Flai · 3D Spatial Intelligence

AI that understands the 3D world.

One AI engine, nine models — semantic mapping across aerial, mobile, national, utility, forestry and defence. Point clouds in, decisions out: on the cloud, on your GPU, or air-gapped on your hardware.

Pretrained model domains
9
Throughput
1M pts/sec
Parallel workers

Trusted by teams mapping the world

AeroGraphics
Aero Toyota
Bluesky
Cartodata
Dewberry
Republic of Estonia, Land and Spatial Development Board
European Space Agency
Flycom Technologies
Forsite
Helimap by Sixense
Nakanihon Air
New Compass
NM Group
Photomap
Pickett, an ESP Company
Propeller
SAM
SIGLA
Visimind
WGI
Woolpert
Kucera International
Socarto Mapping
Barr Geospatial Solutions

Supported by

EUDIS — European Defence Innovation Scheme
NATO DIANA — Defence Innovation Accelerator for the North Atlantic
Raw data in, decisions out

See your world classified in minutes.

Feed in raw 3D point clouds — from any sensor, on any platform. Flai's AI hands back a clean, segmented, typed picture of the terrain and everything on it. No manual labelling, no guesswork.

Raw 3DPC · unclassified, unprocessed
Processed · segmented, typed objects
Why Flai

Stop spending days on what AI does in minutes.

Flai replaces repetitive manual classification with automation — so your team focuses on analysis, not data prep.

Operational efficiency at scale

Process massive point cloud datasets in minutes, not days. Batch process hundreds of files with repeatable workflows and zero manual bottlenecks.

<3minper km² at 16 pts/m²

Coverage nothing else offers

A broad library of object classes out of the box, many that automated tools could never reliably detect before. Need something specific to your work? Additional retraining is available for your own categories.

60+object categories

Repeatable, auditable results

F1 scores of 0.98+ across key classes. Consistent output regardless of operator — every project, every time. Fully auditable.

0.98+F1 accuracy
Capabilities

Built for professional geospatial workflows.

Nine pre-trained AI models, 180+ processing nodes, a visual workflow builder. Everything you need from raw LiDAR to analysis-ready deliverables.

29 classification categories

Ground, buildings, vegetation, water, bridges, powerlines, and more — the most comprehensive LiDAR classification out of the box.

180+ processing nodes

Readers, writers, filters, rasterisers, vectorisers, classifiers. Build any workflow with drag-and-drop in the visual editor.

Visual flow builder

Drag-and-drop editor for processing pipelines. Pre-built templates for common use cases. Set up once, run on every project.

Online 3D viewer

Visualise billions of points in your browser. Measurement tools, classification colouring, interactive editing, region isolation.

Batch processing

Process hundreds of files in a single run. Automated queuing, progress tracking, error handling for production-scale workloads.

Custom model training

Need better results on specific terrain or vegetation? The retraining service creates models tuned to your data and environment.

The models

Nine models, one engine.

Every vertical is served by a dedicated model, all powered by the same FlaiCortex engine — shipping today, in beta, or on the near roadmap.

Compatible hardware

Supported sensors.

Flai works with the LiDAR sensors you already fly. Airborne platforms, mobile mapping rigs, handheld scanners — agnostic to manufacturer, scan pattern, and density.

Sensor manufacturer 1
Sensor manufacturer 2
Sensor manufacturer 3
Sensor manufacturer 4
Sensor manufacturer 5
Sensor manufacturer 6
Sensor manufacturer 7
Sensor manufacturer 8
Sensor manufacturer 9
Sensor manufacturer 10
Sensor manufacturer 11
Sensor manufacturer 12
Sensor manufacturer 13
Sensor manufacturer 14
Sensor manufacturer 15
Testimonials

What customers say.

As GIS experts for more than 20 years, we recently discovered the power of Flai’s AI platform. It was used for ultra-fast and convincing point cloud classification for damage mapping after a terrible storm severely hit one of our main cities. We had urgent need for this data to manage this disaster and Flai’s team delivered to our entire satisfaction.
Marc RiedoHead of GIS · State of Neuchâtel — Switzerland
Using Flai cut down the time for our manual point cloud classification by a factor of four — radically improving our turn-around and project profitability.
Hal WoodingManaging Director · Wooding Geospatial Solutions Inc
Flai’s automated, accurate vegetation classification seamlessly integrates into our workflow, providing the precise data we need to make informed maintenance decisions for our power network.
Aleš ZupančičTechnical system expert · ELES
Get started

See it work on your data.

Upload your LiDAR and get classified results in minutes. No credit card. No commitment.

Free tier · 10,000 processing units / month · no GPU required