SAMPLING, ANALYSIS & LAB ASSAY
verify outputs, especially in ambiguous cases. The systems should also be integrated with existing workflows to ensure the data feeds into geological modelling software, sampling protocols and exploration decision-making systems. And these models require ongoing refinement as the data source changes, with processes to monitor performance and retrain as new core arrives.
The industry is at an inflection point, according to Nitz, where ML4DrillCore systems have moved from research projects to commercial deployment, yet adoption remains uneven.
“ Many organisations are still manually logging core the way it was done 20 years ago – not because the technology is unavailable, but because implementation requires upfront investment in data standardisation and organisational change,” he says.
For operations leaders and chief geologists, the strategic question is no longer whether to adopt machine learningbased core analysis, but how quickly and comprehensively to implement it.“ The competitive advantage goes to organisations that redeploy their geological talent from routine classification work to strategic interpretation,” Nitz says.
Looking ahead, the next frontier for these trailblazers will be integration into wider decision-making environments.
“ When combined with hyperspectral scanning, X-ray Fluorescence( XRF) analysis, 3D geological modelling and real-time grade tracking, automated core interpretation becomes part of a seamless decisionsupport ecosystem,” Nitz says.“ Geologists working with these integrated systems will operate at a fundamentally different level – interpreting mineral systems, validating and challenging models, and driving exploration strategy with data density and speed previously deemed impossible.” through our‘ data frame’ to deploy the scanners in the field, we’ ve got access to all the instrumentation needed to carry out the scanning,” he said.
“ If you look at each of the sensors that we work with, they all have a different physical response, so the LIBS sensor from Lumo offers microsecond-style responses that allow you direct measurements of light elements like lithium and rare earths.” LIBS is a rapid, non-destructive technique that uses a high-energy laser pulse to vaporise a tiny spot on a rock, creating a short-lived plasma that emits light unique to the elements present when it cools. This provides immediate chemical information about the material, meaning faster geochemical insights directly at the source, enabling more efficient exploration and resource evaluation, according to the company.
“ This‘ needle point graph’ analysis is complemented with the two million metres of XRF scanning we have carried out on suitable elements, plus the findings we had with our other sensors,” Sanden said.“ That style of information is perfect for supplementing the other.”
GeologicAI’ s sensor stack‘ completion’ was part of the Lumo deal rationale, but the LIBS databank now available to the company will also prove tremendously valuable.
“ All of this data, when combined, is rocket fuel for AI,” Sanden said.“ The way we structure the data for clients, ensuring that there is co-location information as well as the required context – plus standardising the scanning and analysis process as much as possible – ensures they can utilise highquality AI outcomes to solve problems.”
While core scanning is where the company’ s experience currently lies, this problem solving is set to go way beyond that.
“ Our big focus this year is on our geometallurgical and geotechnical proof cases,” Sanden said.“ This ensures the data‘ loops’ around the value chain, providing value further downstream of where the original scans take place.”
Measuring the presence of valuable metals / minerals and assessing the rock type can be combined with scanning for deleterious elements that may affect downstream processing in many instances.
“ Doing a better job of interpretation at this stage of mining, plus carrying out relevant QA / QC practices, provides horsepower for the engineering side of the business,” Sanden said.
And this type of holistic analysis is becoming increasingly feasible due to the falling cost of carrying out these scans and rightsizing the data for machine-learning algorithms, he added.
“ Right now, it’ s very much about hitting these resources with everything as the incremental cost of scanning is very low once you already have a unit on site and have established the data processing flow.”
The company’ s aim – to help the mining sector make better decisions, faster – is being further helped by an ability to test out the AI-backed analysis within weeks of getting data from the first hole.
“ This is helping us get to mineral mapping, finding sulphides, automated lithology, automated domaining, etc in a fraction of the time that most people expect AI to start working,” Sanden said.
AI‘ rocketfuel’
GeologicAI believes it has exactly this type of integrated system at its disposal, following its recent transaction to acquire Lumo Analytics, a provider of the“ most compact and efficient” laser induced breakdown spectroscopy( LIBS) scanner in the market.
Adding LIBS to the company’ s RGB, XRF, hyperspectral and LiDAR sensor solutions cements GeologicAI’ s position as the“ only truly comprehensive and integrated source of sensor data for the critical minerals industry”, the company said in its press release.
Shortly after the deal was announced, IM spoke to Grant Sanden, CEO of GeologicAI, to add colour to this statement.
“ GeologicAI now has all the commercial sensors available for scanning and then,
The LIBS sensor from Lumo offers microsecond-style responses that allow you direct measurements of light elements like lithium and rare earths, according to Grant Sanden
74 International Mining | FEBRUARY 2026