Memory Store (memory.store) is an AI-native persistent memory layer and "company brain" built by the creators of Julep AI. It captures decisions, facts, and context across fragmented tools (such as Slack, Gmail, IDEs, Claude, and ChatGPT) and synthesizes them into structured, auto-updating knowledge accessible by AI agents through the Model Context Protocol (MCP) and REST APIs.
graph TD
subgraph "External Context Sources"
Slack[Slack / Email / Notes]
IDE[IDE / Git / Coding Agents]
Chat[Claude / ChatGPT / Cursor]
end
Slack & IDE & Chat --> Ingest[Memory Store Ingestion API]
Ingest --> Synthesizer[Synthesis & Fact Extraction]
Synthesizer --> RelationalStore[(PostgreSQL / TimescaleDB)]
Synthesizer --> VectorStore[(Vector Embeddings)]
Synthesizer --> Briefs[Living Docs / Auto-Updating Briefs]
subgraph "Agent Access Layer"
MCP[MCP Server: record / recall / check_in]
REST[REST API & SDKs]
end
RelationalStore & VectorStore & Briefs --> MCP & REST
MCP & REST --> Agent[AI Agents & Workflows]
record, recall, check_in), making it instantly compatible with MCP-compliant clients like Claude Code, Cursor, and custom agent harnesses.pgvector for state persistence).When integrated into an MCP client, Memory Store exposes three fundamental cognitive primitives:
| Tool | Purpose | Example Use |
|---|---|---|
record |
Stores a discrete decision, user preference, or fact. | Recording an architectural choice or database schema constraint. |
recall |
Queries memory using semantic search and entity filters. | Looking up prior conventions or decisions before generating code. |
check_in |
Syncs current session status, tasks, and progress. | Saving checkpoint state when wrapping up a development turn. |
To configure Memory Store in an MCP client configuration (such as claude_desktop_config.json):
{
"mcpServers": {
"memory-store": {
"command": "npx",
"args": ["-y", "@julep/memory-store-mcp"],
"env": {
"MEMORY_STORE_API_KEY": "your-api-key"
}
}
}
}