ANO NTSK ISU has developed two AI-powered services designed to automate ERP system management tasks. These tools focus on generating operational documentation and building conceptual data models, replacing manual labor previously performed by analysts.
The documentation service reduces preparation time from weeks to hours. During testing, it processed 43,000 files, including 69 documents and 185 directories, to produce a 328-page user manual. It performs six automated checks, including compliance with GOST standards.
The conceptual modeling service cuts development time from six-eight weeks to approximately one week. It uses the BM25 algorithm for data retrieval and incorporates four expert control points. The system has already delivered its first model for the Treasury functional block.
Both services operate within the client's closed environment and are accessible via a web interface. They utilize a hybrid architecture combining Python code with local language models, ensuring that all results are traceable to original source materials.
