@Tim Westermann Thanks for the informstions!

@Henry James Thanks

@Sunday Kennedy I'm building this directly in n8n . I'm using the AI Agent node with a Gemini model connected via a Messenger Webhook

@Saulo C. This is brilliant. The timestamp check logic sounds like exactly what I need to prevent duplicate replies. I'll try to implement this 'last message' validation logic. Thanks for the detailed breakdown

Hello community, I am currently building an AI Sales Chatbot for Facebook Messenger using the AI Agent node. my workflow triggers immediately on every incoming Webhook message. If a user sends multiple short messages in a row (e.g., "Hello", "I want to buy", "a drone"), the workflow runs 3 times separately. This wastes AI tokens and results in the bot replying 3 separate times, which creates a bad user experience. I want the following sequence of events: 1. When a message arrives, it is stored/appended temporarily. 2. The workflow initiates a short delay (e.g., 5-10 seconds). 3. If a new message arrives from the same sender_id during this delay, the timer resets, and the new message is appended to the previous ones. 4. Only after the silence period (no new messages) does the workflow send the full concatenated text block to the AI Agent for processing. My Question: What is the simplest way to achieve this workflow? Thank you for your advice!

thanks