Building an AI powered x-ray for the Earth’s subsurface.

Sparse data in, layers and traps out. The same model serves oil and gas, carbon capture and storage (CCS), and geological hydrogen.

JEPA inference Measured: three wells and their logs Inferred by JENNIFER

JENNIFER: Joint Embedding Neural Network Inference For Earth Resources

Summary

Between wells, the rock is unmeasured, new acreage is sparsely measured, someone still has to decide what is there before a well is drilled.
We are building a “World Model” of the Earth’s crust that can be queried like a system, not like an exploration survey.
It will be trained on open data from the Norwegian DISKOS database, the best public basin-scale database in the world.
Norway is the market entry point, selling model licenses that will work for subsurface intelligence needs in CCS, geological hydrogen, and oil and gas.
Ships as a licence pack and an API / MCP endpoint into existing subsurface tools.

50% of oil remains undiscovered in Norway (Sodir, 2026)
70 new offshore areas available for exploration in 2026 (Norwegian Ministry of Energy, 2026)
Expert Analytics offers market connections (e.g., Vår Energi meeting October) through their own clients and relationships with oil and gas industry in Norway
Total Addressable Market: $19B/year globally on exploration (Wood Mackenzie, 2026)
Serviceable Available Market: $3.5B/year exploration in Norway (Norsk Petroleum, 2026)

Pre-seed Ask: $1M

Founder’s published data fusion research on serpentinization/white hydrogen (Aiken, 2022; 2025; 2026)
Founder developed prototype model with multivariate zero-shot prediction on well quantities like seismic velocity and lithology (Aiken, in-review)