Human-like memory for AI agents — semantic, episodic & procedural. Experience-driven procedures that learn from failures. Free API, Python & JS SDKs, LangChain, CrewAI & OpenClaw integrations.

The cognition layer that turns your agent into a HyperAgent

Benchmark suite for evaluating retrieval quality and latency of AI agent context systems

Synapse Layer — Continuous Consciousness Infrastructure for AI Systems. Persistent. Secure. 1-line integration.

Maximem Synap is the memory layer that makes AI agents remember. #1 on LongMemEval (90.2%). Works natively with LangChain, LlamaIndex, CrewAI, Google ADK, AutoGen, OpenAI Agents, Semantic Kernel, Haystack, and Pydantic AI.

SQL Native Memory Layer for LLMs, AI Agents & Multi-Agent Systems

Memory and recall layer for AoA: explicit memory objects, provenance threads, temporal relevance, salience, and reviewable recall contracts.

Ultimate memory layer for local LLM's ~ mitigate limitations of Context Window.

Persistent identity and memory across AI tools - mcp-native, local-first, framework-agnostic, production-ready.

A memory layer for LLMs that extracts entities, relationships, and decisions from conversations and stores them in a semantic knowledge graph, giving any AI persistent, structured context across sessions.

Standalone memory middleware for the Poncho AI agent (FastAPI + SQLite + Claude Haiku).

The AI agent memory engine that deliberately stays small. SQLite + sqlite-vec + Ollama, CJK-native, three entry points. Apache 2.0.

Cross-host memory fabric for Claude and Codex, with provenance, recall, and unified CLI workflows.

🧠 Universal memory infrastructure for AI agents. Give your AI a memory that never forgets.

smrti (स्मृति) — non-dual memory for structured knowledge elicitation

Belay Protocol reference — deduplication engine, canonicalization, pluggable VectorStore. MIT. A component of Belay, memory for the AI you already have.