AI Knowledge Base: Automate with Ollama, PGVector & Telegram

Transform your documents into an intelligent, AI-powered knowledge base that answers questions instantly through Telegram. This advanced n8n workflow combines Ollama, PGVector, and Retrieval-Augmented Generation (RAG) to create a searchable repository from your uploaded files, delivering contextual AI responses in seconds.

What this workflow does

Upload documents through a simple n8n form interface and watch as your files are automatically processed into a semantic knowledge base. The workflow extracts content from PDFs, JSON, CSV, XLS, XLSX, and plain text files, then intelligently chunks the data for optimal retrieval. Using the nomic-embed-text embedding model, it generates vector embeddings stored in PostgreSQL with PGVector extension for lightning-fast semantic search.

Ask questions directly through your Telegram bot, and the system performs semantic similarity searches to find the most relevant document context. Llama 3 running via Ollama then generates accurate, contextual responses based on your actual documents, delivering personalized answers straight to your Telegram chat.

Use cases

• Personal knowledge management: Create a searchable library of research papers, notes, and documents
• Team documentation: Build an AI assistant for company policies, procedures, and manuals
• Customer support: Instant answers from product documentation and FAQ databases
• Educational resources: Transform course materials into an interactive learning assistant
• Research automation: Query large document collections for specific information

Technical details

Built with essential n8n nodes including Form, Telegram, If, Switch, No Op, and Sticky Note components. Requires Ollama with Llama 3 model, PostgreSQL with PGVector extension, and Telegram Bot API credentials. Optional Cloudflare Tunnel or ngrok setup enables public access to your knowledge base system.

This workflow leverages cutting-edge RAG technology to ensure responses are grounded in your actual documents, eliminating AI hallucinations while providing accurate, context-aware answers from your personal knowledge repository.