Supermemory

What is Supermemory?

Supermemory is a context engineering platform that gives AI agents a powerful, persistent memory—like a "second brain" that remembers everything across conversations, tools, and sessions. Unlike traditional RAG or vector databases that treat each interaction as isolated, Supermemory builds a dynamic, evolving knowledge graph that understands users, merges facts, resolves contradictions, and retrieves context in under 300ms.

Built for developers, teams, and individuals, it works with any AI model and integrates seamlessly via API, plugins (like for Claude or Cursor), or a personal app. Whether you're building enterprise AI assistants or just want your personal AIs to remember your preferences, Supermemory ensures your agent always has the right context—fast, accurate, and up to date.

What are the features of Supermemory?

• Memory & Continual Learning: Persistent, structured memory built as a knowledge graph using dynamic dreaming—facts evolve, merge, and get forgotten like human memory.
• SuperRAG (Retrieval): Hybrid search with reranking and structured context retrieval in under 300ms, far faster than alternatives like Zep or Mem0.
• Real Filesystems: Mount a semantic filesystem where grep becomes smart search and files auto-index—works natively on macOS and Linux with standard commands.
• User Profiles: Agents remember user identity, preferences, and behavior across sessions—not just the current chat.
• Connectors: Auto-sync data from Slack, Notion, Gmail, GitHub, Drive, S3, and more—no manual imports or stale data.
• Smart Extractors: Turn PDFs, web pages, images, and audio into meaningful memory objects with intelligent chunking that preserves context.
• Qualitative Analysis: Analyze, cluster, and summarize user signals directly inside your memory—without exporting data or compromising privacy.

What are the use cases of Supermemory?

• Build AI assistants that adapt over time by remembering user history, preferences, and past interactions.
• Create self-updating knowledge bases that sync live from internal docs, wikis, or APIs—always current, never stale.
• Power real-time agent reasoning with millisecond access to the latest facts from multiple sources.
• Give engineering teams instant access to institutional knowledge—searchable, unified, and always in sync.
• Enable personal AI workflows where everything you teach one AI (via Chrome extension or plugins) is remembered by all your AIs.
• Replace traditional RAG pipelines with a unified memory system that reduces latency by 70% and cuts token usage by 40–50%.

How to use Supermemory?

• Install the SDK with npm install supermemory or use the Python package.
• Initialize the client and start ingesting data—from files, apps, or raw text—using automatic extraction.
• Connect your tools (Notion, Gmail, etc.) so Supermemory stays updated in real time.
• Query memory using natural language; retrieval pulls from profiles, documents, and conversation history in one unified graph.
• Deploy via API, use the Personal App, or integrate plugins for Cursor, Claude, and more.
• Monitor usage in the Developer Console and manage API keys, top-ups, and teammates.