Telegram RAG AI Agent with Custom Knowledge

Overview

This n8n workflow creates a powerful AI chatbot accessible via Telegram. It leverages Retrieval Augmented Generation (RAG) to allow users to upload their own documents (PDFs, CSVs) and then ask questions that the AI answers based only on the provided knowledge base. The workflow breaks down into several key stages:

1. Document Ingestion and Processing: When a user uploads a file via the Upload your file here form, the workflow uses the Recursive Character Text Splitter to break down the document into manageable chunks.
2. Vector Embedding and Storage: The Embeddings HuggingFace Inference node converts these text chunks into numerical vectors. These vectors are then stored in an in-memory vector store (Simple Vector Store), indexed by vector_store_key.
3. User Query Handling: Incoming messages from Telegram (Telegram Message Trigger1) are captured. The user's query is then processed by an AI Agent1.
4. Information Retrieval: The AI Agent1 uses the Retrieve documents node to search the vector store for relevant information based on the user's query.
5. AI Response Generation: The Groq Chat Model is utilized by the AI Agent1 to generate a response, strictly adhering to the system prompt which confines its knowledge to the retrieved documents.
6. Telegram Reply: The final answer is sent back to the user in Telegram using the Reply in Telegram1 node.

This workflow is ideal for creating custom Q&A bots, internal knowledge base assistants, or any application where an AI needs to answer questions based on a specific, user-provided set of information.