§ Executor · The discovery layer of your stack
PulseRadar.
Scans your market continuously, surfaces the signals worth a response, and drafts the response itself. The signal-to-draft loop closed.
01 What it does
PulseRadar is the platform that decides what's worth a response in your market — and writes the first draft of that response. A continuous monitor across the sources your domain cares about, a ranking layer that filters noise from signal, and a draft generator that hands the operator a starting point, not a blank page.
Most "market intelligence" tools deliver alerts. You still have to read every alert, decide which matter, and start writing from zero. PulseRadar does the reading, the deciding and the first 80% of the writing — so the operator only spends time on the work that demands their judgment: the final 20%.
Continuous source monitoring
A standing scan across the sources your domain depends on — competitor sites, industry feeds, regulator updates, partner channels, primary databases. New signals enter the queue within minutes of being published.
Signal ranking
Every signal is scored against criteria you define — relevance to your audience, novelty against your existing coverage, urgency, competitive gap. Noise drops below the threshold, signal stays visible.
First-draft generation
Signals that clear the threshold trigger a draft — researched, written, sourced — that lands as a ready-to-edit asset in your CMS. The operator picks up at editing, not at staring.
Entity discovery
When a signal references an entity you don't have coverage of, PulseRadar surfaces it to HubForge so the entity hub gets built before the article that mentions it goes live.
02 How it works
PulseRadar runs as a durable workflow on a continuous schedule. The signal-to-draft loop is closed inside the platform — no operator handoff required to get from "this is new" to "this is a draft."
Monitor. The platform pulls from your configured source set — RSS, API endpoints, scraping targets, partner feeds — on a fast cadence and surfaces what's new since the last sweep.
Score. Each new signal gets a relevance score, a novelty score, and an urgency score against your domain. Items below the threshold drop out; items above enter the ranked queue.
Cluster. Signals that touch the same underlying event are clustered so the operator sees one ranked item, not five duplicates from different sources. Provenance is preserved.
Draft. Top-ranked items trigger a draft workflow that researches the angle, writes the body, attributes the sources, and hands the asset to your CMS as a draft ready for editing.
Route. The draft lands in your queue. Atlas posts a brief note. New entities flow to HubForge. The operator opens the asset and starts where the platform left off.
03 What it replaces
PulseRadar collapses the function of "what's happening in our market and what should we say about it" — usually distributed across a junior analyst, an alert tool, and the operator's morning RSS reader — into a single operated layer.
04 Live receipts
PulseRadar currently operates the discovery layer of a live reference deployment — an editorial publishing platform under Umbra Group.
Reference deployment — sustained since 2026-05
- Continuous source set live across primary domain feeds, with new signals landing in the queue within minutes of publication.
- Ranked surfacing filters volume to the items worth a response — the operator never sees raw alert noise.
- Draft generation produces ready-to-edit assets in the CMS with sources attached, body drafted, angle scoped.
- Entity discovery feeds HubForge — new entities that appear in the queue get hub pages built before the article that mentions them ships.
- Resilience built in — early-abort budgets and per-source error classification keep one bad endpoint from poisoning the whole sweep.
Source — Internal observatory + CMS draft audit · 2026-05
05 Pairs with
PulseRadar sits at the front of the content pipeline — what it surfaces flows downstream to the rest of the stack.
HubForge
New entities surfaced in the queue get hub pages built before the article that mentions them ships.
Feeds — distributionPressRelay
Drafts that go live land in the distribution pipeline with metadata and per-channel adaptation queued.
Operated by — cockpitAtlas
New top-ranked signals and draft outputs surface in the Atlas morning brief.
Feeds — structured dataSchemaForge
Drafts that ship to the CMS get schema applied on publish, with entity references already resolved.
06 Deploy in your org
PulseRadar is deployed during a Lighthouse Sprint that scopes your source set and your ranking criteria — what signals matter to you, what doesn't, and what "draft-worthy" looks like in your voice.
Source scoping (Pulso). We catalogue the sources your domain depends on, the cadence each can support, and the historical signal-to-noise ratio.
Scoring & draft tuning (Sprint). Ranking criteria are configured to your domain. Draft generation is tuned to your voice and angle conventions. The handoff to your CMS is wired with provenance preserved.
Ship (Watch). Platform goes live behind Atlas. Draft volume calibrates to your editorial capacity. The operator stops monitoring; they edit.
07 Pricing line
PulseRadar is included as an executor in any Lighthouse Sprint that touches your discovery, intelligence or editorial workflow. Operating cost is bundled into your Lighthouse Watch tier and scales with source count and draft volume.
Standalone deployment (PulseRadar alone) is available from the base Sprint tier. Detailed pricing in the Sprint framework.
08 Engage
See signal-to-draft closed on real sources.
We walk you through the live reference deployment: a fresh signal entering the queue, scoring against domain criteria, clustering with related items, and the draft landing in the CMS minutes later with sources attached.