The tempting shortcut
When a team adopts AI tools, it is tempting to explain friction through labels: one person is exploratory, another is cautious, and a third prefers automation. Those observations may start a conversation, but they do not tell a team who owns a prompt, who validates a result, or how a disagreement is resolved.
Replace labels with working agreements
Write down which decisions require a written rationale, which outputs need a second reviewer, and how a person can challenge an agent-produced recommendation. These agreements are usable whether a teammate identifies with a type, rejects typology entirely, or changes working style under pressure.
A better retrospective question
Instead of asking whether two personalities clashed, ask where information was lost, where authority was ambiguous, and whether the team had a safe path to raise uncertainty. That produces a repairable system rather than a story about fixed traits.