The briefing, explained
-
Start here
How to read an edition
The four parts of every edition, and the fastest path to what matters to you.
22s · rendered presenterTRANSCRIPT Read it instead▸
Welcome to DAYBREAK, the morning read on the fights that matter. Every edition has the same four parts. The ledger up top shows how the edition was built. The executive summary gives you the through-line in a few sentences. The priority stack tells you what to do first. Then the feed carries the day's items, each one with a plain action at the bottom. If you only have two minutes, read the summary and the stack.
-
Reading the ratings
What the source ratings mean
Every item carries a letter and a number. Here is how to read them at a glance.
27s · rendered presenterTRANSCRIPT Read it instead▸
Every item in DAYBREAK carries two marks, and they mean different things. The letter rates the source: A is reliable, B is usually reliable, and it goes down from there. The number rates the information itself: one is confirmed, two is probably true, and higher numbers mean less certainty. So a B-two is a usually-reliable source reporting something that is probably true. Developing never reads as confirmed here. When you see a lower rating, treat the item as a lead, not a fact.
-
For parents and programs
The child-protection lens
How child-priority items are flagged, and where a real emergency should go instead.
29s · rendered presenterTRANSCRIPT Read it instead▸
Some items in every edition are flagged as child-priority. They are marked in red and routed to the top, and you can filter the whole edition down to just those with the child-protection lens button. Please hear this clearly. DAYBREAK reports on policy so that people can act on it. It never handles case material and it is never a place to make a report. If a child is in danger right now, call nine one one. For exploitation, the NCMEC CyberTipline is the right door, and the number is at the bottom of every page.
-
Behind the brief
How an edition gets built
Nine stages, one human gate. Nothing here publishes itself.
27s · rendered presenterTRANSCRIPT Read it instead▸
Here is how a DAYBREAK edition comes together. Before dawn, the pipeline pulls the morning's sources, clusters the same story across outlets, and drops the duplicates. A classifier tags every item, and child-priority items are sorted first. Then an editor assigns each one a source rating and a status. Every item gets routed to a tool and a concrete action. And then the part that matters most: a person reads the whole edition and approves it. Nothing here publishes itself.
TRANSCRIPT Read what the audio says▸
Here's how a DAYBREAK edition actually comes together. Before dawn, the pipeline pulls the morning's sources, wires, government feeds, courts and markets. It clusters the same story across outlets and drops the duplicates. A classifier tags every item, and the child-priority items get sorted to the top first. Then an editor assigns each item a source rating and a status, so that something still developing never reads as confirmed. Every item is routed to a tool and to a concrete action. And then the part that matters most. A person reads the whole edition and approves it. Nothing here publishes itself.
Narrated by a rendered voice (Laura · eleven_v3), prepared ahead of time. Nothing is spoken live and nothing you do on this page is sent anywhere.
Two marks on every item
From August 1, DAYBREAK uses the Admiralty scale, which rates the source and the information separately. That separation is the whole point: a reliable outlet can pass along something unconfirmed, and an unfamiliar source can be exactly right.
The four earliest editions predate it and carry a single status tag instead — confirmed, developing, preliminary or reported. Those are an editor's call on the item, not a separate reading of the source, and the archive does not pretend otherwise by converting one into the other.
- The letter — source
A · B · C…
A reliable · B usually reliable · and downward from there. It describes the track record of who is reporting, not whether this particular claim is true.
- The number — information
1 · 2 · 3…
1 confirmed · 2 probably true · and downward. It describes this claim, on its own, today.
- Reading them together
B2
A usually-reliable source, reporting something probably true. Treat it as a strong lead, not a settled fact. Developing never reads as confirmed here.
Nine stages, one human gate
The machine gathers, clusters, tags and drafts. A person approves before anything goes out — because a brief that flags child-safety and market risk cannot ship unread.
- Ingest — pull the morning’s sources: wires, government feeds, courts, markets and the calendar.
- Dedupe & resolve — cluster the same story across outlets, drop the duplicates, keep the primary source.
- Classify — tag each item by category and status. The child-priority classifier runs first and routes flagged items to the top.
- Verify & status — assign confirmed, developing, preliminary or reported. Unverified claims are marked, never smoothed. This is the first human touch.
- Route — map every item to its response tool and a concrete action.
- Compose — render the edition, and trigger the embedded modules where the day calls for them.
- Human review — the mandatory gate. Nothing auto-sends. Child-safety items are checked to confirm they route to hotlines and never become a reporting channel.
- Deliver — send, and publish the web edition alongside it.
- Log & measure — track what drove action, feeding tomorrow’s priorities.
What DAYBREAK will not do
- No edition sends unreviewed. There is no autopilot.
- Status tags are required. Developing never reads as confirmed.
- Child-safety items route outward. To NCMEC, to the hotlines, to 911 — DAYBREAK never becomes a reporting channel and never handles case material.
- Corrections are logged and carried in the next edition.
- No claim outruns its evidence. Where a number is a snapshot, the page says it is a snapshot.
Human trafficking — National Hotline 1-888-373-7888 · Childhelp 1-800-422-4453
Crisis — 988 · Emergency — 911