Product

Geological Intelligence

A dynamic lithostratigraphic lexicon, cross-survey geological correlation, and 2,000 metallogenic models translated into executable geoprocessing rules.

Core System

Dynamic Lithostratigraphic Lexicon

A continuously updated compendium of geoscientific knowledge extracted from decades of accumulated PDFs — articles, theses, technical reports. One record per geological unit, not per document.

  • Age, lithology, depositional environment, contacts, mineral potential — all with source and confidence score
  • When two articles disagree, the system detects and logs the conflict instead of overwriting
  • Multilingual framework: SGB/CPRM, BRGM, BGS and any geological survey
  • Foundation for AI systems: LLMs, exploration copilots, metallogenic analysis

Example: Serra Geral Formation

Official Lexicon (CPRM)4 paragraphs

Age: 135–97 Ma · Main lithotype: dacite

Infomineral Dynamic Lexicon102 articles
Conflict 1: Articles describe basalts (90%+ volume), not dacite
Conflict 2: Formation vs. Group — real debate, documented
Conflict 3: Ar-Ar dates constrain event to 127–138 Ma

Full traceability to every source document

Cross-Survey Integration

CPRM & BRGM Data Correlation

Brazil's SGB/CPRM uses structured fields. France's BRGM compresses everything into free text. LLMs bridge this gap at scale.

40%
French lithology processed
0.91
Average confidence
0
Failures/350 batches
16,000
Geological elements

Fields Extracted from Free Text

Main lithology
Secondary lithologies
Stratigraphic age (eon→epoch)
Sedimentary environment
Deposit type
Metamorphism
Magmatism
Tectonics

Before (BRGM raw)

"Ensemble carbonaté massif comprenant des calcaires bleus… Monts de Lacaune — Lower Cambrian"

After (structured)

Lithology: Limestone, dolomite
Age: Lower Cambrian
Environment: Carbonate platform
Region: Monts de Lacaune

Prospectivity

2,000 Metallogenic Models as Geoprocessing Rules

Theoretical metallogenic models transformed into executable spatial analysis rules by LLM agents.

1

Model Compilation

2,000 metallogenic models compiled via bibliographic research and RAG techniques, covering skarns, orogenic gold, porphyry systems, and more.

2

Rule Translation

LLM agents convert theoretical models into executable geoprocessing rules. Example: create 200m halo around microgranite, clip with carbonate rocks.

3

Spatial Execution

Sequential spatial queries: buffer → intersect → clip → rank. The model translates geological models into automated processing steps.

4

Validation

Results validated against known occurrences. Rules refined and expanded to new regions — South America and Africa.

GIS + AI

Natural Language GIS Queries

An interface for searching 1,594 South American mineral deposits using natural language. Translates questions to SQL, maps results, and generates contextualised geological analysis.

Example query:

"Where are the largest copper deposits in Chile?"

→ AI generates SQL · maps results · provides geological insights

Try the Portal →
SAmMD — 1,594 deposits Chuquicamata 68 Mt Cu · Porphyry

Need geological data integration?

Whether correlating surveys, building a lithostratigraphic foundation, or translating metallogenic models — we can help.

Discuss Your Project