ForceX AI

ForceX AI is Indonesia's first AI-for-energy platform. We build physics-informed AI agents that replace hours of numerical simulation with millisecond-level predictions for geothermal reservoirs, nuclear reactors, power grids, and oil & gas operations. Our flagship product GeoForce predicts reservoir temperature and pressure fields 282,350x faster than traditional TOUGH2 simulation.

GeoForce is a physics-informed CNN (57,802 parameters) trained on 1,000 reservoir simulations. It takes reservoir properties (temperature, permeability, porosity, depth, well locations) and predicts 20 years of thermal-hydraulic evolution in 3 milliseconds on CPU. It achieves R²=0.994 for temperature and R²=0.997 for pressure on held-out test data, with zero physics constraint violations. The model and training dataset are open-source on Hugging Face under Apache 2.0.

GeoForce is free for all users with no registration required — real CNN predictions, not a demo. The platform includes 12 AI agents across geothermal, nuclear, oil & gas, renewables, and general physics. Additional validated models are available on Pro tier.

ForceX AI covers 5 energy domains: Geothermal (GeoForce), Renewable Energy (GridForce, StorageForce, WindForce), Oil & Gas (SubsurfaceAI, DrillAgent, PipeGuard), Nuclear (NeutronX, ReactorMind, SimForce), and General Physics (PhysicsForce, PlasmaForce). Each product uses physics-informed models trained on real simulation data.

Yes. GeoForce CNN v1.1 model weights and the 1,000-simulation geothermal reservoir dataset are published on Hugging Face under Apache 2.0 license. The model card includes full training methodology, validation metrics, and a documented failure analysis of the v1.0 version.