Conference

We’re speaking at WTC Montreal

May 17–21

GroundedAI + Leapfrog Works

Move from underground capture to ground model without manual transcription. Lithos captures geotechnical conditions in 3D, Strata structures the dataset, Leapfrog Works imports it effortlessly for tunnel design risk, ground behaviour modelling, and infrastructure decision-making.

Decision anchoring

Leapfrog Works is where tunnel engineers

Leapfrog Works is where tunnel engineers and geotechnical engineering consultants build the ground model that design decisions depend on. Soil and rock units and groundwater conditions define what the infrastructure project will encounter and how it needs to respond.

The model is only as current as the data imported to it. When underground observations arrive days or weeks late, reformatted through manual steps, the ground model is already behind the excavation. Design validation runs against the wrong picture. Risk assessments rely on stale assumptions.

When the handoff from the field to the model is seamless, the model stays up to date with what's encountered.

The manual gap today

Underground data destined for Leapfrog Works

Underground data destined for Leapfrog Works follows the same fragmented path as any other modelling input: photos in folders, notes on paper, meshes processed in one tool and georeferenced in another. Geotechnical data, such as rock mass classifications, discontinuity sets, and ground conditions, get transcribed from field forms into spreadsheets before anyone in Leapfrog Works ever sees it.

For tunnelling projects, this delay has direct consequences. Support design decisions, RES reviews, and baseline (GBR) comparisons all depend on a ground model that reflects what's encountered.

The Workflow Stages

The data process

Four stages. One handoff. No manual translation.

Stage 1 - lithos

Capture

Location-based 3D scanning of underground conditions in less than five minutes. Offline-first. Accuracy in the 5-10 mm range.

Stage 2 - strata

Structure

The dataset syncs to Strata where it's georeferenced, mapped, and organized. Round naming and location hierarchies tie every capture to its project location. The geotechnical record is reviewed, validated, and prepared for export, structured so every observation links back to its location, date, and observer.

Stage 3 - Handoff Point

Export to Leapfrog Works

Strata exports georeferenced and mapped textured meshes as OBJ with their MTL material files, and 3D annotations as CSV with full attributes. No intermediate tools. No file cleanup.

STAGE 4 - LEAPFROG WORKS

Model in Leapfrog Works

Import directly into Leapfrog Works. Geotechnical layers, discontinuity data, and ground-condition observations arrive structured and attributed. Build or update the ground model from current underground data, not from last month's field notes.

How Lithos and Strata improve Leapfrog Works

Lithos

Lithos, for underground data capture

Lithos captures geotechnical conditions at the point of observation including rock mass classifications, 3D scans, and annotated HD images with time-stamps. The data that imports to Leapfrog Works carries its full field context.

Strata

Strata, the export pipeline

Strata is the structured project record where plan meets reality. For tunnelling projects, Strata organizes data captured underground by chainage, and exports to Leapfrog Works in compatible formats, preserving textured meshes, mapped annotations, and attributes.

What you get

Leapfrog Works-ready geotechnical datasets

Georeferenced textured meshes (OBJ + MTL), 3D annotations with full attributes (CSV), rock mass classification data, discontinuity geometry, and ground-condition observations.

Frequently asked questions.

Who is this Leapfrog Works workflow built for?
Can GroundedAI's datasets export to Leapfrog Works?
Does this connect to the RES meetings workflow?
Related Workflows

Used across key workflows.

See how structural mapping connects to the broader GroundedAI platform.

See the capture-to-Leapfrog Works data process end to end.

Walk through the full handoff with a tunnelling dataset. We'll show you what imports, what's preserved, and how fast the ground model catches up to the excavation.