From Generative AI to Agentic AI: A Paradigm Shift

Summary

AI models have evolved rapidly and are now not only capable of solving discriminative problems and generative requests by understanding natural language instructions, but are also capable of making decisions and executing actions with increasing autonomy, giving rise to the concept of agents. With the increasing complexity of the use cases being solved, two concepts naturally arise: specialization and orchestration of agents, and this is where agentic AI comes into play. The seminar will introduce the essential concepts of agent construction: instruction engineering and the configuration of the tools that agents invoke; as well as the decision-making, routing, delegation, and coordination mechanisms of these solutions as indispensable components of the various existing orchestration patterns. It will also review the different technological alternatives available for developing multi-agent solutions and their distinguishing characteristics, from options that do not require code and allow rapid development of personal or departmental productivity solutions, to code-oriented options that support complex solutions with enterprise-level scalability. Date: March, 13th, 2026.