Build a Telegram AI Chatbot for Text and Voice with n8n

Telegram message trigger → text/voice routing → AI Agent response → reply back to Telegram

Who

Operators or developers who need to quickly deploy an AI conversational bot on Telegram

You get

A Telegram bot handling both text and voice messages with typing indicators, defaulting to GPT-4o

You need

n8n · telegram

Steps

1. Prepare credentials and bot

   Contact @BotFather in Telegram to create a bot and obtain the bot token. In n8n Credentials, add a telegramApi credential (paste the bot token) and an openAiApi credential (paste the OpenAI API Key). If you need voice transcription, also prepare credentials for a speech-to-text service (the template defaults to OpenAI Transcribe).

2. Import workflow and configure nodes

   Import workflow.json from the template. Verify these key node configurations: - **Telegram Trigger**: ensure updates includes message, select your telegramApi credential. - **Text Or Voice (Switch)**: keep default branch conditions — upper branch checks `{{ $json.message.text }}` exists (text), lower branch checks `{{ $json.message.voice.file_id }}` exists (voice). - **OpenAI Chat Model**: default model is gpt-4o; click to switch to gpt-4 or another available model. - **Simple Memory**: sessionKey set to `={{ $json.sessionId }}`, contextWindowLength is 10. - **AI Agent**: promptType is define, text input is `{{ $json.message.text }}`; options.systemMessage is in the next prompt. - **Send Response To Telegram**: text is `{{ $json.output }}`, chatId references Telegram Trigger's chat.id, additionalFields.appendAttribution is false. Update all credential nodes to real credentials; do not keep the credential-id placeholder.

   Execution prompt · Source text。AI Agent node options.systemMessage field

   Copy prompt

   =# Overview
   You are a personal assistant that helps user to fulfill their request.
   
   When you are asked to perform a task on the current date, please use the current time and date: {{$now}}
   
   ## Output
   You should output the result and don't include any link. Just mentioned where you added it to.

3. Configure Telegram webhook and public access

   Ensure Telegram can callback to your n8n instance: - If using n8n Cloud, webhooks are usually auto-configured. - If self-hosting, expose n8n's webhook URL via reverse proxy or tunnel (e.g., ngrok), and set the Telegram webhook (use `https://api.telegram.org/bot<TOKEN>/setWebhook?url=<WEBHOOK_URL>`). The Telegram Trigger's webhookId is auto-generated by n8n after import and usually does not need manual change.

4. Test the workflow

   Activate the workflow. In Telegram, send to your bot: 1. A text message and check for an AI reply; 2. A voice message and check for a transcribed text reply; 3. Verify a "typing…" indicator appears before the reply. If the voice branch fails, check that the Get Voice node correctly retrieves the file_id and that the Transcribe The Voice node's openAiApi credential is valid.

You are assisting with reproducing an n8n template: a Telegram AI chatbot. First confirm the user has: a publicly accessible n8n instance, a Telegram bot token, and an OpenAI API Key. Then guide them through importing workflow.json, updating credentials, configuring webhooks, and explain the AI Agent system message purpose. Do not perform sensitive operations for the user (e.g., exposing tokens).