DAIR.AI describes a Salesforce AI Research paper, Just-in-Time Memory, that keeps raw trajectories and writes a short memory payload only after the next task arrives. It reports gains of 16.2, 16.3 and 3.9 success-rate points over the strongest baseline on ALFWorld, WebShop and tau2-bench.
The Jev-Mem paper proposes an agentic memory architecture in which a lightweight controller makes storage and retrieval decisions and an LLM is used only for final reasoning. It reports 6.6x faster memory construction, 36.7% lower query latency, and a LoCoMo score of 0.777.
WFM turns an LLM wiki into a graph and retrieves with query-conditioned message passing, so page text and the links between pages shape the result together. The team says a GPU-to-GPU training protocol trains 10.5 times faster and reports strong results on five agent-memory and multi-hop reasoning benchmarks.
