A hosted memory layer that stores and retrieves context for AI applications using four parallel search strategies: vector similarity, temporal recency, keyword matching, and knowledge graph traversal. It automatically extracts entities and relationships from stored memories, ranks results using cognitive science activation models, and trains ML models on your usage patterns to improve retrieval quality over time. Memories have lifecycle stages and gain or lose confidence based on corroboration. You get eight actions including store, recall, update, forget, and graph traversal. Connect via MCP or REST API with your server URL and API key. Useful when you need persistent memory across sessions that actually learns which retrieval strategy works best for different query types rather than just doing basic vector search.
claude mcp add --transport stdio aiappsapi-adaptive-recall uvx adaptive-recall