Plintir // Actionable intelligence // Interface visuals use synthetic data

Actionable intelligence data

Every source you hold, turned into intelligence you can act on.

Plintir connects classified, legacy, sensor and open sources into one governed mission model — then gives analysts, watch floors and commanders the tools to interrogate it, decide from it, and account for every decision afterwards.

10Products,
one mission model
PBScale of governed
mission data
24–72hAutonomy at
the tactical edge
Mission feed — all-source — evt/s
Objects 4,182,904Links 11,940,332 Resolved today 0Pending review 7

The difference that matters

Most platforms give you more data. Plintir gives you the next decision — and the evidence to stand behind it.

Raw collection is abundant. What is scarce is a picture that is current, sourced, releasable and specific enough to act on. Everything Plintir builds is measured against one question: did this shorten the distance between an observation and a decision someone can defend?

Where Plintir operates

Headquarters and the far edge run the same mission model.

Platforms built for one operating context leave everyone else with a read-only view of someone else's work. Plintir assumes all four contexts run at once, on one ontology, one policy engine and one audit trail — including when the network is gone.

CONTEXT A

Strategic

Headquarters analysis across the full historical picture — trend, readiness, capability and resource judgements built on years of data, not last week's spreadsheet.

Full corpus · Petabyte scale

CONTEXT B

Operational

Command and intelligence fusion on the watch floor. Live alerting, common intelligence picture, collection management and planning in one place.

Watch floor · Seconds to update

CONTEXT C

Tactical edge

Execution where connectivity fails. Local inference, pre-positioned mission packages and deferred synchronisation that never overwrites what it cannot see.

Disconnected · Days of autonomy

CONTEXT D

Coalition

Partner and inter-agency work where every object carries its own handling, releasability and need-to-know — enforced, not trusted.

Per-object releasability

The product suite

Ten products. One intelligence picture.

Each product solves a distinct problem and shares the same mission model, security policy and audit trail. Adopt one, or adopt the suite — nothing has to be re-modelled to add the next.

SOURCESNORMALISEONTOLOGY

PLINTIR NEXUS

Intelligence Data Fabric & Mission Ontology

Connects classified, legacy, streaming and geospatial sources into one governed mission model, so analysts stop reconciling spreadsheets and start answering questions.

Explore Nexus

NAI-07COMMON PICTURE

PLINTIR CITADEL

All-Source Intelligence & Mission Operations

The analyst workbench and common intelligence picture: search, link analysis, geospatial and temporal fusion, alerting and finished intelligence production in one place.

Explore Citadel

PROPOSED ACTIONHUMAN AUTHORITY AUTHORISED RETRIEVAL

PLINTIR AUGUR

AI Decision Intelligence

Retrieval, copilots and agents operating on classified networks — with authorisation applied before retrieval, citations on every claim, and a human approval gate on every consequential action.

Explore Augur

SENSOR-TO-DECISIONTRACK FUSION

PLINTIR SENTINEL

ISR Fusion & Sensor-to-Decision

Correlates sensor observations into tracks, fuses multi-source returns against the mission picture, and carries governed decision-support dossiers through review and approval.

Explore Sentinel

COLLECTION WINDOWCOVERAGE FOOTPRINT

PLINTIR CONSTELLATION

Space & Multi-Sensor Collection Orchestration

Matches collection requirements to satellite and airborne opportunities, submits and tracks requests, and routes returned products straight back to the analyst who raised the gap.

Explore Constellation

LINK DOWN LOCAL INFERENCE

PLINTIR OUTPOST

Tactical Edge Intelligence

Runs the mission picture, local search and on-device inference on a disconnected field node for days, then synchronises safely — surfacing conflicts instead of silently overwriting them.

Explore Outpost

EXPEDITIONARY NODERAPID EMPLACEMENT

PLINTIR VANGUARD

Expeditionary Ground Station & Deployable Node

Ruggedised, rapidly emplaced hardware and software that gives a deployed headquarters direct sensor access and full local analytic capability within hours of arrival.

