Optimize n8n WhatsApp Chatbot Interaction

have an n8n workflow that works very well as a WhatsApp customer service chatbot:
• It’s connected to WhatsApp (Cloud API / webhook).
• It’s integrated with OpenAI API for AI responses.
• It saves chat transcripts to SQL.
• It also sends the conversation by email.

Right now, there is one main issue:

When a customer sends multiple messages in a row (for example:
“Hello”
“I have a problem with my account”
“This is my username…”),
the bot answers each message separately, and it ends up bombarding the user with many replies.

What I want:

I want to improve the workflow so that:
• The chatbot waits and collects all user messages for a short period (for example 5–15 seconds of inactivity, or configurable).
• Then it sends ONE combined reply using OpenAI, instead of replying to every single message.
• This should work per user / per chat session, not mixing different users.

Requirements:
• Strong experience with n8n (self-hosted)
• Experience with WhatsApp Cloud API / webhooks
• Familiar with OpenAI API (chat completions)
• Comfortable working with SQL (the transcripts part is already working, just keep it compatible)

Deliverables:
• Updated n8n workflow (or new version) that:
• Groups incoming messages from the same user for a configurable time window.
• Sends one AI reply after the user finishes typing.
• Avoids duplicate or spammy messages.
• Short documentation or Loom/video / screenshots explaining:
• What you changed.
• Where I can change the “wait time” before replying.

Please start your message with “n8n WhatsApp” so I know you actually read the description.
Share any similar n8n / chatbot projects you’ve done before.