Skip to content

Compete on the World Stage with tNavigator – Find out more about our Global Technology Summit Roadshow!

AI Assistant

AI-Powered Support for Reservoir Modelling

A secure, multi-agent AI assistant built into tNavigator — with a local MCP server so third-party AI tools can work with your projects too.

Request a Demo Request a Demo
Back to Software

AI Built into tNavigator

tNavigator’s AI Assistant is a secure, multi-agent AI assistant for engineers and geoscientists, built directly into the tNavigator interface and powered by a large language model. It answers questions in plain language, helps set up and troubleshoot models, and can write or edit Python workflow code on request – all without leaving the modules you already work in.

Because requests are processed on a local server inside your own network by default, AI Assistant can be used without sending data outside your organisation. In tNavigator version 26.1 and beyond, it also gained a Local Model Context Protocol (MCP) Server, letting third-party AI applications query tNavigator’s documentation and manage projects through the Model Context Protocol.

Context-aware, multi-agent chat

Built into every stage of the workflow – from seismic to surface network, AI Assistant understands tNavigator’s documentation and your live project, and can generate, edit and rerun Python workflow code on request.

MCP (Model Context Protocol) support

The Local MCP Server lets third-party AI applications interact with tNavigator’s knowledge base, search its documentation, and manage tNavigator projects via API functions.

AI-assisted troubleshooting

The AI Summary feature reads simulation calculation logs, explains warnings and errors, and recommends fixes – directly from Model Designer’s Cases and Jobs tabs, or from the Remote GUI.

Additional Tips and Information for Using tNavigator AI Assistant:

  • Chat history is saved across sessions, to quickly QC and repeat tasks.
  • Connects to external LLM (large language model) and ASR (speech recognition) servers via the OpenAI API protocol, with voice input supported for requests.
  • Can be invoked directly from the workflow Custom Code editor to write or edit code, and used as a step inside an automated workflow.
  • Runs on a local server within your own network by default, keeping requests and data inside your organisation.