Telegram AI Chatbot with RAG using OpenAI & PGVector

This n8n workflow creates a sophisticated Telegram AI chatbot enhanced with Retrieval-Augmented Generation (RAG) using OpenAI and PGVector in PostgreSQL. It ingests documents from sources like Google Drive, chunks and embeds them into a vector database for semantic search, enabling the bot to provide accurate answers grounded in your private data, reducing hallucinations common in generic AI.

Key benefits include personalized responses for support, onboarding, and knowledge access; seamless integration with Postgres (Neon, Supabase, or self-hosted); and contextual memory for ongoing conversations. It builds on a Telegram starter template, adding vectorization and retrieval logic for high-relevance results.

Ideal use cases: Customer help desks, internal FAQ bots, client-facing knowledge assistants. Setup involves configuring credentials for Telegram, OpenAI, Postgres/PGVector, and Google Drive. The workflow handles message receipt, context retrieval, AI generation, and responses, with optional expansions for Notion or Airtable ingestion.