Machine learning
Inspect the regional response and where two similar ML methods assign different relative rankings.
← MindSync work
LodeCraft by MindSync
LodeCraft by MindSyncExploration intelligence · Perth, Western Australia
Focus your next exploration decision.
MindSync uses LodeCraft, its internal scientific system, to analyse geological, geochemical, geophysical, drilling and spatial evidence. Your team receives ranked areas for review and GIS-ready maps, with the reasoning, uncertainty and data gaps visible. Methods can include machine learning when the question and evidence support it.
At Yamarna, the October study brings measured gold, magnetic context, a current ML response and candidate review areas into one inspectable record. Examine the evidence and next steps →
Based in Perth, Western Australia. Led by geologist Jacob Cichosz. GIS-ready outputs for your team.
01 / Mineral exploration data analysis
Evidence your team
can work with.
Each review priority arrives with its supporting evidence and its limits. Inspect these examples from the Yamarna case study.
- Spatial outputs
- Ranked review areas and maps for your geographic information system (GIS), with coordinate context and limits.
- Evidence report
- The question, sources, methods and reasoning behind each agreed review priority.
- Uncertainty and gaps
- Contradictory evidence, missing information and what could change the interpretation.
- Review presentation
- A walkthrough of the findings and the next investigations the evidence supports.
The October Yamarna package contains 97 checked QGIS layers, current candidate geometry and supporting evidence. Explore the current handover and its limits →
02 / Yamarna · October 2026
Historical data.
A fresh exploration view.
Current candidate areas overlap the recorded Bloodwood and Renegade South prospects. See the model views, documented drilling milestones and the next questions for review.
Measured evidence
Inspect gold values at sample locations and regional magnetic context. Keep measured observations distinct from model predictions.
Candidate review areas
See current geometry alongside named sites, source context and the limits of point-based comparisons.
Bynoe–Finniss · 28 September 2026
See the broader response.
Follow the evidence.
Five Bynoe maps tell the story from geological context to connected model responses and wider investigation areas. See how the workflow is designed to reduce repeated desk work, focus mapping and sampling questions, and hand over a QGIS package your team can keep using. Customer time and cost savings have not been quantified in this study.
See the Bynoe case study →03 / Working together
Your question.
An inspectable answer.
Your team brings the exploration question and available evidence. MindSync performs the analysis with LodeCraft and hands the work back to your geologists.
Working with MindSync
How an engagement starts.
Geologist Jacob Cichosz leads the scope, analysis and handover. We aim to respond to your first enquiry within two business days (Perth time).
Start with your question
Send a non-confidential summary of the area, commodity, available evidence and decision your team needs to make.
Discuss the fit
A free, no-obligation discussion explores whether LodeCraft could help. It does not include a completed data assessment or analytical report.
Agree how to share data
Before project files are exchanged, agree confidentiality terms and a suitable transfer method.
Review a written proposal
Receive the scope, fee, timetable and agreed deliverables. Timing depends on the question and the available evidence.
Approve before analysis begins
Paid analysis starts only after you approve the written scope. If the evidence is unsuitable, we may suggest a narrower question, data remediation or not proceeding.
Explore the analytical stages
Stage 01 / Expected outcome
Agree a responsible scope.
- Area of interest and exploration context
- Company and relevant public data
- Suitability, permissions and evidence gaps
Review your geological, geochemical, geophysical and drilling data, the project scope and the agreed handover.
Mineral exploration data analysis and target generation ↗04 / Rigour & professional limits
Keep the reasoning.
Keep the uncertainty.
A compelling map can still mislead. The analytical record needs to remain inspectable long after the initial recommendation.
01 / Validation that respects geography
Nearby training and test areas can overstate performance. Spatial validation must reflect the intended use area, and model challenge must examine leakage from clues connected to known occurrences.
02 / Unknown ground stays unknown
Unexplored ground is not proven barren. Evidence gaps, disagreement, novelty and abstention remain visible where relevant; one score must not hide weak support.
03 / A reconstructable decision record
Sources, transformations, exclusions, methods and run identity stay attached to the result. Client work is scoped as a distinct body of evidence for later professional review.
Learn how methods, spatial validation and uncertainty should be handled in Machine learning for mineral exploration: methods, validation and practical limits.
What the work supports
A narrower investigation.
Ranked areas for geological attention, supporting and contradictory evidence, and a documented basis for the next review.
What it cannot establish
Proof of an orebody.
Ranks are not probabilities of mineralisation. Discovery and commercial success are not guaranteed. Geological judgement and field validation remain required.
Start with your exploration question
Bring your next
decision into focus.
Send a non-confidential summary of your area, commodity and decision. We aim to respond within two business days (Perth time). The initial fit discussion is free and carries no obligation.