This n8n workflow creates an intelligent HR and IT helpdesk chatbot that leverages PDF document extraction, vector embedding, and AI language models to provide accurate responses to employee queries. It streamlines knowledge management by importing internal policy PDFs, transforming them into a searchable vector store, and integrating with Telegram for real-time communication.
The workflow begins with downloading and extracting text from internal policy PDFs using an HTTP request and PDF extraction node. The extracted text is split into manageable chunks, transformed into numerical vectors with OpenAI embeddings, and stored in a PostgreSQL vector database. This makes it possible for the AI to quickly retrieve relevant policy information when answering questions.
It features a Telegram trigger to listen for employee messages, with logic to handle both text and voice inputs. Voice messages are transcribed with OpenAI’s speech-to-text model, ensuring all queries are processed as text. An AI agent then uses retrieval-augmented generation (RAG) with the vector store and maintains conversation context via a chat memory stored in PostgreSQL. The AI assistant provides relevant, context-aware answers related to HR and IT policies, troubleshooting, and FAQs.
This workflow is highly practical for organizations looking to automate employee support, improve response times, and maintain consistent knowledge base interactions via chatbots integrated into common messaging platforms like Telegram.
Reviews
There are no reviews yet.