Just-in-Time Memory: Learning to Curate Task-Adaptive Memory for LLM Agents
Yefan Zhou, Yang Li, Zeyu Leo Liu, Semih Yavuz and Shafiq Joty (Salesforce AI Research) propose Just-in-Time Memory (JitMem), which stores raw trajectories and decides what to extract from them only when a new task arrives.
Ask this paper
Write-time problem. Reflection, workflow and skill memories are distilled when a task ends, before the future query is known, which discards information and yields one summary for all later uses.
Read-time curation. Given retrieved traces and the new task, a curator writes a compact payload tailored to that task.
Direct training signal. Because the payload is used on the same task, the curator can be trained from immediate task success, avoiding delayed credit assignment across many later tasks.
Results. On ALFWorld, WebShop and tau2-bench, JitMem beats the strongest baseline by 16.2, 16.3 and 3.9 success-rate points.
Untrained curator. Even without training, read-time curation is competitive with or better than write-time baselines, and training adds further gains.
Abstract
Agentic memory systems reuse past experience to improve future performance, yet most existing designs curate memory at write time: once a task is completed, its trajectory is distilled into a fixed artifact, such as a reflection, workflow, skill, or reasoning strategy, that is later retrieved by similarity. This forces the system to decide what is worth remembering before the future query is known, irreversibly discarding information and producing a query-independent summary that must serve many possible downstream tasks. Learning such a write-time curator is also difficult because the value of a storage decision may only become apparent when a relevant query arrives, potentially many tasks later, creating a long-horizon credit-assignment problem. We instead retain raw trajectories and defer curation until read time, when the current task is known. Given the retrieved traces and the new task, a memory curator synthesizes a compact, task-adaptive payload tailored to the immediate need. Because this payload is consumed on the same task, the curator can be trained directly from immediate task success, avoiding delayed utility signals and the need to artificially group related tasks. Across ALFWorld, WebShop, and $τ^2$-bench, our Just-in-Time Memory (JitMem) consistently outperforms no-memory agents as well as heuristic and learned write-time memory methods, improving over the strongest baseline by 16.2, 16.3, and 3.9 absolute success-rate points, respectively. Notably, even an untrained curator is already competitive with or surpasses these baselines, showing that task-adaptive read-time curation itself is a major source of the gain; training the curator further compounds the improvement.