USING ARTIFICIAL INTELLIGENCE AS A TOOL FOR CRITICAL ANALYSIS AND SUPPORT OF SOFTWARE SYSTEM DESIGN

Abstract

The article investigates the use of artificial intelligence (AI) as an instrument for critical analysis and decision support in the process of software system design. Modern software systems are characterized by high complexity, scalability requirements, and intensive integration with external services, which significantly increases the risks associated with architectural and design decisions. Traditional approaches to system analysis and software architecture design largely rely on expert experience, which may lead to subjectivity, overlooked dependencies, and inconsistencies in requirements.

The study focuses on the application of machine learning techniques and large language models (LLMs) for analyzing technical documentation, user stories, functional and non-functional requirements, UML diagrams, and architectural descriptions. AI-based tools are examined as a means of identifying ambiguities, conflicts, and incompleteness in requirements, as well as predicting architectural risks and potential technical debt at early stages of the software life cycle.

A comprehensive case study is presented based on the design of a medium-complexity web-oriented CRM system. The research covers the full initial design cycle, including requirements analysis, architectural modeling, and evaluation of alternative architectural patterns such as microservices, CQRS, and event-driven architecture. Natural language processing models were used to analyze textual requirements and classify them by priority, business value, and risk level. The results demonstrate that automated AI-based analysis enables the detection of a significant proportion of potential defects and inconsistencies before implementation begins.

The findings confirm that artificial intelligence can effectively support system analysts and software architects by reducing subjectivity, improving the quality of design decisions, and mitigating long-term architectural risks. The article concludes that AI should be considered a complementary analytical tool that enhances expert judgment rather than replacing it. The proposed approach contributes to improving the reliability, maintainability, and scalability of software systems and outlines directions for further research in integrating AI tools into early stages of software engineering processes.