The holding "Rosel" presented an industrial AI assistant.

      The state corporation Rostec, through its holding "Rosel," announced the launch of a new artificial intelligence-based platform called ShokinGPT. The solution is designed to work with documents and data within enterprises. Developers claim that the system can reduce the time spent searching for information by 80% and save each specialist between three to five working hours a day. Moreover, ShokinGPT operates entirely within the organization's IT perimeter and does not require internet access. The technology has already received a registration certificate from Rospatent.

      How the system is structured

      ShokinGPT is not built as a single universal service, but as a set of specialized digital assistants, each responsible for its specific task. Some modules prepare documents and regulations, others process technical support requests, and others work with nomenclature and analytics. All of them are integrated into a common platform. The current version already includes tools for processing texts, PDFs, and spreadsheets, as well as knowledge bases. The system can transcribe audio recordings with speaker identification, create diagrams, and analyze programming code. Developers promise to gradually expand the functionality without changing the basic architecture.

      Initial results and feedback

      The platform was created by the scientific and production enterprise "Istok" named after A.I. Shokin, which is part of "Rosel." Company employees are already using the new system. The idea is that it reduces the workload on departments where a lot of time is spent on repetitive operations. For example, ShokinGPT is already helping to automatically distribute technical support requests and standardize nomenclature for procurement. This, as they say in the holding, improves data quality and saves labor costs.

      Vitaly Alexandrov, the director of digital transformation at "Rosel," emphasizes that their task was to create a tool for the practical needs of industry and government structures, rather than another general-purpose chatbot. According to him, the system takes into account industry terminology, internal regulations, and the specifics of managing industrial assets.

      In the future, ShokinGPT is planned to be applied much more broadly. The platform aims to be integrated into engineering and production processes. Plans include teaching the system to work with design and technological documentation, assisting in the preparation of production materials, and participating in the automatic selection of solutions for production planning systems. A separate direction is the analysis of telemetry from equipment. It is expected that AI will be able to identify signs of possible failures in advance, predict equipment performance, and help make decisions based on the accumulated data of the holding. So far, as is often the case with such announcements, the most interesting promises are for the future, but the first steps have already been taken.