Supreme Exterior Cleaning: GHL CRM, Automation & 4-Channel AI Chatbot
Single AI brain deployed across all 4 inbound communication channels for a UK exterior cleaning business — WhatsApp, SMS, Website, and Messenger — using one OpenAI agent and one Supabase knowledge base, with per-customer conversation memory maintained across every session.
The Problem
- Supreme Exterior Cleaners was receiving inbound messages across four different channels (WhatsApp, SMS, website chat, Facebook Messenger) with no automated response system.
- Each channel required manual monitoring and manual replies.
- Leads that came in outside business hours went unanswered.
- The owner was personally handling every initial inquiry before any qualification had happened, creating a bottleneck from first message to booked job.
The Solution
- 4 parallel systems sharing a single AI core.
- Architecture is identical across all channels: customer message received in GHL on a channel-specific trigger
- GHL sends payload to n8n via Custom Webhook
- n8n normalises the payload in an Edit Fields node
- AI Agent (OpenAI) queries the Supabase vector store knowledge base for relevant business information
- generates a contextual response with conversation memory
- response returned to GHL
- GHL sends reply back on the originating channel.
- WhatsApp and SMS use GHL-side workflows with n8n handling the AI logic.
- Website chatbot uses a pure n8n webhook response pattern with no GHL dependency.
- Messenger routes through GHL with the same n8n AI backend.
- All 4 channels share the same OpenAI agent and Supabase knowledge base — consistent answers regardless of how a customer contacts the business.
Screenshots
Results
All 4 inbound channels (WhatsApp, SMS, Website, Messenger) handled by one AI agent — no channel monitored manually
Single knowledge base in Supabase vector store serves all channels — answer consistency guaranteed regardless of how the customer reaches out
Per-customer conversation memory maintained across sessions on every channel
Website chatbot operates independently of GHL via pure n8n webhook — no dependency on GHL availability
Owner no longer manually responding to initial inquiries — AI handles qualification before any human involvement
Tech Stack
Want something like this?
Tell me what you're building. I'll tell you honestly if I can help.