OpenAI + Pinecone RAG for Social & Email Sales Replies (n8n)

Turn Social & Email Inquiries into Grounded, Ready-to-Send Replies with OpenAI + Pinecone (n8n)

This n8n workflow automatically syncs your Google Sheets knowledge base into Pinecone, then uses OpenAI-powered RAG to draft accurate, context-grounded sales replies for Instagram/Facebook, TikTok, and inbound email—posting the response back to the original channel and logging every interaction to Google Sheets.

What this workflow does

• Builds a Pinecone knowledge index: Manually pulls offerings/FAQs from Google Sheets, chunks each row, generates OpenAI embeddings, and upserts vectors with metadata (including buy_link) into a Pinecone index.
• Receives new customer inquiries: Listens for Instagram/Facebook events via a Meta webhook and normalizes comments/DMs into a common interaction schema.
• Handles TikTok too: Receives TikTok comment events via webhook or polls for new messages via IMAP email, acknowledging events and normalizing them into the same interaction schema.
• Generates replies with RAG: Embeds the incoming inquiry, retrieves the top matching chunks from Pinecone, and assembles a context block including any buy_link metadata.
• Returns a structured, grounded answer: Uses an OpenAI chat model to produce a structured JSON reply that stays within retrieved context and flags buying intent when appropriate.
• Includes checkout CTA when relevant: Appends the buy link CTA when buying intent is true, sends the reply via the correct channel (Meta Graph API, TikTok API, or SMTP email), and logs the full interaction to Google Sheets.
• Alerts on failures: Sends an email alert via SMTP when any workflow execution errors occur.

Use cases

• Auto-respond to Instagram/Facebook comments and DMs with product-specific answers drawn from your Sheets.
• Reply to TikTok comments with accurate FAQs and the right checkout link for interested buyers.
• Draft and send inbound sales email replies grounded in your catalog, while keeping a complete log in Google Sheets.

Technical details

• RAG stack: OpenAI embeddings + OpenAI chat model + Pinecone vector retrieval
• Integrations: Google Sheets (KB + logging), Meta webhook + Meta Graph API, TikTok webhook/API, IMAP email polling, SMTP email
• n8n nodes mentioned: code, merge, switch, webhook, email send, sticky note