Editor-in-chief
Editorial leadership — the review queue, sensitive coverage, and the calls that set the tone for the desk.
Title-wise, the editor-in-chief is at the top of the editorial floor. Role-wise, you're typically admin (or owner if you also run the workspace).
What you do
Set editorial standards
What gets approved, what gets rejected, where the bar is. Your decisions become the reference.
Handle sensitive coverage
Sensitive-flagged stories — deaths, politics, legal, conflict, medical — route through editorial review by design. You make the call.
Approve / reject at the top
Final approvals on prominent stories, especially when desk editors aren't sure.
Spot patterns
Coverage patterns the team should watch for, gaps in the source set, regions or beats that need more attention.
Coach reviewers
When a desk editor's review is loose or rushed, your annotations and feedback raise the bar.
Daily workflow
Open the editorial board
/app/editorial. Filter to high-priority and sensitive-flagged drafts first.
Read provenance, not just the body
Every draft has source list, claim-level attribution, conflict markers, and a confidence read. Use them — they're how you triage at speed without losing rigor.
Approve, reject, or annotate
Approve when the draft holds. Reject with notes when something's off. Annotate without state change when the draft is fine but you want to coach.
Pull stories from the feed when needed
Sometimes a story breaks and nobody has filed a draft yet. From the feed, you can compose, refine, and submit yourself.
Watch the room
Glance at the editorial board through the day to see what's piling up. If a desk is backed up, redistribute.
How to read provenance fast
Source count + tier
Two wires + a tier 1 outlet > five tier 3 blogs. The source list shows tier badges.
Conflict markers
If sources disagree on a substantive fact, you see both versions. Decide which to publish — the AI does not decide for you.
Claim-level attribution
Click any sentence in the draft to see which source supports it. Unsupported claims should be your first concern.
Confidence read
A weighted signal, not certainty. Low confidence means slow down, not auto-reject.
Sensitive coverage
Always reviews
Sensitive-flagged stories cannot bypass editorial review, regardless of confidence. The flag fires for:
- deaths or casualties,
- political figures,
- legal proceedings,
- medical or health claims,
- armed conflict.
There is no override. The decision is yours; the queueing is non-negotiable. See Sensitive coverage.
When to reject
The cleanest reasons to reject:
- Unsupported claims — a substantive sentence has no source.
- Conflicts smoothed — two sources disagreed and the draft picked one without flagging.
- Wrong angle — the draft missed the story your audience actually needs.
- Tone or voice — sounds like a chatbot, not your publication.
- Sensitive call — the cluster involves a sensitive topic and the draft handles it the wrong way.
Always include a note when you reject. The author needs to know what to fix.
When to annotate without rejecting
When the draft is publishable but you want to coach:
- "This is fine; next time check the AP wire too — it had a different angle."
- "Approved, but the conflict marker should have been more prominent."
- "Good lede; cut the third paragraph."