RAG Support Copilot Workflow for n8n: Claude + OpenAI Embeddings

Turn your internal docs into a support-grade RAG chat with Claude + OpenAI embeddings (n8n)

This RAG Support Copilot workflow for n8n ingests your internal documentation, indexes it in an in-memory vector knowledge base using OpenAI embeddings, and powers a public support chat powered by Anthropic Claude that answers only from your knowledge—escalating anything missing to Gmail and logging every Q&A to an n8n Data Table.

What this workflow does

• Builds a knowledge base from an n8n Form trigger: receives a source title and free-text content as knowledge-base entries.
• Creates embeddings and a vector store: splits submitted content into chunks, generates OpenAI embeddings, and stores vectors in an in-memory vector store keyed as company_kb.
• Runs a public support chat endpoint: customers submit support questions via a chat trigger.
• Retrieval-Augmented Generation (RAG) with Claude: the Claude agent searches the company_kb vector store and responds using retrieved passages, with short-term conversation memory.
• Escalates when needed: if the knowledge base can’t answer (or a user requests a human), the workflow sends an escalation email via Gmail with the question and context.
• Logs everything for operations: records the question, Claude’s answer, session ID, and timestamp to an n8n Data Table (recommended name: support_chat_log).

Use cases

• Customer support teams who want Claude-powered answers grounded in internal docs (no guessing).
• SaaS operators managing growing documentation and needing reliable escalation to a support inbox.
• Automation engineers deploying an n8n RAG support chatbot with auditable chat logs.

Technical details

• Integrations/tools: OpenAI (Embeddings), Anthropic (Claude), Gmail OAuth2, n8n Data Table.
• n8n nodes include: Form trigger, LangChain Agent (n8nn8n-nodes-langchainagent), Chat trigger (n8nn8n-nodes-langchainchat trigger), Gmail tool, and supporting nodes (e.g., Data Table, Sticky Note).
• Target vector store key: company_kb.