Researchers at MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) introduced JAZ, a minimalist agent framework that scored 70% on the StuLife long-term recall benchmark versus 62% f…
A raw-conversation-turn memory system using a single Postgres table and three MCP search tools scored F1 = 0.665 on the LoCoMo benchmark with Claude Sonnet (full 10 samples), beating Omni-SimpleMem's …
Researchers at MIT CSAIL led by Zhening Li, Omar Khattab and Armando Solar-Lezama built JAZ, an agent framework whose only primitive is an LLM-backed `invoke` function that lets the model write arbitr…
Researchers built JAZ, a minimal LLM agent framework whose single primitive, invoke, lets the model write arbitrary executable code including recursive invoke calls, and reported that JAZ invoke outpe…
A developer built a local, dependency-free system for searching AI coding-assistant chat transcripts without a vector database, using compact per-chat extracts, IDF keyword ranking with first-message …
A new benchmark called RENDER, introduced in an arXiv paper (arXiv:2608.23568v1), shows that the format of reader-facing memory artifacts can swing LLM memory evaluation scores by up to 72.6 points. O…
Researchers from Shanghai Jiao Tong University and Ant Group have demonstrated a new attack technique called InjecMEM that lets hackers plant hidden instructions in an AI agent's memory with a single …
A developer introduced MemoryBench, a modular benchmark suite for rigorously evaluating AI agent memory systems beyond marketing claims. The suite includes four core tasks—factual recall, temporal rea…
AI agent memory has become a crowded corner of AI infrastructure funding in 2025 and 2026, with startups like Letta, Mem0, and Zep building persistent memory systems on vector databases and retrieval …
Researchers introduced AgentMemBench, a benchmark evaluating five long-term memory strategies for conversational AI agents across three datasets and 491 annotated question turns, finding that external…
A developer explains that LLMs are stateless and that the intelligence of a long-running agent lives entirely in the harness that manages memory injection. The post introduces Smriti, an open-source t…
AI agent memory architecture splits information into working memory (context window) and long-term memory (external stores), enabling stateless large language models to persist and recall data across …
TRACE, an open-source Python library for self-organizing long-term memory in LLM agents, was released on PyPI. It organizes conversation history into a hierarchical B+Tree to enable efficient, multi-p…
Developer Simon Willison built agent-historian, a tool that lets AI coding agents query their own session history, after repeatedly encountering agents that forget solutions from previous sessions. Th…
OpenClaw-memgpt, a new plugin for the OpenClaw agent framework, implements MemGPT's three-tier memory architecture—core, recall, and archival memory—to provide persistent, searchable memory across ses…
Independent researcher Haijun Wen of Light Ark Technologies is seeking a cs.CL endorser on arXiv to post a preprint on a model-agnostic framework for persistent AI personality, addressing memory and p…
Memory systems in AI assistants are evolving beyond the short-term versus long-term split, with OpenAI and LangGraph SDKs advocating for a three-tier stack of working memory, durable state, and retrie…
Researchers have developed LANTERN, a lightweight memory layer that recovers facts lost when large language models compress long conversations, achieving 78.3% recovery of verifiable facts without req…
A developer has found that AI agent memory systems optimized for retrieval accuracy can produce more unsafe actions than less accurate retrieval strategies, because relevance and authority are separat…
Large language models reset context with every new conversation, forcing developers to repeatedly re-explain preferences, codebases, and project constraints. Persistent context windows, retrieval-augm…