n8n Website Chatbot Workflow: GPT-4o-mini + Session Memory

Turn your website visitors into real conversations—instantly

This n8n website chatbot workflow provides a public chat webhook that answers user messages with OpenAI’s gpt-4o-mini, while using session memory to keep the dialogue coherent across multiple turns.

What this workflow does

Designed for n8n automation engineers and SaaS operators, this workflow connects your site to an AI-powered support/chat experience:

• Receives visitor messages through the public n8n chat webhook trigger (via your embedded chat widget URL).
• Interprets the message using an AI agent with a customizable system prompt, so the chatbot can follow your website’s tone and support boundaries.
• Maintains recent context by pulling conversation history from buffer memory (the last 10 messages) to avoid repetitive, disconnected replies.
• Generates a response by calling the OpenAI Chat Model node configured with gpt-4o-mini.
• Returns the generated reply back to the chat widget for the visitor.

Use cases

• Website visitors ask questions about your product—your chatbot responds consistently within the same session.
• Early-stage SaaS support triage: answer FAQs and guide users based on your defined system prompt boundaries.
• Lead qualification style chat: use the system prompt to steer conversations toward the next best action.

Technical details

• n8nnodes-langchainchat trigger: exposes the chat webhook that your website widget connects to.
• n8nnodes-langchainagent with a custom system prompt for response control.
• n8nnodes-langchainlm (OpenAI Chat Model): configured to use gpt-4o-mini.
• n8nnodes-langchainmemory (buffer window): session memory using the last 10 messages.
• Setup steps: add an OpenAI API credential, select it in the OpenAI Chat Model node, update the agent system prompt, activate the workflow, and embed the chat widget using the public URL.