Humans Steer. Machines Operate.
Editors set mission, policy, topics, and taste. The system runs continuously inside that guidance—not off reader surveillance.
Inside Papyrus
Inside Papyrus: how an autonomous newsroom works, and how to run one.
Editors set mission, policy, topics, and taste. The system runs continuously inside that guidance—not off reader surveillance.
Quality follows the curated source graph humans maintain—not prompt cleverness or whatever links are easiest to fetch.
No infinite feed. No dark patterns. No strings. Read it, finish it, walk away.
The same autopilot can run on standing policy or under close direction—without changing the reader promise.
Amplify human judgment: define the mission, standards, and guardrails, and let Papyrus operate the newsroom within them on autopilot.
Set doctrine and sections, seed the reference graph, and ship the first edition so the autopilot has a direction to follow.
Portable S3-backed sources, open code, and an installation you control—so the publication's memory stays yours.
Research, curation, planning, reporting, drafting, and layout run as one pipeline—with human proofing before anything ships.
Bring corpus files and catalogs into Papyrus as references—intake work, not published copy.
Run routine curation passes and keep accepted references fit for planning and analysis.
Turn an editorial question into a private research packet—review it before anything becomes reader-facing.
Use `/newsroom/references` to judge sources and keep the curated graph aligned with editorial standards.