This project implements a hybrid memory AI system using n8n, combining both semantic (vector-based) and symbolic (SQL-based) memory to create a context-aware conversational agent. It uses OpenAI's embeddings and GPT models alongside PGVector in PostgreSQL to store and retrieve past conversation data both semantically and literally. The agent dynamically generates SQL queries to recall specific past messages, while also leveraging vector search to understand user intent from previous interactions. This dual-memory architecture allows the AI to respond accurately and contextually, reducing hallucination and enabling intelligent dialogue across sessions.

Hello everyone I am Ali