Typed ontology objects
Every point becomes a typed, queryable object — the shared 3D ontology across verticals.
FlaiCortex is the AI engine behind Flai — nine specialised models, each tuned to the objects and standards of its domain, turning raw point clouds into typed 3D objects and vertical deliverables.
Shipping today (GA), in beta, or on the near roadmap. We label maturity honestly — you always know what is production-ready.

Full-scene airborne classification
A single airborne-LiDAR capture, fully classified in one automated run — proven at national scale.

Cross-cutting cleanup
Isolated and building-edge noise detected and removed ahead of every downstream model.

Conductors, towers & insulators
Transmission & distribution corridors resolved down to insulators, guy wires and tension wires.

Roads & roadside detail
Dense, close-range mobile LiDAR classified to asset-inventory granularity.

3D situational awareness
Terrain and typed objects for GEOINT, OCOKA and JIPOE workflows — deployable fully air-gapped.

Through-water seabed mapping
Water surface, column and seabed — plus vessels, piers, cranes, buoys and breakwaters, all typed.

Stems, canopy & biomass
Forest structure resolved into ground, vegetation strata and individual tree stems.

Catenary system & track
Rail corridor classified into track bed, rails, catenary wires, masts and trackside infrastructure.

Industrial plant · scan-to-BIM
Industrial-facility scan classified into piping, HVAC, cable trays, beams, gratings and supports.
Every model does more than label points. The same engine vectorises and generalises the result into a typed 3D ontology and the deliverables your workflow actually consumes.
Every point becomes a typed, queryable object — the shared 3D ontology across verticals.
Building footprints, breaklines, centrelines, wire and catenary vectorization.
DTM / DSM, hillshade, CHM, slope and contours.
Aerial topographic objects · forestry single-tree inventory · powerline tower & attachment points.
Fine-tune FlaiCortex to your terrain, sensor or classification schema — three paths, supervised by senior AI engineers.
Send us your annotated data — or we annotate it for you. Our AI team tunes a module to your terrain and schema, delivered into your existing setup.
Fastest path · no ML expertise neededTrain on your own infrastructure with your own data. The full training pipeline runs alongside self-hosted Flai. Data never leaves your network.
Defence · NMAs · data sovereigntyInteractive point-and-click corrections in the web UI feed back into model improvement on your specific data — iterative, no code.
Continuous improvement over timeUpload a sample and get classified results in minutes — or book a call to scope a custom model.