Telegram AI Bot with Memory-Based Customer Replies

How it works

1. 1.Telegram Trigger receives incoming customer messages.
2. 2.The AI Agent processes the message text with a system message enforcing concise, friendly behavior.
3. 3.The gpt-4o-mini chat model generates the response based on the conversation context.
4. 4.The buffer window memory stores the last 10 exchanges keyed by the Telegram chat ID.
5. 5.The Telegram node sends the agent's output back to the originating chat.

Use cases

• Lightweight first-line customer support for a small business on Telegram
• Automated replies to general questions while deferring pricing and policy details
• Multilingual chat assistant that mirrors the customer's language

Requirements

• A Telegram bot token created via BotFather configured in the Telegram trigger and sender credentials
• Valid OpenAI credentials with access to the gpt-4o-mini model
• The workflow must be activated for the trigger to receive messages automatically

Setup and testing in n8n

1. Send a text message to the bot and verify a reply under 120 words arrives
2. Ask about prices or business hours and confirm the bot says it will check with the team
3. Message the bot in another language and verify the reply matches that language
4. Send several follow-up messages to confirm memory of the prior conversation within 10 messages

Service Value

Ideal as a smart automation service combining integrations and AI to produce ready-to-use results.