Lamassu

Cities already operate thousands of cameras, acoustic arrays, air-quality monitors, vibration sensors and RF-capable infrastructure — continuous data streams currently unexploited for defence situational awareness. Today the practical options are purpose-built sensor deployments — costly, slow, geographically limited — or nothing at all. Commercial smart-city analytics favour breadth over defence-grade auditability: they typically lack signed provenance, deterministic enforcement, or a chain of evidence sufficient for operational trust.

The core problem is exploiting the sensor infrastructure that already exists, at defence-grade reliability and explainability, without new procurement — a sovereign capability Canada currently sources from no domestic supplier.

Site IFCs — one per service area — ingest directly from already-existing sensors through a signed connector model speaking ONVIF, MQTT, SensorThings, NGSI-LD, OPC-UA and DDS. Sensor-specialist micro-models across vision, acoustic, RF, spatial and environmental modalities propose candidate observations; a site coordinator composes them; and the site’s deterministic adjudication engine — never a model — commits the regional indicator.

A single Central IFC correlates across every site with larger models, fine-tunes the deployed roster, and holds corporate memory. If the reachback link is cut, a Site IFC keeps ingesting, inferring and committing under local authority, and reconciles when the link returns. Every committed indicator carries a machine-readable confidence and resolves, through a signed trace tree, to its source observations.

A portable Site IFC deployment profile — a containerized build of the compute tier for ruggedized field compute or a cloud-edge instance — targets operational readiness against local feeds within hours, not weeks.

Determinism over the models

Sensor specialists propose; a deterministic, non-model engine commits every indicator against a verified evidence chain. Detection behaviour is governed by versioned, empirically determined parameters — never an opaque model score.

The city as the sensor

A forward/rear fusion pattern recast for civilian infrastructure: regional commit authority at every Site IFC, partition-tolerant reachback to one Central IFC — and a floor that never moves: no Lamassu-owned hardware below the Site IFC.

Governed for civilian data

A deny-by-default privacy and policy framework built for civilian-source data, PIPEDA-aligned, with a signed chain of custody per inference — and a mission-agnostic governing framework that is Canadian-owned IP.

Fusion & adjudication

Validate the fusion and adjudication logic in a purpose-built five-layer simulation environment — and empirically determine the confidence-aggregation and threshold parameters the engine commits with.

The connector model

Prototype every target protocol against representative implementations — validating the connector model where real sensor APIs and firmware exceed the standard, the phase’s most significant technical risk.

Tier hardware & profile

Viability prototypes exercised on the reference tier hardware, and the portable Site IFC profile — the physically deployable configuration that anchors the witnessed demonstration.

Component 2 — build

Entering at TRL 6 with a build-ready design package. The build carries Lamassu through TRL 7 — a controlled pilot in an actual operational environment — to TRL 8: the technology qualified in final form against operational requirements.

One platform, many missions

Urban counter-UAS, critical-infrastructure anomaly detection and environmental early warning — and the same governed platform extends to summit security, public safety and allied urban operations, deployable across any city without new procurement.

Canadian-owned

IP is Canadian-owned and retained by the innovator, with Crown rights per the IDEaS contract terms; the sovereign-capability objective is served by the IP staying in Canada.