Turn promotional emails into a short audio update sent to Telegram. Teams can hear new offers and deals in minutes, without digging through the inbox. Great for marketers and busy managers who want quick insight while on the go.

The flow checks Gmail every minute for messages with the promotions label. It pulls the full email content, removes extra fields, then uses an AI summarization chain powered by OpenAI to create a concise, newsletter style summary. The summary text is sent to a text to speech API, the audio is converted from base64 to a file, and the text and audio are merged. The output is sent to a Telegram chat as an audio message with the summary as the caption. This creates a spoken digest of promotions that is easy to consume and share.

Setup needs Gmail access, an OpenAI API key, and a Telegram bot token. You choose the Gmail label, the Telegram chat, and the voice used for speech. Expect less time spent on email scanning and a faster way to share deals with your team. Useful for daily promo roundups, competitive monitoring, and quick internal updates.

• Gmail polling every minute with a label filter for promotions
• Fetch full email content for accurate context and summarization
• Code step removes unneeded fields to keep payload clean
• AI summarization chain with OpenAI and a custom newsletter prompt
• Text to speech via HTTP request to generate an audio file
• Convert base64 response into a binary audio file ready to send
• Merge summary text and audio and aggregate outputs when needed
• Send audio to Telegram with the summary as the caption

• Reduce manual review from 60 minutes to 5 minutes per day
• Automate up to 90% of repetitive inbox scanning
• Improve summary consistency by 80% using a standard prompt
• Handle 10 times more promotional emails without extra effort
• Connect Gmail, OpenAI, and Telegram in one stream

1. Import the template into n8n: Create a new workflow in n8n > Click the three dots menu > Select 'Import from File' > Choose the downloaded JSON file.
2. You'll need accounts with Gmail, OpenAI and Telegram. See the Tools Required section above for links to create accounts with these services.
3. In the n8n credentials manager, open the Gmail Trigger node and the Gmail node. In the credential dropdown, click Create new credential, choose Gmail OAuth2, and follow the on screen steps to grant access to your Gmail account.
4. Open the OpenAI Chat Model node. In the credential dropdown, click Create new credential, select OpenAI API Key, and paste your API key from the OpenAI account dashboard.
5. Open the Telegram node. In the credential dropdown, click Create new credential, select Telegram API, and add your bot token from BotFather. Save the credential.
6. In the Gmail Trigger node, set Poll Times to every minute and set Filters to the label used for promotions, for example CATEGORY_PROMOTIONS. Save the node.
7. In the Get message content node, confirm operation is get and that messageId is mapped to {{$json.id}}. This lets the node pull the full email body.
8. Open the Summarization Chain node. Confirm the OpenAI model is selected as the language model and review the prompt. Keep chunking mode as advanced for long emails.
9. Open the HTTP Request node for text to speech. Set method to POST, URL to the provided endpoint, set Content Type to application/json, and include the text and voice fields in the JSON body.
10. Check the Convert from base64 to File node. Ensure the source property is data so the audio is turned into a binary file.
11. Open the Telegram node and set operation to sendAudio, binaryPropertyName to data, and enter your target chatId. Keep the caption expression to post the summary text.
12. Run a test: label one email with CATEGORY_PROMOTIONS, then execute the workflow once. Verify the audio and caption arrive in your Telegram chat.
13. If messages do not send, check Gmail scopes, confirm your Telegram bot can post to the chat, verify OpenAI quota, and ensure the TTS API returns a data field. Review the n8n execution log for any node errors.
14. Enable the workflow so it runs on schedule and monitor early runs to confirm timing, length of audio, and caption formatting.