Explore Vanguard

RELEASABLEPARTNER APARTNER B

PLINTIR ENCLAVE

Coalition & Cross-Domain Collaboration

Publishes a releasable subset of the mission picture to partners and other agencies under per-object handling rules, with complete dissemination audit and controlled cross-domain transfer.

Explore Enclave

CLOUDHIGH-SIDEAIR-GAPPED

PLINTIR BASTION

Accredited Sovereign & High-Side Hosting

Runs the suite in sovereign cloud, on-premises, high-side and air-gapped environments with separate key hierarchies, zero-trust controls and an accreditation path already walked.

Explore Bastion

BUILDSIGNSTAGEFLEET SIGNED DELIVERY

PLINTIR CONDUIT

Continuous Delivery & Fleet Assurance

Delivers signed, policy-gated software updates across cloud, classified and disconnected estates, with staged rollout, verified provenance and one-step rollback.

Explore Conduit

Core capabilities

What the platform does, in plain terms.

Twelve capabilities that show up in every deployment, whatever the mission and whatever the classification environment.

Every source, connected

Classified reporting, legacy systems, streaming sensors, imagery, documents and open sources arrive with their markings, reliability and lineage intact.

One mission model

Entities, observations, tracks, events, relationships and assessments modelled once and used by every analyst, application and model.

Entities resolved, not guessed

Eight records about one vessel become one vessel with eight sources — scored, reviewable and reversible.

Geospatial and temporal truth

Map, track, replay, geofence and correlate across space and time, with freshness and confidence visible on every symbol.

Sensor to decision

Observations correlate into tracks, tracks into assessments, assessments into governed decision support with a full evidence chain.

Collection that closes gaps

Intelligence gaps become collection requirements, matched to real sensor opportunities and tracked until the answer comes back.

AI you can defend

Authorised retrieval, cited output, visible confidence and a human approval gate on every consequential action.

Alerting that earns attention

Correlation rules over live streams produce priority alerts in seconds, with assignment, escalation and incident workflow attached.

Sharing with proof

Publish releasable slices to partners under per-object policy, with a dissemination record an inspector can query.

Operations that survive the link

Days of autonomous capability at the edge, honest degradation and conflict-safe synchronisation on reconnection.

Readiness in context

Supply, maintenance and movement constraints traced directly to the missions and plans they threaten.

Auditable end to end

Every derived value traces to its source and every action to a named person, months after the fact.

See it working

Live interfaces, synthetic data.

Three screens from across the suite, running in your browser on fabricated data. Full detail for each product sits on its own page.

Plintir · CITADEL · Common intelligence pictureAO GRANITE · 3 active tracks Synthetic data
NAI-07 · GEOFENCE ACTIVE FAC-0093 EVT-2288 OBS-9014 TK-4471 TK-2019 TK-7730

Tracks · 3 active

TK-4471AIR · UNK

Signals + imagery correlate
Entity: ENT-8890 · 4 s ago

0.87
TK-2019SURFACE

Transponder + imagery
Entity: VES-4471 · 11 s ago

0.74
TK-7730GROUND

Edge sensor NODE-11
Entity: unresolved · 2 s ago

0.41

Geofence alerts

14:02:11 · NAI-07 · P1TK-4471 entered geofence from west · watch officer notified

Unresolved stays unresolved. TK-7730 has no confident entity association, so it is shown that way rather than attached to the nearest plausible candidate to make the picture look tidier.

Plintir · AUGUR · Mission copilotPrivate on-premise model · domain-authorised Synthetic data
Summarise what changed in AO GRANITE over the last 12 hours, and flag anything that contradicts the 0400 assessment.

Advisory output · retrieval scoped to caller

Three material changes in the last 12 hours.

1 · Track TK-4471 was associated to ENT-8890 at 0912Z on correlated signals and imagery, confidence 0.87 OBS-2291.

