On December 2, 2025, a lecture titled “How to Build an AI Agent with LLMs: From Idea to Production” was held at the premises of Base42 in Skopje, co-organized by EuroCC MK – the National Competence Center for High-Performance Computing and Google Developers Group Skopje. The event attracted significant interest from both the academic and industrial communities, highlighting the relevance of agent-based systems and modern AI technologies.

The lecturer was M.Sc. Stefan Andonov, an assistant and researcher at FINKI and a member of the National Competence Center. He presented the process of developing AI agents based on large language models (LLMs) – from the initial idea, through model and architecture selection, to the deployment of a production-ready system.

The lecture covered the theoretical foundations of large language models, agent-based systems, and the software engineering required to develop an agent system, as well as a practical component (demo). In the practical part, examples of both a simple and a complex agent were demonstrated, including a multi-agent scenario aimed at automating a manual process that can take several hours and consists of data analysis, creation of visualizations and documents, and scheduling meetings with clients. These examples illustrated how AI agents can significantly improve workflows and reduce operational effort across various industries.

The event represents a significant contribution to the mission of EuroCC MK to enable knowledge dissemination and workforce development in the fields of HPC, machine learning, and advanced AI systems. Participants expressed strong interest in future activities, confirming the importance of continuous education in this dynamic field.