Tag
This paper presents a longitudinal evaluation instrument for agent memory that avoids label-error and contamination by generating facts before text, and demonstrates that short-horizon benchmarks can mis-rank memory architectures compared to long-horizon performance. It also releases the Veracium open-source library for agent memory evaluation.
This article discusses the design approach for AI memory systems, advocating for letting good memory systems emerge naturally from evaluation rather than designing memory architecture from the top down. The author argues that memory is a second-order effect evolved under pressure and proposes a longitudinal evaluation framework.