Turbo Cloud and MGRI Develop AI Assistants for Geologists
Geologists may discover oil, gas, or ore deposits, but before extraction can begin, they must prove that the reserves are real by submitting a comprehensive report to a specialized commission. These documents can run to hundreds of pages and contain complex calculations, regulatory requirements, and references to legislation.
AI solutions can now help prepare these reports. Turbo Cloud, a cloud service provider owned by Rostelecom, is to supply computing capacity, while Sergo Ordzhonikidze Russian State University for Geological Prospecting (MGRI) is to contribute its expertise in geology and subsurface resources. The two organizations have signed a three-year agreement.
AI Reviews and Explains
Turbo Cloud and MGRI are set to implement two main products. The first, NedraGPT, is an AI assistant that checks geological exploration reports before they are submitted to the State Commission on Mineral Reserves. For example, the system can identify contradictions in a document and explain which requirements have not been met and how serious the issue is. Geologists can then correct the errors, helping the report pass review on the first submission. NedraGPT draws on a database of 145 million books, articles, dissertations, and patents. Its capabilities cover nine fields, ranging from geology and geophysics to legislation and industrial safety.
The second algorithm is Selena AI, an office assistant. It processes scanned documents and tables, extracts the required figures, drafts letters and orders from templates, and tracks license deadlines. Tasks that would otherwise take specialists many days to complete can be handled much faster.
Geological exploration involves vast amounts of documentation, and errors in reports can carry significant consequences. The new technologies are therefore intended to help users correct inaccuracies and prepare responses to reviewers’ comments. Two deployment options are available: a cloud-based Software as a Service (SaaS) model or installation within the customer’s own IT infrastructure. Mining and subsurface resource companies will be the first to test the digital tools.
A Deep Layer of Information
Several AI applications are already being used in Russia’s mining industry, while others remain at the prototype stage. Rosneft has developed a service similar to NedraGPT. Called LitoText, it uses large language models to automatically read written descriptions of drill core, or rock samples, and extract 16 characteristics, including rock type, color, texture, grain size, and fracturing. Expert lithologists trained the model by teaching it how to interpret terminology correctly. According to the company, the technology has made analysis five to seven times faster while reducing the impact of human error. The tool is already being used to process new descriptions and check legacy databases.
MGRI is also developing its own AI project. The university has introduced Geointellect, a multi-agent platform built around an ontological core. The goal is not simply to digitize geological archives. Rosnedra’s collections contain 3.38 million archival items, but only about 12% have been digitized. The platform is designed to connect data based on their meaning. Specialized software modules handle maps, facts, and verification, helping prevent the AI from getting confused by the information it processes. Geologists will be able to connect a new data source in about an hour by moving blocks around in a visual editor.
Rosnedra has been developing a platform for collecting, storing, and analyzing geological information for several years. Its Digital Twin of Russia’s Subsurface integrates more than 120,000 state geological maps, information on 650,000 mineral deposits and occurrences, and 144,700 licenses. The data are accessible through web services. This provides the underlying infrastructure, since databases of this kind do not analyze information on their own, but they give industry-specific AI tools the data they need to work.
By 2027, Russia is also expected to deploy a unified digital platform for subsurface data that will support up to 30 AI agents for analyzing geological information. The platform is being built on the Federal State Information System for the Unified Fund of Geological Information (FGIS EFGI), maintained by Rosgeolfond.
Scope for Greater Efficiency
Russia has a substantial mineral resource base, and geological exploration requires specialists to process large volumes of historical data and regulatory documentation. Developing specialized digital tools is therefore important for improving industry efficiency by speeding up procedures, increasing staff productivity, and making more effective use of reserve data.
Turbo Cloud and MGRI are developing IT solutions at the intersection of language models, cloud computing, and specialized industry knowledge. For geological exploration, these tools offer a way to reduce routine workloads. For government agencies, they could speed up the processing of documents related to reserve assessments. For companies developing mineral resources, they could reduce the amount of staff time required.
