Key Takeaways

  • Lithology, structure, alteration, ore zone and geotechnical domaining are all built with implicit geological modelling in the same suites: Leapfrog Geo, Datamine Studio RM and Vulcan. The AI story for all five is largely the same tool wearing different hats.
  • Seequent’s Driver module is the clearest mainstream AI feature here. It clusters and classifies drillhole and assay data to speed up lithology and alteration-proxy modelling.
  • Structural modelling has a genuinely emerging research edge. A 2026 paper, Implicit Structural Modeling via Generative Diffusion Frameworks, uses diffusion models for complex fault geometry, but it’s not in commercial tools yet.
  • Alteration and proxy modelling increasingly leans on hyperspectral (SWIR) data run through scikit-learn-style classifiers before the results reach the modelling package.
  • Geotechnical domaining rides the same backbone, with ML regression on RQD and RMR data as an emerging (not yet standard) add-on.

TL;DR

To bring AI into implicit geological modelling, keep building your models in Leapfrog Geo, Datamine or Vulcan, but run your drillhole data through Seequent’s Driver first. Where you can, classify hyperspectral or automated-mineralogy data upstream, so better inputs reach the model. That is where most near-term AI gains live.

What Is Implicit Geological Modelling?

A mine geological model describes where each rock type, fault, alteration zone and ore body sits in 3D. Geologists build it from sparse drillholes, so most of the volume is inferred rather than observed. Older workflows drew every boundary by hand, section by section.

Implicit geological modelling instead treats each boundary as a mathematical function. The software uses radial basis function interpolation to generate continuous 3D surfaces between the data points. When new drilling arrives, the surfaces rebuild automatically. The same approach produces five model types: lithological, structural, alteration or proxy, ore zone and geotechnical.

How Do I Improve Geological Modelling With AI?

AI enters the workflow in three places, from mature to research-stage.

Building the base implicit geological modelling workflow

First, load drillhole intervals, assays and logging codes into an implicit-modelling package. Leapfrog Geo is the dominant tool, with Datamine Studio RM and Maptek Vulcan as the main commercial alternatives. A geologist always reviews the final wireframes. None of these are hands-off outputs.

Adding ML clustering with Driver and classifiers

Next, add the concrete AI layer. Seequent’s Driver is a cloud-based machine learning module that clusters drillhole and assay data into continuity-based domains. It flags likely vein zones, alteration boundaries and lithological contacts, with confidence scores attached. Run your data through Driver, then use its domains as a starting point in Leapfrog. For alteration models, many teams also train a scikit-learn classifier on SWIR hyperspectral data or geochemistry. Its alteration-proxy predictions then feed the model.

Watching research on structural modelling

Structural modelling is the one genuinely research-stage item here. The June 2026 diffusion-model paper handles fault intersections, branching and thrust nappes that trip up conventional implicit methods. However, it is published research, not a commercial feature. If your structural geology is complex, with multiple fault generations or overturned sequences, watch this space rather than deploy.

Where Implicit Geological Modelling Fits in Resource Estimation

The geological model is the frame for everything downstream. Its boundaries become the domains that geological domaining tests statistically before estimation. Mineralogy and processing behaviour then build on the same volumes, for example when you are predicting geometallurgical domains. As a result, errors in the geological model carry into every grade and recovery estimate.

Expect AI to save time rather than replace judgement. Driver speeds up interpretation, and classifiers sharpen alteration inputs. Nevertheless, the geologist still decides which contacts and faults make geological sense.

FAQs

What software is used for implicit geological modelling?

Leapfrog Geo is the dominant package, with Datamine Studio RM and Maptek Vulcan as the main alternatives. All three interpolate drillhole data into 3D surfaces. Seequent’s Driver adds machine learning on top of Leapfrog.

How does AI help implicit geological modelling?

AI clusters and classifies drillhole data into likely domains before modelling starts. This gives the geologist a data-driven starting point instead of choosing parameters from scratch. Classifiers on hyperspectral or geochemical data also improve alteration inputs.

Is implicit geological modelling fully automatic?

No. The software generates surfaces automatically, but a geologist sets the rules and reviews the result. Contacts, fault relationships and continuity still need expert judgement.

Can AI model complex fault structures?

Not yet in commercial tools. Research using diffusion models shows promise for fault intersections and thrust nappes. For now, complex structural models still rely on the geologist’s interpretation.

Try It With Geocluster

Juggling five interconnected model types, each pulling from different data sources and AI tools, is cross-referencing work that gets lost across separate software licences and spreadsheets. Geocluster gives you a single research harness to reason across your geological models. Read how the Geocluster Research Harness works to see how.