Supermemory is an advanced memory and context engine designed to provide AI agents with persistent, evolving knowledge. Unlike stateless RAG systems, Supermemory utilizes a specialized memory engine that extracts facts from conversations, resolves contradictions, and handles temporal updates through automatic forgetting. The system achieves state-of-the-art performance, ranking #1 on LongMemEval, LoCoMo, and ConvoMem benchmarks. Notably, on LongMemEval, it achieves 95% Recall@15 while providing a 99.4% context reduction, requiring only ~720 tokens of context. The platform features a hybrid search mechanism combining RAG with personalized memory and offers sub-50ms retrieval for user profiles. Developers can integrate the engine via a single API that manages embeddings, chunking, and vector database configurations automatically. It supports multi-modal extraction for PDFs, images, videos, and code, alongside real-time connectors for platforms like GitHub, Notion, and Google Drive. Supermemory is available as a managed service or can be self-hosted as a single binary with zero configuration, supporting local embedding models like Xenova/bge-base-en-v1.5 or external providers like Ollama for fully offline operation. Additionally, it provides an MCP server and various plugins to enable persistent memory for tools like Cursor, Claude Code, and Windsurf.