Legacy “Society of Mind” (Minsky) AI paradigms are again rapidly evolving for greater power and scalability, with security and interpretability as critical challenges. This topic is to bring together interested parties to review old and new Agentic Architectures research and introduce novel concepts and new directions.
Current state of the art especially involves mapping increasingly complex cybernetic-agentic loops (Sense=>Act (SA), Sense=>Plan=>Act (SPA), Observe=>Orient=>Decide=>Act (OODA), etc.) to GPU Data Center compute using CUDA, Slurm, LangGraph, and so on.
Case-based Reasoning (CBR) and Rule-based Knowledge Base (KB) integration to containerize Foundation Models are specific Agentic Architecture sub-topics. Much current work is already directly AI-driven (AI creating AI).