Proceedings of the International scientific and practical conference ―Cambridge Science and Education Conference‖ (May 15-17, 2026) / Publisher website: www.naukainfo.com. - Cambridge, United Kingdom, 2026. - 429 p.

124 COMPUTER AND SOFTWARE ENGINEERING UDC 004.415.53:004.81 Kutaiev Serhii Valentynovych senior lecturer Lviv State University of Internal Affairs Lviv, Ukraine DYNAMIC VISUAL REGRESSION ANALYSIS IN DESKTOP SYSTEMS USING AI AGENTS Abstract. The maintenance of a consistent user interface across diverse desktop environments remains a significant challenge, especially when dealing with operating system-level rendering variations between Windows and macOS. Traditional pixel- based visual regression methods often fail due to anti-aliasing differences, font rendering discrepancies, and minor graphical updates that do not impact functionality. This paper proposes a methodology for utilizing AI Agents to perform dynamic visual regression analysis in desktop systems. The approach leverages computer vision and multimodal large language models (LLMs) to distinguish between insignificant cosmetic shifts and critical UI defects. By integrating AI Agents into Appium-based automation frameworks, organizations can achieve a higher level of visual verification that adapts to complex layouts and dynamically generated content, significantly reducing the maintenance overhead associated with visual test suites. Keywords: visual regression, AI Agents, computer vision, UI consistency, visual verification, multimodal LLMs.