Public information, organised into an awareness graph with evidence behind every claim.
This network works on the public-source, non-classified layer only. News, official releases, hazard and weather data, business registries and threat-intelligence sources flow in continuously. Entities, relationships, events and narrative leads each keep their source attached, so an analyst can trace any conclusion back to the original evidence.
What analysis teams deal with daily
The shortage is rarely information. It is structure an analyst can act on.
Sources spread across dozens of doors
News, notices, hazard data and company records each live somewhere else. The manual patrol gap is exactly when an event grows.
Conclusions without evidence
Summaries mutate as they are forwarded. When review cannot get back to the original text, the agency cannot brief from it.
Events & entities that will not line up
The same organisation, place or event is written five ways across sources, and the relationship graph never forms.
Access that is all or nothing
Mandates differ by department. One dataset for everyone fails compliance; locking everything away fails the mission.
Network make-up
Every layer, from intake to delivery, keeps its source.
Multi-source intake
Public web pages, official releases, hazard and weather data, business registries and threat-intelligence sources connect under licence. Each source is registered with its permission, coverage and update frequency.
- News and public web sources
- Official releases and notices
- Hazard, weather and water data
- Business registries and threat-intelligence sources
Entity & relationship graph
Organisations, places, events and public figures become entities. Relationships and timelines are queryable, and one asset serves many tasks.
- Four entity classes: organisations, places, events, public figures
- Name variants normalised to one entity
- Relationships and timelines queryable
- Graph changes kept under version control
Narrative & disinformation screening
Disaster rumours, influence activity and abnormal spread paths are flagged with the propagation chain and the evidence, never with an unsourced verdict.
- Abnormal spread paths flagged apart
- Rumour leads carry the propagation chain
- Every conclusion carries an evidence link
- No verdict without a source
Tiered workspaces & alerts
Data is supplied by agency, role and task. New events push under set rules; exports and access are logged end to end.
- Partitioned by agency, role and task
- New events pushed under set rules
- Exports and access logged end to end
- Permission changes take effect at once, on record
The data asset card
With intelligence, read the data before the features. Each item is fixed before a project starts.
Coverage & sources
Jurisdiction, language, source type and licence are listed item by item. Unlicensed sources are not connected.
Updates & retention
Update frequency, retention period and deletion rules are registered per source. No single blanket promise is printed.
Evidence & scoring
Methods and scoring rules are explainable. Conclusions carry evidence links an analyst can re-check step by step.
Delivery route
Land one task layer first, then grow it into the network. Every step leaves something you can inspect.
Task pilot
Pick one of real-time event detection, disaster-disinformation response or influence-activity screening, and run it on real public data.
Deliverables: the pilot run report, with the detected-event list, evidence-chain samples and the error record.
Graph build
Expand sources and entities on the pilot pattern, with role partitions and audit live from the start.
Deliverables: the graph asset pack: entity and relationship inventory, source registry, permission partition table.
Platform supply
Workspaces, APIs and scheduled awareness products run side by side, issued by agency mandate.
Deliverables: API documentation, the workspace account model, and sample scheduled awareness products.
Analyst operations
Agencies that need continuous watch can delegate daily monitoring and briefs. Formal judgements stay with agency staff.
Deliverables: the watch roster and briefing samples, and the judgement escalation flow.
- Deployment can sit on approved cloud, on premises or in a dedicated network, with single-tenant and customer-held key options.
- Platform build, data subscription and analyst operations can be contracted separately.
- The first engagement targets one task layer, then consolidates upward. No whole-network prime bid on day one.
Use & export limits
The stronger the capability, the earlier the limits get written.
Public-source scope
The network gathers public-source information. Every source is registered with its licence, coverage and update frequency, and data is supplied by agency mandate.
Use limits
The system is designed without facial-recognition template generation and without mass personal-surveillance functions. Personal data follows source licence and local law.
Export assessed per deal
Supply to a foreign government passes EAR, ITAR and destination-law review first. No review, no negotiation.
Judgement stays with the agency
The system supplies evidence and leads. Formal briefings and response decisions belong to the client agency.
Questions, answered
Do you handle classified information?
No. The classified layer is out of scope. The network gathers public-source, non-classified information only. Where a classified layer is needed, it belongs to licensed partners; this system neither connects to it nor retains classified data.
How far can a conclusion be traced?
One step, back to the original text. Entities, relationships and events each carry their source, so an analyst can re-check every conclusion in the workspace, and the scoring methods and rules stay explainable.
Are the sources lawful to use?
Every source is registered under its licence, or it is not connected. Jurisdiction, language, source type and licence are set out in the data asset card before work starts, with update frequency and retention registered per source.
Can different departments see different data?
Supply is partitioned by agency, role and task. Partitions go live during the graph build, exports and access are logged end to end, and anything outside a mandate stays invisible.
Does the system make the judgement call?
No. It supplies evidence and leads; the agency decides. Formal briefings and response decisions belong to the client agency. Analyst operations cover daily monitoring and briefs only.
Where does an engagement start?
With one task layer, run as a pilot. Choose real-time event detection, disaster-disinformation response or influence-activity screening, prove it on real public data, then grow the network on the pilot pattern.
Where to next
Trust & compliance
Deployment models, data sovereignty, security operations, supply-chain screening and use limits, in one place.
MIANGAI active defence
Intelligence feeds it, defence acts on it. Both run on one situational core.
Hazard & public safety warning
The same event detection, pointed at emergency agencies.
Start with one task pilot
Send the agency mandate, the regions watched and the current source list. The reply covers pilot scope and the data asset card.
