Inspired by human dreaming as a memory reconsolidation process, Anthropic has its AI agents use downtime to "dream." During this background process, the agent reviews its memories, identifies and prunes contradictions, and cleans up the information to improve the coherence and utility of its long-term memory.

The leaked architecture shows a sophisticated memory system with pointers to information, topic-specific data shards, and a self-healing search mechanism. This multi-layered approach prevents the common agent failure mode where performance degrades as more context is added over time.

Effective agent memory is not merely a storage layer. It's an encapsulated system for learning and adaptation that integrates embedding models, re-rankers, databases, and LLMs, all working in concert to hold, move, and store data.

Retrieval-Augmented Generation (RAG) is just one component of agent memory. A robust system must also handle dynamic operations like updating information, consolidating knowledge, resolving conflicts, and strategically forgetting obsolete data.

Instead of treating memory as a component, adopt a "memory-first" approach when designing agent systems. This paradigm shift involves architecting the entire system around the core principles of how information is stored, recalled, and forgotten.

To manage context costs, Tasklet summarizes agent history with decreasing granularity over time. Recent interactions are sent verbatim, while older conversations have tool calls, thinking steps, and messages truncated or summarized. This is done in cache-aware buckets to minimize cost.

Claude's "Dreams" feature is not automatic learning but an explicit API call to review past sessions and synthesize memories. This design gives developers direct control over when and what an agent learns, transforming memory management from a black box into a deliberate, auditable action.

M0 organizes agent knowledge into two distinct layers: a high-level "Experience" summary outlining strategy and cautions, and a detailed "Skill" layer with structured operational steps. This allows an agent to load the compact strategy first and only retrieve operational details when necessary, keeping the active prompt lean and efficient.

A cost-effective AI strategy involves using a powerful, expensive model once to solve a complex task, then using a system like M0 to distill that solution into reusable "experience" and "skill" records. Cheaper models can then leverage this pre-packaged knowledge to execute the same task with higher success rates and significantly lower token costs.