---
name: agent-meeting-runner
description: Prepare, run, and close meetings with agents so each one produces decisions and owned actions instead of discussion. Use when meetings consume time and nothing is decided or recorded.
---

# Agent meeting runner

A meeting that ends without decisions has cost everyone's time and
produced nothing. Agents cannot attend the conversation, but they can
prepare it, hold its structure, and turn it into recorded outcomes,
which is where most of the value and nearly all of the failure lives.

## Method

1. **Refuse the meeting without a decision to make.** An agenda that is
   entirely updates should be a written brief instead (see
   agent-executive-briefing).
2. **Prepare the pack in advance.** The decision, the options, the
   evidence, and the recommendation circulated before, so the meeting
   discusses rather than discovers (see meeting-agenda).
3. **Time-box per item and hold it.** Named durations per decision, with
   overruns pushed to a follow-up rather than consuming the last items.
4. **Capture decisions as they happen.** What was decided, by whom, and
   why, written during rather than reconstructed afterwards (see
   agent-decision-log).
5. **Convert every action into a commitment.** Owner and date, or it is
   not an action (see agent-accountability-loop).
6. **Send the record the same day.** Decisions, actions, and open
   questions, short enough to be read.
7. **Review the meeting's own worth periodically.** A recurring meeting
   that has produced no decisions for a month should be cancelled.

## Boundaries

Agents prepare and record; humans meet, argue, and decide. Recording
people speaking has consent and privacy implications that vary by
jurisdiction (see pii-handling), and sensitive discussions about
individuals must not go into shared agent-readable notes. An agent
cannot read the room, which is often where the real decision happens.