Supermemory is a memory engine and context platform that gives AI assistants persistent memory across conversations. Founded by a 19-year-old developer with backing from Google executives, it ranks first on all three major AI memory benchmarks — LongMemEval, LoCoMo, and ConvoMem. The core problem it solves is fundamental: every AI conversation starts from scratch, losing all context from previous interactions. Supermemory automatically extracts facts, preferences, and project context from conversations, builds evolving user profiles, handles knowledge updates and contradictions, and forgets expired information without manual intervention.
The platform operates as both a consumer app and a developer API. On the consumer side, a browser extension, desktop apps, and mobile companions let you save links, chats, PDFs, images, and documents that Supermemory indexes and makes searchable through natural language queries. For developers, the API provides the full context stack in a single system: memory extraction, RAG with hybrid vector and keyword search at sub-300ms latency, auto-maintained user profiles, file processing for any format, and connectors for Google Drive, Notion, Slack, Gmail, and S3. SDKs are available for TypeScript and Python with just three lines of code to integrate.
Supermemory's MCP server is its most powerful distribution channel — a single configuration line in Claude Desktop, Cursor, Windsurf, VS Code, Claude Code, OpenCode, or OpenClaw gives any MCP-compatible client instant access to persistent memory. The server supports both OAuth and API key authentication with project-scoped memory containers for separating work and personal contexts. Built on Cloudflare Workers with Durable Objects, the MCP server handles over 100 billion tokens monthly. Plugins for Claude Code, OpenCode, and OpenClaw ship as open-source implementations, and the team also created MemoryBench, an open evaluation platform for benchmarking memory systems.