2 · Vehicle count at FAC-0093 rose from 4 to 11 IMG-7714.

3 · CONTRADICTION — the 0400 assessment ASM-0088 placed the same entity 240 km south, citing one report whose source reliability was downgraded at 1120Z. Both cannot hold. I have not resolved this; an analyst should adjudicate.

▲ Proposed action · requires human authorisation

Create collection request CR-2291 against gap G-14 and route to the collection manager.

Policy check · PASSED at execution time
Authority required · Collection manager or above

Authorise Decline

Audit trail · this session

14:22:41identity

Query received · caller a.reyes · clearance attributes evaluated

14:22:41retrieval

Scoped to caller — 2,104 of 8,880 candidates authorised

14:22:42routing

Routed to private on-premise model, domain-authorised

14:22:51output

7 citations attached · sensitivity inherited from sources

14:22:52gate

Action proposed — awaiting human authorisation

The gate is architectural. The copilot can propose a collection request; it cannot create one. Authorisation is checked when the action executes, never inherited from the model's earlier reasoning.

Plintir · CONSTELLATION · Collection requirementsCycle 2026-08-15 · 7 active requirements Synthetic data
RequirementIntelligence gapPriorityEligible sensorsCoverageStatus
CR-2291Pattern of life, NAI-07 north approachP1SAT-A2 · UAS-114
82%
TASKED
CR-2287Vessel association, TK-2019P1SAT-C1 · AIS
64%
PARTIAL
CR-2280Facility status change, FAC-0093P2SAT-A2
95%
RETURNED
CR-2276Ground movement corridor, grid 44SP2UAS-114 · NODE-11
38%
PARTIAL
CR-2270Infrastructure survey, ORG-2205P2Withheld — compartment
NO ACCESS
CR-2264Throughput assessment, FAC-0121P3SAT-C1
71%
TASKED

Compartmentation without blind spots. CR-2270's sensor list is withheld from this caller, but the requirement stays visible — so the collection manager still knows the gap exists and can escalate it.

Mission outcomes

Twelve things teams do with Plintir on day one hundred.

Not features. The work itself — each of these is a thread we will run end to end against your own data during evaluation.

Fuse conflicting reporting into one defensible assessment

Reports, events, tracks and imagery-derived products resolve into entity dossiers with contradictions surfaced rather than averaged away, and an assessment that cites its evidence.

Hold a picture the watch floor actually trusts

A live geospatial and temporal view where freshness, confidence and source status are visible on every symbol, so nobody briefs stale data as current.

Turn an intelligence gap into a tasked sensor

Gaps become collection requirements, matched to real sensor opportunities, tracked to completion and linked back to the assessment that needed them.

Map a network and defend every link

Expand relationships from any seed, identify key nodes and communities, and show the observation behind each connection under scrutiny.

Compare courses of action honestly

Resource, risk, timeline and intelligence confidence scored side by side, with assumptions recorded and decision authority tracked.

Cut alert noise without missing the one

Correlation rules and confidence thresholds turn a flood of events into a short list a watch officer can actually work.

Share with a partner in minutes, not weeks

Publish a releasable subset of the picture under per-object policy, with dissemination recorded for later account.

Keep working when the link drops

Deployed teams run the mission picture, local search and inference for days, then rejoin without corrupting the enterprise record.

Exploit a satellite return where it lands

Direct downlink and local processing at a deployed node, with products routed straight to the requirement that justified them.

See a logistics failure before it becomes a mission failure

Shortages and maintenance constraints traced downstream to the tasks, plans and timelines that depend on them.

Answer a mission question with citations attached

A copilot that retrieves only what the caller is cleared for, cites every claim, names the gaps and proposes rather than acts.

Run an investigation that survives disclosure

Controlled cases with evidence chain, timeline correlation, review workflow and managed release.

Built for the whole floor

Sixteen roles, and none of them is "the user".

