WhatsApp Memory-Aware Chatbot Workflow (n8n + OpenRouter GPT-4.1)

Turn WhatsApp chats into memory-aware conversations with n8n + OpenRouter (GPT-4.1)

This WhatsApp Memory-Aware Chatbot Workflow listens for incoming WhatsApp messages, retrieves each sender’s saved chat memory from an n8n Data Table, generates a thoughtful reply using an OpenRouter GPT‑4.1 agent, and then sends the response back on WhatsApp—optionally saving new facts for future context.

What this workflow does

• Triggers on new WhatsApp messages via the WhatsApp trigger.
• Splits incoming payload items and routes only text messages for processing.
• Loads sender-specific chat history from an n8n Data Table and aggregates prior messages into a chat-memory list.
• Calls an OpenRouter (GPT‑4.1) agent with the user’s text plus the aggregated memory to generate a reply.
• Optionally stores new “memory” entries extracted by the agent back into the Data Table (sender, recipient, summary, timestamp).
• Sends the generated response back to the user using the WhatsApp Cloud API send step.

Use cases

• Customer support where the assistant remembers prior requests and context per WhatsApp sender.
• Lead qualification chats that retain key facts (e.g., preferences or answers) for follow-up messages.
• Internal ops or SaaS notifications where agents maintain a lightweight conversation history.

Technical details

• n8n nodes: set, switch, split out, WhatsApp (trigger + send), aggregate, data table.
• Model: OpenRouter GPT‑4.1 agent for reply generation and memory extraction.
• Memory storage: an n8n Data Table with columns for sender, recipient, message, created_at (and updated with extracted memory entries as configured).
• WhatsApp integration: requires WhatsApp Cloud API credentials for both trigger and send nodes, plus correct mapping of the recipient WhatsApp ID and your WhatsApp Phone Number ID.