Supermemory provides a universal memory layer focused on ingesting user content from external sources.
| Feature | Rice | Supermemory |
|---|---|---|
| Philosophy | Cognitive substrate for autonomous agents | Second brain for user personalization and RAG |
| Memory Model | Four-component cognitive architecture | Unified knowledge graph with user profiles |
| Execution | Server-side WASM skills | None (uses MCP for external tools) |
| Data Ingestion | Agent-driven (commit, focus) | Integration-driven (Notion, Drive, Slack connectors) |
| Underlying Tech | RiceDB with HDC | Postgres + standard vector search |
Rice is built for autonomous agents that need to think, learn skills, and manage their own state. Supermemory is built for user personalization and RAG, excelling at aggregating user data from external sources.
Rice uses a cognitive architecture with distinct Working, Episodic, Procedural, and Semantic memory. Supermemory uses a unified knowledge graph approach to link entities and generate user profiles.
Rice includes server-side WASM execution via Procedural Memory. Supermemory relies on the Model Context Protocol (MCP) to connect agents to external tools but does not host or execute logic itself.
Rice is agent-driven. Agents actively commit traces and focus on items. Supermemory is integration-driven with connectors for Google Drive, Notion, Slack that automatically ingest and index user content.
Rice is for autonomous agents that need to learn how to perform tasks. Supermemory is for "Chat with your Data" applications that need to know what's in your Notion or who you emailed.