Turn Telegram into an on demand AI speech coach. Share your draft as text or voice, get clear feedback, and request a fresh version that applies your notes. Ideal for marketers, founders, and teams who need fast, consistent messaging.

Messages from Telegram start the process. Text goes straight to analysis. Voice notes are downloaded and transcribed to text with OpenAI. A switch routes the text to one of three roles: start a new speech, give feedback, or produce a refined speech from feedback. The Google Gemini model writes the reply, while a memory window keeps recent context and a reset step can wipe it when needed. Two code steps clean special characters and split long answers into parts under 4000 characters so Telegram can deliver them in order. The system answers back in the same chat.

Setup only needs a Telegram bot, an OpenAI key, and a Google Gemini key. Expect faster drafts, fewer edits, and easy collaboration in one chat. Use it for event talks, sales pitches, or company updates without switching tools.

• Telegram message trigger listens for new chats and replies in the same thread
• Voice detection routes audio to download and transcription before analysis
• OpenAI transcribes voice notes into accurate text
• Switch node sends requests to new speech, feedback, or synthesis paths
• Dynamic system prompts tailor the AI role for each path
• Google Gemini generates fast, clear responses for each request
• Memory buffer keeps recent context and a wipe step can clear it on demand
• Python code cleans special characters to avoid Telegram formatting issues
• Output splitting sends long responses in multiple messages under 4000 characters

• Reduce drafting time from 2 hours to 10 minutes per version
• Accept text and voice so users can work hands free
• Keep up to 25 recent messages for consistent feedback
• Deliver long answers by splitting over 4000 characters into clean parts
• Cut formatting errors in Telegram by removing special characters
• Support many users in one bot with per user sessions

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 Telegram, Google Gemini and OpenAI. See the Tools Required section above for links to create accounts with these services.
3. In Telegram, open BotFather, create a new bot, and copy the bot token.
4. In the n8n credentials manager, create a new Telegram credential and paste the bot token. Save it.
5. In your OpenAI account, create an API key. In n8n, create a new OpenAI credential and paste the key. Save it.
6. In Google AI Studio, generate a Gemini API key. In n8n, create a new Google Gemini credential and paste the key. Save it.
7. Open the Telegram Trigger node and select your Telegram credential. Confirm updates are set to message.
8. Open the Download Audio File and Respond to Telegram Message nodes and select the same Telegram credential.
9. Open the Transcribe Audio File node, choose your OpenAI credential, and confirm the operation is transcribe.
10. Open the Google Gemini Chat Model node, choose your Gemini credential, and keep the model set to models/gemini-2.0-flash-001.
11. Review the Route Flow Based on Message Content node. Add the phrases your users will send to trigger new speech, feedback, or synthesis, and make sure each output connects to the right Set prompt node.
12. Check the memory nodes. Keep the session key set to the Telegram user id and confirm the context window length is 25. Test the wipe step if you plan to use a reset command.
13. Activate the workflow. Send a short text to your bot and confirm a clean reply returns.
14. Send a voice note. Verify it transcribes and returns a response. If nothing returns, check the execution logs and the OpenAI credential.
15. Paste a long prompt. Confirm the bot replies in multiple messages. If formatting looks odd, review the clean and split code nodes for any edits.