Mining Resource Estimation and Ore Characterization Copilot

AI solution for mining exploration and resource estimation that improves grade and tonnage estimation, supports prospectivity modeling, enables real-time mineral mapping from drill cores, and provides ore lithology soft sensing for downstream crushing decisions.

The Problem

“Mining Resource Estimation and Ore Characterization Copilot”

Organizations face these key challenges:

Sparse and heterogeneous subsurface data across assays, logs, geophysics, imagery, and plant systems

Long delays in manual core logging, mineral interpretation, and assay-driven updates

Resource estimates are difficult to refresh continuously as new drilling arrives

Exploration target prioritization is subjective and inconsistent across teams

Crushing circuits lack direct, real-time ore lithology measurements

Model outputs are often siloed from operational decision workflows

Uncertainty is not consistently quantified or communicated to decision makers

Impact When Solved

Improve early-stage grade and tonnage estimation with uncertainty-aware predictionsRank exploration targets faster using integrated geological, geochemical, geophysical, and remote sensing dataAutomate mineral mapping from drill cores using hyperspectral imaging and spectral classificationInfer ore lithology in near real time from crushing-circuit process signalsReduce turnaround time between data acquisition and geological or operational decisionsProvide a unified copilot interface for geologists, resource modelers, and plant operators

The Shift

Before AI~85% Manual

Human Does

•Review every case manually
•Handle requests one by one
•Make decisions on each item
•Document and track progress

Automation

•Basic routing only

With AI~75% Automated

Human Does

•Review edge cases
•Final approvals
•Strategic oversight

AI Handles

•Automate routine processing
•Classify and route instantly
•Analyze at scale
•Operate 24/7

Real-World Use Cases

Real-time ore lithology soft sensor for crushing circuits

An AI system watches crusher signals like motor current and speed to estimate what mix of rock types is entering the crusher right now.

Multiclass probabilistic estimation from time-series process signalsprototype validated on industrial crushing-process data with strong reported accuracy; positioned as deployment-feasible rather than broadly commercialized.

10.0

AI-based grade and tonnage estimation

AI estimates how much ore is in the ground and how rich it is, using exploration measurements and geological context.

Continuous prediction and uncertainty-sensitive estimationemerging; promising in literature but typically complementary to established geostatistical methods rather than a full replacement.

10.0

Mineral exploration prospectivity modeling resource hub

A curated toolkit showing how machine learning can help geologists combine many rock, chemistry, geophysics, and text clues to guess where valuable mineral deposits may be found.

Predictive ranking and multimodal pattern discovery for exploration targeting and subsurface interpretation.early-to-mid maturity enablement asset; it is a curated repository of applied resources rather than a packaged production system.

10.0

Real-time mineral mapping from drill cores using hyperspectral imaging

A hyperspectral camera scans drill core samples and software turns the images into near-real-time mineral maps, helping geologists see what rocks and minerals are present faster than manual inspection.

Computer vision classification and spectral pattern matching for mineral identificationdeployed applied workflow at institute level, indicating early commercial maturity for operational geological analysis.

10.0