AI Agent: Telegram PDF RAG Q&A with OpenAI, Groq & Pinecone

Unlock Intelligent Document Interaction with this AI Agent

This AI Agent transforms your Telegram into a smart assistant capable of understanding and answering questions about PDF documents you provide. Simply upload a PDF, and the agent will process it using OpenAI for embeddings and store it in a Pinecone vector database. Then, ask questions in plain language, and our Groq-powered Llama 3.1 language model will retrieve relevant information from your documents and provide concise answers. This AI-driven automation is perfect for anyone looking to build a Retrieval Augmented Generation (RAG) system for their PDFs without complex coding.

Key Abilities & Benefits

• Document Ingestion & Processing: Automatically extracts text from uploaded PDFs, splits it into manageable chunks, and generates powerful semantic embeddings using OpenAI. Benefit: Makes your PDF content ready for intelligent search.
• Vector Storage & Retrieval: Leverages Pinecone to store document embeddings, enabling ultra-fast and semantically relevant information retrieval. Benefit: Quickly find the exact information you need, even in large documents.
• Advanced Question Answering: Utilizes Groq's high-performance Llama 3.1 model to understand your questions and generate human-like answers based on the content of your PDFs. Benefit: Get instant, context-aware answers directly within Telegram.
• Seamless Telegram Integration: Interact entirely through Telegram – upload PDFs, ask questions, and receive answers in your chat interface. Benefit: Easy access and use within your existing communication workflow.
• AI-Driven Automation: A complete RAG pipeline automated with n8n, handling everything from document loading to answer generation. Benefit: Save time and resources by automating complex document analysis tasks.

• An n8n instance (Cloud or self-hosted).
• Telegram Bot API Token.
• OpenAI API Key (for embeddings, e.g., text-embedding-ada-002 or newer).
• Pinecone API Key and an existing Pinecone index (e.g., "telegram" with appropriate dimensions).
• Groq API Key (for LLM access, e.g., llama-3.1-70b-versatile).

1. Download the n8n workflow JSON file.
2. Import the workflow into your n8n instance.
3. Configure the 'Telegram Trigger' node: Create a new Telegram credential or select an existing one using your Bot API token.
4. Configure the 'Telegram get File', 'Telegram Response', and 'Telegram Response about Database' nodes with the same Telegram credential.
5. Configure 'Embeddings OpenAI' (both instances): Create or select OpenAI credentials using your API Key.
6. Configure 'Pinecone Vector Store' (both 'insert' and the one connected to 'Vector Store Retriever'):
   • Create or select Pinecone credentials using your API Key and environment.
   • Specify your Pinecone Index name (e.g., "telegram") in the 'Pinecone Index' parameter. Ensure this index exists in your Pinecone account with the correct dimension for your OpenAI embedding model (e.g., 1536 for text-embedding-ada-002).
7. Configure 'Groq Chat Model': Create or select Groq credentials using your API Key. You can adjust the model if needed.
8. Review the 'Recursive Character Text Splitter' for chunk size/overlap if your PDF content requires specific tuning.
9. (Optional) Customize the pineconeIndex value in both Pinecone nodes if you used a different index name.
10. Activate the workflow.
11. Send a PDF document to your Telegram bot to ingest it. Then, ask questions about the document's content.