AP Research: Study of LLM Memory in Agentic Tasks

Conducted a 90-trial quantitative experimental study comparing Stateless, Recurrent, and Retrieval-Augmented Generation memory architectures for LLM agents in a full Catan simulation.