This paper presents a conceptual approach to transforming standard educational courses into a microlearning format with quest elements, inspired by cybernetic principles and the potential of large language models (LLMs). Modern digital education faces challenges associated with traditional content delivery methods that often fail to engage learners effectively and don't align with contemporary learning preferences for more dynamic, accessible, and personalized experiences.

This approach's foundation is a methodology for decomposing educational content into micromodules. Practical implementation includes FFmpeg-based tools for automatically segmenting video lectures into logically complete fragments. The structured sequence of learning materials in Moodle ensures logical connections between micromodules, forming a cohesive educational path with quest elements.

The innovative aspect is the conceptual integration of LLMs to enhance system capabilities. The paper examines prospects for using language models for automated assessment of student responses, generation of personalized explanations, and creation of recommendations for further learning trajectories.

This research represents the first stage in developing a comprehensive approach that combines automatic video content segmentation technologies with microlearning principles and the potential of LLMs to create more interactive and personalized educational trajectories.

Keywords: Microlearning, large language models, cybernetics, control loops.