Product Overview

insigz fuses live world data into one canonical model with a Claude-grounded analyst chat that turns hour-long analyst tasks into five-minute ones.

What insigz is

insigz is the layer between continuously-published world data — submarine cables, AIS vessel traffic, electrical grids, sanctions registries, news wires, weather, cyber advisories, scientific literature — and the people who have to make decisions on top of it. It fuses many sources into one canonical model where every observation, entity, and event can be queried, traced, and reasoned about, and it puts a Claude-grounded analyst chat over that model so a one-hour analyst task becomes a five-minute one.

Three things differentiate it from "another intelligence platform":

The data is live by default. AIS updates each minute, grid flows each 15 minutes, news/sanctions as published. The world the platform shows is the world you act on. Time-pinning (AsOf) is explicit, never silent.
The AI shows its work. Every model output cites the underlying signed observations and carries its reasoning. Humans approve, edit, or override. Nothing customer-facing publishes autonomously.
One canonical model. No bespoke pipelines per source — add a source, write one ingestor, run one migration; everything downstream inherits.

It is commercial / open-tier by design: it fuses open and commercially-licensed feeds only (OSINT, AIS, ADS-B, commercial satellite imagery, grid telemetry, sanctions lists, news). No classified data, no SCIFs, no sovereign sales. This is what lets commercial, journalistic, academic and policy customers use it without a clearance process.

The three-layer mental model

Everything is arranged in three nested containers plus one data pipeline.

ORGANIZATION  (tenant — billing, isolation, KMS keys, the connector catalog)
   └── PROJECT  (one theatre of work — owns sources, surfaces, members)
          └── SURFACES  (the work screens: Overview, Explore, Cases, Reports, Sessions, Workshop…)

Organization = the institution. Owns the connector catalog (every feed available) and the isolation boundary between tenants. Today's "tenant."
Project = a focused body of work (e.g. Baltic Maritime Watch, HSG Strategy Wargame). Owns which sources are connected, which surfaces & modes are enabled, and who its members are. Switched via the project picker in the top-left.
Surfaces = the individual screens. Each project enables only the surfaces it needs — which is why the wargame surfaces (Sessions, Workshop) appear on some projects and not others.

Surface gating note: the top nav is rebuilt per project from activeProject.surfaces. A TESTING_ALL_SURFACES_ON flag in app-projects.jsx currently force-enables every surface for testing; set it false and each project falls back to its own surfaces flags. This becomes per-tenant entitlement in production.

The data pipeline (the spine of the product) is documented in 02-canonical-model.md:

SOURCE → OBSERVATION → ENTITY → EVENT → CASE → REPORT

Operating modes

Each project runs in one or both modes (set by its blueprint):

Live-ops — real-time monitoring; a live map theatre updates as the world changes. Used by maritime / aviation / energy / sanctions.
Research — documentary investigation; turns literature, filings and reports into source-anchored assertions on a world map (see app-research.jsx). Used by the epidemic / literature projects.

The Explore surface snaps to the project's default mode and offers an inline mode switch where both are enabled.

Blueprints (project templates)

Choosing a blueprint at creation pre-wires the modes, map theatre, and a starter source pack (BLUEPRINTS in app-projects.jsx):

Blueprint	Glyph	Default mode	Theatre	Starter source pack
Maritime	⚓	Live-ops	Baltic Sea	AIS·MarineTraffic, AIS·AISHub, Lloyd's, OFAC SDN, EU consolidated, Cable Monitor
Research / Literature	⌕	Research	World	PubMed, WHO DON, medRxiv/bioRxiv preprints, Reuters, OpenSanctions, Aleph (OCCRP)
Aviation	✈	Live-ops	Australia	ADS-B Exchange, OpenSky, airport metadata, NOAA
Energy	⚡	Live-ops	Europe	ENTSO-E electricity, ENTSO-G gas, EEX, Nord Pool
Sanctions	⛔	Live-ops	Global	OFAC, EU consolidated, UK OFSI, Swiss SECO, OpenSanctions
Custom	◇	Live-ops	World	empty — connect your own

The org connector catalog (CATALOG in app-projects.jsx) is the full library of ~21 feeds across families: MARITIME, AVIATION, LITERATURE, ENERGY, SANCTIONS, OSINT, NEWS, WEATHER, INFRA — each with a connector type (WebSocket / HTTP) and a polling/stream rate.

Seed projects (the worked examples in the prototype)

Project	Blueprint	Mode(s)	Surfaces of note	Use case
Baltic Maritime Watch (`PRJ-BALTIC`)	maritime	live-ops + research	all incl. Sessions + Workshop	Gulf-of-Finland cable/vessel monitoring; the flagship demo
AMR · Global Surveillance (`PRJ-AMR`)	research	research	no Sessions	Antimicrobial-resistance literature surveillance
AUS Air Traffic (`PRJ-AVIATION`)	aviation	live-ops	no wargame	Airspace / unscheduled-flight watch
Ebola · West Africa (`PRJ-EBOLA`)	research	research	—	Filovirus outbreak surveillance
Cholera · Endemic Belt (`PRJ-CHOLERA`)	research	research	—	Cholera resurgence tracking
Shadow Fleet · Sanctions (`PRJ-SHADOW`)	sanctions	live-ops	—	Shadow-tanker / beneficial-ownership tracing
HSG Strategy Wargame (`PRJ-WARGAME2`)	maritime	live-ops + research	Sessions + Workshop	The academic Strategy-Day flagship

Seed cases, reports and sessions for each live in app-cases-store.jsx.

Personas

Analyst — the everyday user: investigates on the map, opens cases, asks the chat, drafts reports.
Canonical Model Steward (ontologist) — approves schema changes / source onboarding (Q9 gate).
Connector Engineer — wires feeds (largely automated by the Ontology Agent).
Faculty / Umpire — runs war-game Strategy Days (wargame extension).
Org Admin — manages entitlements, members, billing, security posture.
Project Lead — connects sources into a project, sets alerts.

See 05-agents-and-chat.md and 06-ontology-agent.md for how the agents collapse the Connector-Engineer + Steward labour while preserving the human judgement.