Each role gets a working surface built for its job and an authorisation profile that matches its clearance — including oversight, which can query the record without touching operations.

Intelligence analyst

Search and fuse sources, build dossiers, test hypotheses, publish finished intelligence.

All-source fusion analyst

Correlate multi-source reporting, tracks and incidents against source reliability.

Collection manager

Turn gaps into requirements, match sensors, track requests and returns.

Watch officer

Hold the picture, triage alerts, coordinate operational response.

Commander

Consume decision-ready views, compare options, review risk and assumptions.

Mission planner

Build plans, synchronise resources, map dependencies and contingencies.

Geospatial analyst

Analyse terrain, imagery-derived products, tracks, routes and time-series.

Counterintelligence analyst

Resolve entities, analyse networks, manage cases and controlled release.

Logistics analyst

Assess readiness, supply, maintenance, movement and mission dependencies.

Cyber analyst

Correlate indicators, infrastructure, identities and operational impact.

Data engineer

Connect sources, build pipelines, certify data products, own quality and markings.

Ontology engineer

Define mission objects, relationships, actions and policy-aware operations.

AI engineer

Register models, build retrieval and agents, evaluate performance, monitor drift.

Security administrator

Own identity, classification, compartments, transfer policy and audit.

Platform operator

Operate environments, deploy releases, manage deployed nodes and recovery.

Oversight and inspection

Query audit and dissemination records independently of operational systems.

The intelligence cycle, actually closed

Eight stages, and the one that gets skipped is always the last. Plintir treats consumer feedback as a first-class object, so collection priorities improve instead of drifting.

STAGE 1

Direction

Requirements, priorities, authorities, constraints and deadlines captured as objects, not email.

STAGE 2

Collection planning

Sources and sensors identified, coverage gaps quantified, opportunities matched.

STAGE 3

Collection

Authorised reporting, sensor returns, documents, media and feeds arrive with provenance.

STAGE 4

Processing

Parsed, normalised, geolocated, entity-extracted and enriched, with markings preserved.

STAGE 5

Analysis and fusion

Sources correlated, entities resolved, hypotheses tested, confidence scored, gaps named.

STAGE 6

Production

Assessments, dossiers, maps and briefs generated with citations and handling controls.

STAGE 7

Dissemination

Released under per-object policy to approved consumers, with a complete record kept.

STAGE 8

Feedback

Consumer response captured, requirements updated, collection re-prioritised. The loop closes.

Responsible AI in the mission

AI output is advisory until an authorised person accepts it. Plintir may correlate, summarise, forecast and recommend — consequential action requires a human.

This is an architectural constraint rather than a policy note. The authorisation gate sits in the action layer beneath every application and every agent, so there is no route to a consequential action that goes around it.

Retrieval before generation

Authorisation is applied before any document reaches a model. Nothing is retrieved and then hidden — it is never retrieved at all.

Domain-bound models

Classified prompts never route to an endpoint that is not authorised for the same handling level, regardless of how well that model performs.

Execution-time authorisation

Actions are checked when they execute. An agent's earlier reasoning is never treated as permission to act.

Derivative sensitivity

Generated text inherits the sensitivity of what it retrieved, and derivative review runs before anything is released.

Untrusted content stays untrusted

Retrieved documents, uploaded files and external content are treated as data, never as instructions to the agent.

No autonomous employment of force

Effects workflows are decision support with human approval throughout. The platform does not select, authorise or execute.

Engineered against known failure modes

The nine ways platforms in this category go wrong.

Published because they are predictable, and because any vendor who cannot name them has not delivered one. Each answer is an engineering commitment, not an operating procedure written after go-live.

RiskHow it shows upHow Plintir answers it

Over-centralisation

One global system becomes a bottleneck, or an unacceptable concentration of risk.

Federated domains, data products, explicit replication decisions, independent capability at the edge.

Classification leakage

Derived analytics or AI output combines sources and exposes what should have stayed closed.

