ZEP Memory

This video demonstrates a complete workflow where an AI agent is built using n8n integrated with Zep for long-term relational memory. The system retrieves and filters user data via HTTP requests and code nodes, combining short-term memory from a PostgreSQL store with a dynamic knowledge graph from Zep. It uses Telegram for session-based interactions and optimizes API token usage with two memory management methods. The workflow, created by Nate Herk, is available as a free template via his Skool community.