agentusecasesAll 965 use cases
Sales, Marketing & Support

Answer WhatsApp customer questions and escalate to humans

An n8n and OpenAI agent answers WhatsApp questions from a knowledge base and escalates low-confidence cases to a human on Telegram with full context.

Done withn8n

What they did
An n8n workflow takes each incoming WhatsApp message and assembles the conversation history plus the company knowledge base. OpenAI drafts a reply grounded in that content, and a confidence check decides whether it can be answered reliably. Confident answers go back on WhatsApp. Otherwise the case goes to the human team on Telegram with the full conversation attached.
How it went
The README gives no measured results, only design goals: cover routine questions, give round-the-clock coverage, and keep a human on unreliable cases. The exported n8n workflow JSON was still being added.
Worth knowing
The repo documents architecture only, with no workflow file yet and no description of how confidence is scored, so you would have to design that check yourself.

Try it yourself with n8n

Build an n8n workflow that answers customer questions on WhatsApp using an AI agent grounded in [my knowledge base, e.g. a Google Doc or FAQ sheet]. When the agent's confidence is low or the customer asks for a human, forward the full conversation to me on Telegram and let me reply from there. Done when a test question gets a correct answer and a deliberately tricky one gets escalated to me with context.

Read the original ↗

Source: github.com · Jul 1, 2026

Five of these in your inbox every morning

The best things people got an AI agent to do, each with the prompt to try it.

More like this