idea behind it
The dangerous errors are the ones everyone assumed someone else caught. Saying handoffs out loud, and naming the odd thing, is what catches them.
Where it comes from
Aviation crew resource management. AI handoffs pass quickly and quietly. Saying them out loud makes the transfer visible, and the habit of naming the odd thing catches output that looks plausible and is not. One person states the action out loud and another confirms it back. Alongside that, a standing habit: name anything unusual rather than assuming it is fine.
Also fits: Warning signs get explained away