Operating Guide for AI Meeting Prep

This operating guide covers the use of AI meeting prep as practiced by individuals who, prior to a consequential meeting, ask an AI system to generate probable questions, key topics, or anticipated objections, then attend the meeting presenting this preparation as the product of their own prior thinking.

The subject is AI meeting prep: the use of a language model to accelerate pre-meeting preparation in ways that the person doing the preparing has not disclosed to meeting participants and has no specific forum in which to disclose. The help is useful enough to use. It is awkward enough to disguise.

AI meeting prep adds one private manual check. The check consists of reading through the generated questions and removing any that sound generated. The removal is a form of editing. The editing is described internally as ensuring quality. The quality check takes approximately eight minutes and produces a document that is presented as preparation.

That check becomes part of the normal workflow. The first time AI meeting prep is used, it is accompanied by mild unease. By the third time, the unease has been replaced by the editing step. By the sixth time, the editing step has been abbreviated. The preparation is still generated. The abbreviation is not noticed by the meeting's other participants.

The contradiction at the center of this guide is one of attribution. Experience, in a professional context, is the accumulated capacity to anticipate problems. AI meeting prep produces, on demand, a list of probable problems for contexts the user has not personally encountered. The capacity this replaces is not erased. It is simply not the source of this particular output.

Under the governing rule, AI meeting prep becomes procedure once repetition feels safer than reconsidering it. Assisted work becomes routine while disclosure stays exceptional. The exceptional nature of disclosure is maintained not by rule but by the absence of a place to put it.