Per-object labels, derivative policy, output checks, releasability review and negative testing.

Over-trust in AI

Model output is read as fact rather than as an advisory judgement to be checked.

Citations, confidence bands, visible source-versus-inference distinction, human approval, continuous evaluation.

Wrong entity merge

Two distinct entities are fused and quietly contaminate every product built on them.

Scored matching, evidence per merge, analyst review, reversible merges, source records preserved.

Stale picture

A pipeline or link failure creates confident-looking awareness that is hours out of date.

Freshness indicators, feed health, degraded-mode interface, alerting on stale critical sources.

Deployed node compromise

A captured or lost node exposes mission data or trust credentials.

Encryption at rest, device identity, short-lived credentials, remote revocation and bounded local caches.

Vendor lock-in

Mission capability becomes dependent on one model or one proprietary data engine.

Ontology and API abstraction, open storage formats, a model gateway and exportable data products.

Supply-chain compromise

A malicious or vulnerable update reaches classified or deployed estates.

Signed artifacts, bill of materials, staged rollout, offline verification and vulnerability gates.

Cross-domain bypass

Unofficial export becomes easier than the approved release route, so people use it.

Uncontrolled egress disabled, transfer service integration, approval workflow, inspection and audit.

Deployment and scale

Sized from the drivers that actually cost money.

Ingest volume, retention, relationship cardinality, streaming rate, concurrency, inference strategy, deployed fleet size and cross-domain replication — not user count alone. Figures below are planning envelopes, confirmed against your real source throughput during design.

ScaleNamed usersConcurrentIngestPrimary data Objects + linksStreaming
Mission cellSingle team or pilot100–30025–750.1–0.5 TB/day20–100 TB10–100 M5k–20k /s
EnterpriseJoint headquarters1,000–3,000150–5001–5 TB/day0.2–1 PB0.5–5 B50k–200k /s
Multi-domainNational or alliance scale10,000–25,0001,000–5,0005–25 TB/day2–10 PB5–50 B250k–1M /s

Performance you can hold us to

<2s

Entity search response, typical indexed query in authorised scope

<3s

Relationship expansion for a bounded network query

2–10s

Priority feed from platform receipt to map display

<5s

Priority alert generated from a qualifying event

3–8s

AI response begins streaming for common requests

99.95%

Core mission service availability target, monthly

24–72h

Deployed node autonomy without core connectivity

30 / 5 min

Recovery time and recovery point, critical operational state

How a programme runs

Security markings, provenance, audit and attribute-based authorisation are built in the first step. Retrofitting them once mission applications exist is the most expensive mistake available on a programme like this.

STEP 1

Foundations

4–8 weeks

Mission threads, data classification, security domains, authority model, core ontology and target service levels agreed before a line of integration work starts.

STEP 2

Data and ontology

8–16 weeks

Priority sources connected, catalogue and lineage established, core mission objects modelled, search live, attribute-based access in force.

STEP 3

Analyst capability

8–12 weeks

Workbench, mapping and common intelligence picture, entity resolution, timeline, alerting and reporting delivered to real users.

STEP 4

Governed AI

8–12 weeks

Model gateway, authorised retrieval, citations, evaluations, safe tool use, approval gates and red-team testing.

STEP 5

Mission applications

12–24 weeks

Collection management, planning, readiness, investigations, coalition sharing and role-specific views.

STEP 6

Edge and multi-domain

12–24 weeks

Deployed nodes, disconnected operation, signed updates, domain replication and cross-domain workflow.

STEP 7

Assurance at scale

Continuous

Load and failure testing, disaster recovery, accreditation, model assurance and operational exercises.

Start small, prove it fast

Bring three sources and one question you cannot currently answer.

Our first engagement is deliberately narrow: three to five critical sources, a compact mission ontology, analyst search and link analysis, geospatial and timeline, one operational workflow and one governed copilot. If it does not shorten a real decision, it does not scale.