How to Build an AI Messaging / Chatbot Automation (No Code)

Wire a chat trigger to an AI model and reply automatically, using n8n or Make, with no code and one short prompt.

10 minIntermediate

Visit n8n

An AI messaging automation is three pieces: a trigger that fires when a message arrives, an AI node that reads it and writes a reply, and an action that sends the reply back to the same chat. You can build the whole loop in n8n or Make without writing code. This guide uses n8n's AI Agent node as the main path and notes the Make equivalent at each step.

An n8n account (cloud or self-hosted), or a Make.com account
A messaging channel to listen on (Telegram, Slack, WhatsApp, or a plain webhook)
A bot token, for Telegram this comes from @BotFather
An OpenAI API key (or another supported model provider)

Step 1: Choose the messaging trigger

Channel	n8n trigger node	What you need
Telegram	Telegram Trigger	Bot token from @BotFather
Slack	Slack Trigger	A Slack app with event subscriptions
WhatsApp	WhatsApp Trigger	A Meta WhatsApp Business number
Anything else	Webhook	A URL you POST messages to

Step 2: Create the bot and add the trigger

In Telegram open @BotFather, send /newbot, and follow the prompts to name the bot and pick a username ending in bot. BotFather returns an API token. In n8n add a Telegram Trigger node, create a credential with that token, and set the update type to Message so the workflow fires on each new incoming message.

Telegram - @BotFather

$/newbot

Alright, a new bot. How are we going to call it?

$Setuproll Helper

Good. Now let's choose a username. It must end in 'bot'.

$setuproll_helper_bot

Done! Use this token to access the HTTP API:

8123456789:AAH7x...keep-this-secret

Step 3: Add the AI Agent node

Add an AI Agent node after the trigger. Click the plus under its Chat Model connection and pick OpenAI Chat Model, then add your API key (gpt-4o-mini is a safe, low-cost default). Map the incoming Telegram message text into the agent's input, and write a system prompt that sets the bot's scope so it stays on topic. The flow is short: Telegram Trigger to AI Agent to a Send Message action.

AI Agent system prompt

You are a helpful assistant for Setuproll users. Answer briefly and clearly in plain language. If a question is outside automation or no-code topics, say so politely and offer to help with something related. Never invent facts you are not sure about.

Step 4: Reply, test, and go live

Add a Telegram node set to Send Message. Map the chat ID from the trigger so the reply goes to the right person, and map the AI Agent output into the Text field. Run the workflow once, message your bot, and confirm a reply lands within a few seconds. When it works, toggle the workflow Active so it runs around the clock.

Building it in Make instead

The same loop in Make.com is Telegram Bot > Watch Updates as the trigger, then OpenAI > Create a Completion (Chat) for the reply, then Telegram Bot > Send a Message mapping the chat id back. Swap the first and last modules for Slack, WhatsApp, or a Custom webhook to listen on a different channel.

Result: a live chatbot that reads each incoming message, drafts an on-brand answer with your chosen model, and replies in the same chat, all built from three connected nodes and one prompt with no code to maintain.