Build a Retrieval-Based Chatbot with Telegram, OpenAI, and Google Drive PDF Backup
# Telegram RAG Chatbot with PDF Document & Google Drive Backup - An upgraded Retrieval-Augmented Generation (RAG) chatbot built in **n8n** that lets users ask questions via Telegram and receive accurate answers from uploaded PDFs. It embeds documents using OpenAI and backs them up to Google Drive. ## ☀ Who's it for Perfect for: - Knowledge workers who want instant access to private documents - Support teams needing searchable SOPs and guides - Educators enabling course material Q&A for students - Individuals automating personal document search + cloud backup ## ℹ How it works / What it does ### Telegram Chat Handling 1. **User sends a message** Triggered by the Telegram bot, the workflow checks if the message is text. 2. **Text message ➡ OpenAI RAG Agent** If the message is text, it's passed to a GPT-powered document agent. This agent: - Retrieves relevant info from embedded documents using semantic search - Returns a context-aware answer to the user 3. **Send answer back** The bot sends the generated response back to the Telegram user. 4. **Non-text input fallback** If the message is not text, the bot replies with a polite unsupported message. ### ➡ PDF Upload and Embedding 1. **User uploads PDFs manually** A manual trigger starts the embedding flow. 2. **Default Data Loader** Reads and chunks the PDF(s) into text segments. 3. **Insert to Vector Store (Embedding)** Text chunks are embedded using OpenAI and saved for retrieval. 4. **Backup to Google Drive** The original PDF is uploaded to Google Drive for safekeeping. ## 🛠 How to set up 1. **Telegram Bot** - Create via [BotFather](https://t.me/botfather) - Connect it to the Telegram Trigger node 2. **OpenAI** - Use your OpenAI API key - Connect the Embeddings and Chat Model nodes (GPT-3.5/4) - Ensure both embedding and querying use the same Embedding node 3. **Google Drive** - Set up credentials in n8n for your Google account - Connect the "Backup to Google Drive" node 4. **PDF Ingestion** - Use the "Upload your PDF here" trigger - Connect it to the loader, embedder, and backup flow ## 📜 Requirements - Telegram bot token - OpenAI API key (GPT + Embeddings) - n8n instance (self-hosted or cloud) - Google Drive integration - PDF files to upload ## © How to customize the workflow | Feature | How to Customize | |-------------------------|------------------------------------------------------------| | Auto-ingest from folders| Add Google Drive/Dropbox watchers for new PDFs | | Add file upload via Telegram| Extend Telegram bot to receive PDFs and run the embedding flow| | Track user questions | Log Telegram usernames and questions to a database | | Summarize documents | Add summarization step on upload | | Add Markdown or HTML support| Format replies for better Telegram rendering | Built with Telegram + PDF + OpenAI Embeddings + Google Drive + n8n
- Platform
- n8n
- Category
- AI
- Price
- $14.99
- Creator
- Trung Tran
- Chatbot
- RAG
- Telegram
- if
- code
- telegram
- stickyNote
- formTrigger
- googleDrive
- agent
How to import this workflow into n8n
- 1Purchase or download the workflow to get the n8n workflow JSON file.
- 2In your n8n instance, open Workflows and choose "Import from File" (or paste the JSON with Ctrl+V on the canvas).
- 3Open each node marked with a credential warning and connect your own accounts and API keys.
- 4Run the workflow once manually to verify the data flow, then toggle it to Active.
Related AI workflows
- Launch Your First AI-Powered Chatbot with Actionable Tools$9.99
- Automate AI Video Creation and YouTube Upload with Google Sheets$14.99
- Build a WhatsApp Assistant with Memory, Google Suite, Multi-AI, Research, and Imaging$24.99
- Automate Blog Post Creation and Publishing with GPT, Leonardo AI, and WordPress$14.99
- Email Agent$500.99
- Automate SEO Keyword Generation with ChatGPT from Google Sheets$3.99
More from Trung Tran
Need this deployed? We'll set it up for you.
Our automation experts deploy this workflow in your stack, connect your accounts, and verify it works — or build a custom solution from scratch.