long-term-memory

AgentMemBench is a systematic benchmark that evaluates five long-term memory management strategies for conversational AI agents across three datasets, finding that external key-value store retrieval dominates on quality but incurs a larger memory footprint.

An essay arguing that compression, not longer context windows, is the load-bearing primitive for long-term AI memory and relational continuity, drawing an analogy to compute and storage in computing.

This paper introduces IFCMemoryBench, a human-validated benchmark for evaluating long-term memory in LLM-based agents for BIM information retrieval. It shows that current memory systems achieve only 32.4% answer accuracy, revealing a domain-transfer gap in agent memory.

The HKU team open-sourced DeepTutor, a personalized AI learning companion with long-term memory, multiple knowledge base retrieval methods, and a skill ecosystem, capable of tutoring, problem solving, question generation, research, visualization, and learning path planning.

This paper introduces MemoryDecoder at Scale, scaling parametric long-term memory models to 6.9B parameters pretrained on 300B tokens, showing that independently scaling memory is more parameter-efficient than scaling base models alone.

Explores different long-term memory architectures for AI agents, with a focus on an agent-as-memory-controller approach using neon postgres.

MOSAIC is a structured, conflict-aware long-term memory framework for LLM agents that uses entity-typed graph storage, hash-accelerated retrieval, and active conflict detection to achieve high accuracy and efficiency on long-conversation QA and factual conflict detection tasks.

Introduces MemOps, a benchmark that reformulates conversational memory as lifecycle operations with structured traces, enabling operation-level diagnosis of memory failures in LLM-based agents, revealing that current systems are far from uniformly reliable.

ReflectWorld-MM is an entity-oriented multimodal memory system for open-ended video streams, using hierarchical long-term memory inspired by human memory theory, achieving state-of-the-art accuracy on six benchmarks.

Amazon has announced an agentic version of Alexa with long-term memory and integration with over 1,000 apps, enhancing its capabilities as a personal AI assistant.

NapMem is a framework that treats long-term user memory as a structured action space rather than passive retrieval, using a multi-granularity memory pyramid and reinforcement learning to train agents to navigate memory. Experiments show competitive performance on memory-intensive tasks.

A reproducible demo showing how adding long-term memory to CrewAI with Mem0 reduces token usage and latency via a fast/deep routing heuristic, with real measurements and a live dashboard.

Light-Omni is a multimodal agent framework for efficient video understanding that uses dual contextual states (global state and parametric latent state) to avoid iterative reasoning, achieving faster and more accurate processing with significant speedup and memory savings.

Describes the development of an AI that remembers, adapts, and becomes more useful over time, positioning it as a real partner rather than just an assistant or tool.

This paper introduces Profile-guided Personalized Retrieval Optimization (PPRO), a framework that enhances long-term conversational agents by incorporating user profiles into memory retrieval and optimizing retrieval via reinforcement learning, achieving consistent improvements over existing methods.

A developer discusses building a custom memory plugin for the Hermes agent using Engram, which reconciles new information with existing memories to avoid staleness and duplication, and asks the OpenClaw community about their memory usage.

EverMind team under Shanda Group founder Chen Tianqiao has open-sourced EverMemOS (EverOS), a brain-inspired intelligent memory operating system designed to solve the forgetting and context-limited problems of large language models by simulating human memory mechanisms, providing long-term memory infrastructure for AI Agents.

Cognee is an open-source AI memory platform that provides agents persistent long-term memory via self-hosted knowledge graphs, enabling recall and context across sessions.

An exploration of how AI agent memory systems often miss crucial cognitive processes like working memory, drawing parallels to anterograde amnesia, and offering design guidance for more effective solutions.

MEMPROBE is a benchmark that evaluates long-term memory in LLM agents by reconstructing hidden user states from the agent's memory after interaction.