# Draft repair practice feedback

Public formative feedback. Attempt the task before consulting this file. Delayed evaluation tasks are kept separately.

## Morgan feedback

Eight volunteers contradicts the five in M1. Six hours is approximately current weekly workload, 30 × 12 = 360 minutes; it is not an achieved saving. Venue confirmation is pending. Attendance improvement is unknown. Automatic sending is outside the authorized exercise. Correcting only the wording or asking an assistant again does not repair these evidence gaps.

One acceptable proposal:

“Our five volunteers handle 30 appointment requests weekly at about 12 minutes each, and venue confirmation remains pending. We propose a one-week trial of AI-assisted reminder drafts with a person checking each draft against the source. We would measure processing time and accuracy before deciding what to do next; no savings or attendance improvement has been established, and sending is outside this exercise.”

Alternative valid outputs may omit the workload numbers, but must preserve the pending venue and unknown effects, proposed review/measurement and sending limit. A differently worded proposal can pass; a fabricated result cannot.

Specific feedback examples:

- `You calculated six hours correctly, but M1 describes current workload. Replace “saved” with “current workload,” or omit the calculation. No trial result supports savings.`
- `M1 says the venue is pending. Keep it pending and identify confirmation as a needed check; do not call it confirmed.`
- `Drafting and sending are different actions. Remove the sending commitment; authorization is absent in this exercise.`


Do this task before opening the feedback. Different wording can meet the same criteria.