Supermemory drops persistent agent memory engine

Supermemory shifts the paradigm from static vector search to dynamic, stateful memory that actually understands temporal changes.

• –Outperforms standard RAG on benchmarks like LongMemEval and LoCoMo by focusing on fact extraction over simple text chunking.
• –Native MCP support allows the memory hub to function as a universal gateway for tools like Claude Desktop, Cursor, and VS Code.
• –Automatically resolves contradictions and updates user preferences (e.g., location or tech stack) without manual database management.
• –Open-source and self-hostable architecture addresses critical privacy requirements for developers building long-term AI context layers.
• –Backed by $3M in seed funding from industry heavyweights including Google's Jeff Dean and OpenAI executives.