Intelligence — publishing

The Intelligence view is the production half of the intelligence desk. Where Observer harvests and de-duplicates, this view turns the clean pile into finished, on-brand, source-grounded content — and publishes it to a content folder you choose, typically your own site.

The whole desk in one line: sources → items (de-duplicated) → observer reports → writers → your site and your outbox → your approval for anything that reaches another person.

The tabs are Overview, Generate and Images, plus an output directories button in the header that configures where each content type is published.

The content formats

Each writer is a fresh, disposable agent spawned for one run; the schedule, de-duplication, file paths, and send step are all mechanical.

Nothing in that column is running on a fresh install. Every format also exists as a routine, and every routine's schedule ships switched off — the cadences above are what each format suits, not what your machine is doing. You turn a schedule on, and set its interval, in Routines; until you do, the desk only produces when you dispatch it.

Briefings carry an impact tier — notable · significant · major · critical — computed from how many independent sources reported the same story. The story three sources covered outranks the single-source one, mechanically.

Dispatch a writer

1. Open Generate. Each format has its own dispatch panel, and a running dot on the tab shows any writer currently in flight.
2. Optionally steer the run with the free-form instructions field — "focus on enterprise implications", "lean toward critique" — which biases story selection and framing without editing anything by hand.
3. Dispatch. The writer reads the observer reports (and drills into the strongest raw items for specifics), writes a finished piece with complete frontmatter, and exits.

Two properties keep repeated runs safe. A consumption ledger means re-running a writer never re-produces the same story — material already turned into a briefing is skipped, and a force-reprocess run explicitly asks for a new angle or aborts rather than duplicate. And writers are grounded: when the source material is too thin to support an honest piece, the writer aborts instead of padding.

Most formats also exist as routines, so once you enable their schedules the desk runs as a standing publication — the breaking ticker every half hour, briefings each morning, the roundup weekly. Schedules, caps, and source filters are all editable; see Routines & automation. Writer runs spend your own AI subscription, on whichever agent CLI you have configured, and a scheduled writer is token-free when idle — it only spawns when there is genuinely new material.

Publish to your own site

The header's output-directories overlay holds the one setting that decides where content goes: the content site root, plus per-content-type overrides. Point it at your website's content folder and every finished piece is picked up by your site's content loader; point a type somewhere private and it stays on disk, yours only. Newsletters never publish to the site — they are email-only by design.

Your voice comes from your identity files: brand voice, byline and audiences are configuration every writer reads. Fill them in once and the whole desk produces in your register.

Language works in two different ways, and the distinction matters. The translator is driven by the publishing locales in your identity files — it produces a sibling article per locale you list. The newsletter is not: it writes an English edition and a German edition on every run, with distinct editorial content per language rather than a machine translation, whatever locales you have configured.

Images

The Images tab manages the illustration layer: hero images for briefings, reports, and roundups, generated from prompts composed out of the full article text. The image service de-duplicates against images already on disk, patches the image into the article automatically, keeps a cost ledger, and enforces a daily cap. Image generation uses an image-model API key you bring — it is optional, and everything else on the desk works without it.

The AI transparency mark

Every article hero the image service generates is stamped with the European Commission's official "AI GENERATED" mark, composited into the pixels in the bottom-right corner before the file is written. A page-level notice would not survive an image being downloaded or reshared; a mark burned into the file does.

• It is on by default, and it is the one setting on this desk that fails on rather than off: an unreadable settings file falls back to marking, not to skipping.
• The switch is in the Images tab's pipeline settings panel. Turning it off is a lawful choice in some contexts — artwork that could not pass as authentic, or a non-EU audience — which is why it is a switch rather than a constant. The panel states the consequence when you turn it off: new images carry no visible label, while images already generated keep theirs.
• Provenance also lands in the article's frontmatter — that the hero was generated, whether it carries the mark, the style and the model — so it stays greppable even after a delivery optimizer strips image metadata.
• Product covers and product section images are deliberately unmarked. They are commercial artwork rather than news illustration.

Sending the newsletter

Publishing to your own site is automatic — those are your files, on your disk. The newsletter is the one thing this desk produces that goes to other people, and it works differently from everything else here.

The newsletter is a bulk broadcast, not an SMTP send. It does not use the mail accounts in Settings ▸ Email, and configuring an account there does not make it work. It goes out through Resend, a broadcast provider you bring, to the subscriber audience you keep there. That means three settings before the first send, entered in Settings ▸ Email under newsletter sending:

Until all three resolve, the dialog's Send button stays disabled and tells you why, rather than failing after you click it. The same is true when the Email sending master stop is off — see If Send stops working.

The rest of the path:

• Newsletters are always drafts. The writer produces them into your outbox with a draft status; nothing is sent until you review and click Send. There is no autonomous send mode, and the writer has no send capability of its own.
• Send is per edition and irreversible. The dialog shows the subject, preheader and locale of the exact file you are about to broadcast; a broadcast cannot be recalled.
• Every send is recorded. The step stamps the file with a sent status, the send time and the provider's broadcast id — a per-issue audit trail on disk. Delivery numbers are not among them: the broadcast API does not return a recipient count at send time, so the broadcast id is how you look an issue's delivery up with the provider afterwards.

What goes out labelled

This desk labels what it produces in more than one place, and each is on by different terms. It is worth knowing which is which:

• Published articles carry an AI disclosure tag that the writer sets in the article's frontmatter. A site that renders it displays a locale-aware disclosure automatically — enforced per article rather than promised in prose.
• Generated heroes carry the transparency mark burned into the image, as described above.
• Every newsletter broadcast carries a disclosure sentence and a one-click unsubscribe link, rendered unconditionally into both the HTML and the plain-text part of the message. There is no flag that omits either — a broadcast without them is not a broadcast the platform will assemble.

Mail your email agent sends carries its own disclosure and automated-mail markers; see Email.

The Reports viewer

The report archive lives in the Observer view's Reports tab: the dated observer reports the processing layer produces, plus the deep video-analysis reports. This is the mid-layer worth reading directly — writers build on it, but each report is a useful intelligence artifact on its own, and the viewer lets you browse the archive by date without leaving the app.

What this desk deliberately does not do

• It does not invent facts — thin material aborts a run rather than padding it.
• It does not send the newsletter without your click. There is no autonomous send mode on this desk, and no schedule you can set that adds one.
• It does not run itself until you say so — no writer schedule is on when the product is installed.
• It does not machine-translate — translations are written in-agent for tone, which trades raw speed for quality.

Where to go next

• Observer — harvesting — the intake half of the desk
• Routines & automation — schedules, retries, and where you stay in control
• Autonomy & controls — what runs unattended across the platform, and the switches that decide
• Email — the mail cockpit, and the master stop that also governs the newsletter's Send button