Platform architecture

The natural resource intelligence engine.

Thimar reads USGS MRDS, BLM MLRS/LR2000, Texas RRC & GAU, Rule 37/38 filings and satellite imagery, then discovers, scores and monitors natural assets before the market reprices them.

System architecture

Three layers, from commodity data to a defensible position.

Anyone can download the registries. The advantage is what happens after.

Layer 01

Commodity

Automated ingestion & vector geocoding

Connects to federal, state and county repositories through automated ETL. Normalises unstructured legal deeds, GIS shapefiles and geochemical assays into a single geospatial vector database.

Layer 02

Proprietary

Multi-modal scoring

Satellite imagery, production history and title records are evaluated together rather than in isolation, spatial cross-correlation, title curative validation and yield projection in one pass.

Layer 03

Defensible

Continuous monitoring & alerting

New filings, permits and surface activity arrive as alerts against watched parcels, so a position is re-priced when the record changes rather than on a quarterly cycle.

Fig. 1. Three-layer stack; depth marks specialisation.

Data sources

Public registries, read properly.

Five feeds, normalised into one queryable geospatial space.

Source	What it yields	Cadence
USGS MRDS	Geochemical surveys & core drill logs	QUARTERLY
BLM MLRS / LR2000	Federal mineral claim registries	DAILY
Texas RRC & GAU	Well production, spacing, Rule 37/38	DAILY
TWDB & Texas GCDs	Groundwater levels & transfer filings	MONTHLY
Satellite imagery	Surface activity & change detection	WEEKLY

GET /api/v1/parcels/VN-TR-4417/valuation

{
  "parcel_id":             "VN-TR-4417",
  "basin":                 "Vantrel Basin",
  "target_commodity":      "Lithium Brine",
  "geochem_score":         0.96,
  "assessor_baseline_usd": 2580.00,
  "maadin_fair_value_usd": 4200.00,
  "alpha_pct":             62.8,
  "title_status":          "UNENCUMBERED_FEE_SIMPLE"
}

Illustrative data. Not a real asset.

Developer surface

Every valuation is addressable.

Parcel-level scores, title status and fair-value estimates are exposed as REST and GraphQL endpoints for integration into existing land and ERP systems.

Asset classes

Four classes. One engine.

Each class is scored by the same pipeline, with models specialised to its data.

USGS MRDS

BLM LR2000

Texas RRC

TWDB / GCD

Satellite

ThimarScoring engine

Critical mineralsLithium · cobalt · REE

Oil & gas royaltiesPermian · Eagle Ford

Water rightsTX groundwater districts

Carbon marketsVoluntary + compliance

Fig. 2. Sources to engine to classes.

AI-driven mineral rights acquisition across lithium, cobalt and rare earths, aligned with US Inflation Reduction Act and EU Critical Raw Materials Act tailwinds.

Permian Basin and Eagle Ford royalty streams identified through well-performance modelling on Texas RRC public data.

Texas groundwater district monitoring, permit filings, usage trends and transfer activity across conservation districts.

Credit origination and portfolio intelligence for voluntary and compliance markets, modelled alongside subsurface rights on the same parcels.

The model

Royalties fund the next position.

Cash generated by acquired rights is reinvested. Platform subscriptions run alongside it.

01 DiscoverAI scores assets

02 AcquireRights secured

03 Cash flowRoyalties + SaaS

04 ReinvestLarger positions

$1BBy 2030

Fig. 3. The compounding cycle.

01 Discover

The engine scores undervalued tracts and rights against the public record.

02 Acquire

Rights and royalty streams secured at assessed-value baselines.

03 Cash flow

Royalty distributions and platform subscriptions generate recurring revenue.

04 Reinvest

Proceeds fund larger positions, the loop that compounds the portfolio.

Next step

See it run, or read the thesis.

Open the demos Request the data room

Request the seed data roomNDA required

Open the live demos3 prototypes

Work with Arshad2 slots