WhatsApp RAG Enterprise

Who is this for?

WhatsApp RAG n8n workflow for support leads sitting on 40 repeat chats, voice notes and PDFs that you run manually from the editor. Answers live in Google Docs, Drive and MongoDB while agents retype replies in WhatsApp. You import the dg-0013 JSON slug-specific with exact credentials, tested first-try, and Pinecone plus MongoDB vector wiring is ready to verify.

What problem does this solve?

You read each WhatsApp thread, open Google Docs and Drive to find the answer, then retype the reply and re-upload attachments by hand. On manual run, Google Docs Importer and MongoDB Vector Store Inserter sync embeddings through Embeddings OpenAI into MongoDB Vector Search, and Knowledge Base Agent drafts via OpenAI Chat Model. You just approve drafted replies and handle exceptions.

What this workflow does

1. Knowledge Base Agent (AI Agent)
2. OpenAI Chat Model (OpenAI)
3. Embeddings OpenAI (OpenAI)
4. MongoDB Vector Search (MongoDB)

How it works

You run it manually from the n8n editor → Google Docs Importer, MongoDB Vector Store Inserter, into Knowledge Base Agent with MongoDB Vector Search retrieval.

Setup

1. Open danielgrey.shop/app/templates/dg-0013 → Copy JSON → n8n → Import from Clipboard (or Push to n8n)
2. Connect credentials: Google AI, Google Docs, Google Drive, HTTP Header Auth, MongoDB, OpenAI, Pinecone, WhatsApp
3. Run once manually to verify each branch — Time required: ~10 min review — then Activate.

Requirements

- n8n 1.50.0 or later
- Credentials for Google AI, Google Docs, Google Drive, HTTP Header Auth, MongoDB, OpenAI, Pinecone, WhatsApp
- One-time $199 — yours forever, Pro unlocks it too.

Customising

- Tune retrieval thresholds in Knowledge Base Agent and chunk size feeding MongoDB Vector Search.
- Update Map file extensions for your Drive file types and folder IDs in Google Docs Importer.