The pre-publish read your team doesn’t have time to do

Most pieces ship without a second set of eyes. Not because the team doesn’t want them, but because there isn’t the staff. This ability is a standing second-editor: voice, accuracy against sources, context the piece might be missing.

What it does

Reads a draft and the sources the team based it on.
Flags voice drift against the team’s published voice profile.
Flags claims that don’t have a corresponding source in the team’s brain (places the piece is leaning on memory or assumption).
Surfaces relevant context the piece doesn’t mention but probably should: prior coverage on the same topic, related items the brain has seen, factual nuance the draft glosses over.
Returns a short review note, not a redline.

The reveal

A writer files at 5pm. By 5:05 the editor has a half-page review note: three voice flags, two unsourced claims, one piece of context from an earlier piece the brain remembers but the writer didn’t. The edit pass takes ten minutes instead of forty.

What it doesn’t do

No copyediting. No grammar nags. No rewriting. This is editorial review, not proofreading, and it doesn’t replace a human editor where a human editor is available.

Configuration

What to flag: voice, sourcing, missing context. Any subset.
Strictness: every drift, or only the high-confidence ones.
Output: an inline annotated draft, or a separate review note.

Triggers