Loading
WTI Crude–$/bbl|Brent Crude–$/bbl|Henry Hub Gas–$/MMBtu|US Rig Count–rigs|WTI Crude–$/bbl|Brent Crude–$/bbl|Henry Hub Gas–$/MMBtu|US Rig Count–rigs|
Deep Tech

Built for the
Wellbore.

Greenstone's technology stack is purpose-built for oil and gas, not adapted from enterprise software. Every component is designed around the latency, reliability, and precision demands of real-time drilling operations.

See GeoMaster LiveSecurity & Compliance →
Go / WASMReactDjangoFastAPIClaude on AWS BedrockMCPWITSML 1.4.1.1Cloudflare
Core Platform

DataEngine

Wellbore data processing and drilling calculations

DataEngine is the calculation layer behind every GeoMaster workspace: a Python library that turns raw well data into checked, unit-aware results, with a unit-of-measure registry, schema-driven parameter validation and a contract system that rejects bad data before a number reaches the screen.

Python
Library
v0.17
Current version
Typed
Units and schemas
Capabilities
Data ingestion
WITSML 1.4.1.1, LAS 2.0 and CSV, with units converted through a typed registry
Trajectory and plan
Minimum-curvature trajectory with tie-in, distance to plan, bit projection and required steering
Drilling calculations
Drilling mode, MSE, d-exponent, ECD, mud window, trips and gas analysis
Physics-based kick detection
Primary indicators (pit gain, mud weight delta, ECD drop) and secondary trend changes in SPP, ROP, torque, d-exponent and gas. An ML layer is in development.
Calibration
Bayesian tuning of calculation thresholds against real well datasets
AI Core

GeoEngine

AI assistant inside every GeoMaster project

GeoEngine answers questions from a project's own data (wells, trajectories, logs, tops, GeoMatch interpretations and DrillTracker plans), prepares actions the user confirms, and reports on the field from public sources. It runs as its own service next to the GeoMaster backend, and every read runs as the signed-in user.

Live
In GeoMaster
2
Agents
MCP
Tool protocol
Capabilities
Geosteering agent
A supervisor agent with a geosteering playbook and around 50 tools, calling one tool per step and checking every number against the calculation engine
Field intelligence
A wellsite geology agent that searches oil and gas sources for the well's field and keeps only claims a source supports
Human in the loop
GeoEngine never writes on its own: new wells, GeoMatches, versions and imports go through a dialog or a prefilled form the user submits
Claude models
Claude Sonnet for interpretation questions and Claude Haiku for everyday ones, with prompt caching to keep each answer affordable
MCP tools
Every tool group is an MCP server, and an MCP gateway lets users reach GeoEngine from Claude and ChatGPT connectors
Infrastructure

WASM Runtime

The same calculations in the browser and on the server

GeoMaster's geosteering calculations are written in Go and compiled to WebAssembly. The browser runs them for instant interaction, and GeoEngine runs the very same build on the server, so every number it reports matches the screen.

Go
Source language
WASM
Runs in browser
1
Build for app and AI
Capabilities
Trajectory
Minimum-curvature trajectory and positions along the wellbore
Layers and tops
Layer lines, formation tops and segments for the geosteering section
TVT
True vertical thickness and stratigraphic position from the interpretation
Model and stratigraphic logs
Model logs, stratigraphic logs and the linker that ties them to the well
One build everywhere
The same compiled module serves the web app and GeoEngine, so there is one source of truth for every calculation
Roadmap

ML Pipeline

In development

Machine learning is the next layer on top of the physics. It is designed and in development, not yet in production, and we will say so plainly until it ships.

Designed
Pipeline
Physics
First
Roadmap
Status
Capabilities
Kick detection ML
Anomaly detection first, then supervised models, with driller confirm or dismiss as the labelling mechanism
GeoMatch refinement
Automatic refinement of geosteering interpretations, building on the calculation engine
Anomaly detection
Continuous checks on incoming data to flag problems early
Human validation
Every model output will stay a suggestion that an engineer or geologist confirms
Infrastructure

Where it runs

Each part runs where it fits best: the websites at the edge, the platform and AI models on AWS, and GeoEngine on Oracle Cloud.

Websites and edge
Cloudflare

Our websites, DNS and edge security run on Cloudflare

Platform
AWS

The GeoMaster backend and its database run on Amazon Web Services

AI models
AWS Bedrock

Claude models are called through AWS Bedrock

AI and calculation services
Oracle Cloud

GeoEngine and the server-side calculation engine run on Oracle Cloud

Calculations
WebAssembly

The same compiled engine runs in the browser and on the server

Access
Per-user

Every request, including GeoEngine's, runs with the signed-in user's permissions

Ready to see it in action?

GeoMaster is live and processing real wells today. Request a technical demo with our engineering team.

Request DemoSecurity Overview