Show HN: Open-source agent memory layer, 96% on LongMemEval, local-first
A developer shared a fixed harness on GitHub for comparing LLM agent memory systems, providing a standardized way to evaluate and benchmark different memory implementations. The tool aims to help practitioners make informed choices when building agents that rely on memory.
A developer shared a fixed harness on GitHub for comparing LLM agent memory systems, providing a standardized way to evaluate and benchmark different memory implementations. The tool aims to help practitioners make informed choices when building agents that rely on memory.
Show HN: SuperLocalMemory 4.0 "Governed Memory Operating System for AI Agents"
Show HN: SuperLocalMemory 4.0 "Governed Memory Operating System for AI Agents"
Memanto is an AI-powered memory agent that helps users store, organize, and retrieve personal information using natural language. It leverages LLMs to understand and index user data, enabling conversational search and recall. The tool solves the problem of scattered personal data by providing a unified, queryable memory.
Memanto is an AI-powered memory agent that helps users store, organize, and retrieve personal information using natural language. It leverages LLMs to understand and index user data, enabling conversational search and recall. The tool solves the problem of scattered personal data by providing a unified, queryable memory.
MindCache is an open-source memory system that gives LLMs persistent, structured memory across conversations. It helps developers build agents that remember context, user preferences, and past interactions, solving the problem of LLMs having no long-term memory.
MindCache is an open-source memory system that gives LLMs persistent, structured memory across conversations. It helps developers build agents that remember context, user preferences, and past interactions, solving the problem of LLMs having no long-term memory.
Knowl is an agent memory system that uses write-time supersession to manage information updates, achieving a score of 0.90 on the MAB benchmark. It solves the problem of stale or conflicting information in AI agents by ensuring that newer knowledge supersedes older data at the moment of writing.
Knowl is an agent memory system that uses write-time supersession to manage information updates, achieving a score of 0.90 on the MAB benchmark. It solves the problem of stale or conflicting information in AI agents by ensuring that newer knowledge supersedes older data at the moment of writing.