OmniMem: Autoresearch-Guided Discovery of Lifelong Multimodal Agent Memory

This paper introduces OmniMem, a unified multimodal memory framework discovered through an autonomous research pipeline, designed to significantly improve lifelong AI agent performance. It addresses the critical bottleneck of retaining, organizing, and recalling multimodal experiences over extended time horizons, demonstrating capabilities beyond traditional AutoML by focusing on architectural changes, bug fixes, and prompt engineering.