Who’s it for For makers, founders, and productivity nerds who want to listen to their inbox instead of reading it. No servers, no hosting — all done with n8n , a Telegram bot , and AI/ML API (LLM + TTS). What it does / How it works This workflow listens for new Gmail emails , extracts the sender, subject, date, and snippet, generates a 2–3 sentence natural summary using GPT-5 via AI/ML API, then converts the text into a lifelike voice message using Inworld TTS-1-Max . The final audio file is downloaded and instantly delivered to your Telegram via a bot. A separate auxiliary flow captures your Telegram chat_id once, stores it in a Data Table, and the main flow reuses it every time — no hardcoding required. High-level flow: Gmail Trigger detects new incoming email Code node builds a clean emailData block AI/ML API (GPT-5) generates a natural spoken summary AI/ML API (Inworld TTS-1-Max) converts summary → voice Audio file is downloaded Voice message is sent to the same Telegram chat Requirements A working n8n instance (self-hosted or cloud) Gmail OAuth2 credentials Telegram bot token from @BotFather AI/ML API key Base URL: https://api.aimlapi.com/v1 Supports GPT-5 and Inworld TTS-1-Max Set up steps Create a Telegram bot via @BotFather and add Telegram API credentials in n8n Create an AI/ML API credential using your API key Add Gmail OAuth2 credentials for inbox access Import the workflow JSON Open nodes and attach the correct credentials (no hardcoded tokens inside nodes) Run the Telegram Trigger once and send any message to your bot — this saves your chat_id Activate your workflow and receive your first voice-summary How to customize Replace GPT-5 with another LLM from AI/ML API (same schema) Change prompt style: tone, structure, language, or verbosity Add language selection or multi-voice support Add filters (e.g., only important emails, only starred emails) Log all activity to Google Sheets or a database Add IF conditions: voice only during certain hours, weekday vs weekend rules Add rate-limits or user allowlists Change the voice model for different notification styles: corporate, friendly, narrator