Back to Projects

Knowledge retrieval / Customer support

RAG-Powered Customer Support

Business-document retrieval connected to AI responses, channel routing, and human escalation.

Built workflow · Deployment not verified

Business problem

Repeated customer questions need answers grounded in approved business information and a clear route to a person when needed.

What I personally built

I built an n8n support workflow with document retrieval, Supabase/vector search, OpenAI, conversation logging, and response or human-escalation routes.

Tools & integrations

n8nSupabaseOpenAIVector search

How the workflow works

  1. Receive and normalise a customer message.
  2. Retrieve relevant business context for the AI response.
  3. Route the supported response through the appropriate channel.
  4. Log the conversation and escalate requests that need human handling.

Testing & evidence

The supplied screenshot shows the built workflow. A documented evaluation of answer accuracy, unsupported questions, and escalation behaviour remains to be published.

Outcome & current status

Built implementation preview. Live deployment, support-resolution rates, and client outcomes are not verified.

n8n RAG support workflow with document retrieval, multi-channel responses, conversation logging, and escalation
Actual implementation screenshotView Workflow Image