Build a Telegram AI Chatbot with Multi-Format Input & OpenAI

How it works

This n8n workflow enables you to build a sophisticated Telegram AI chatbot capable of processing both text and voice messages. It utilizes the @n8n/n8n-nodes-langchain nodes for AI capabilities and the base n8n-nodes-base.telegram nodes for Telegram integration.

1. Listen for incoming events: The workflow starts with the Listen for incoming events (Telegram Trigger) node, which waits for new messages on your Telegram bot.
2. Determine content type: A Determine content type (Switch) node analyzes the incoming message to differentiate between text and voice inputs.
3. Handle text messages: If it's a text message, it proceeds to Combine content and set properties.
4. Handle voice messages: If it's a voice message, it's downloaded by the Download voice file node and then transcribed to text using the Convert audio to text node (OpenAI).
5. Combine and prepare message: The Combine content and set properties (Set) node consolidates the message content (either direct text or transcribed voice) and identifies the message type (text/voice) and if it's a forwarded message.
6. Engage AI Agent: The AI Agent node, powered by Langchain's Agent and OpenAI's GPT-4o (OpenAI Chat Model), processes the combined message. It utilizes Window Buffer Memory to maintain conversation context.
7. Send final reply: The Send final reply node formats the AI's response into HTML and sends it back to the user on Telegram.
8. Error Handling: If an unsupported message type is received, Send error message is triggered. A Send Typing action provides user feedback during processing.