www.anthropic.comThe components of agent memory \ Anthropic@eddie · nov 09Listen20:00Save1 saveAnthropic describes how Claude uses 'dreaming'—offline reflection on session transcripts—to organize memory notes, correct errors, and build wiki-like knowledge representations that improve performance on long-horizon agent tasks.More onAnthropic956Claude Code747Claude673Letta4#developer-tools4938#software-engineering2268#ai-agents1480#ai-research334#memory-systems2RelatedHow AI Agent Memory WorksAn interactive essay explains how agent memory systems work through context windows, embeddings, and retrieval architectures to overcome LLM statelessness.also on #memory-systems, #ai-agentsThe Shapes of Agent Memory – Files, Stores, and ExperienceStructured stores beat file memory on accuracy and token cost; files win only for small memories or 'I don’t know' answers, and trained experience helps weak models only.also on Letta, Claude Code, #ai-agents, #developer-toolsYour harness, your memoryAgent harnesses are deeply tied to agent memory, and using a closed one cedes control of memory to a third party, creating vendor lock-in that open harnesses avoid.also on Letta, Claude Code, #ai-agents, #developer-tools