AI Automation
RAG Pipeline
An intelligent document Q&A pipeline built on n8n. Google Drive acts as the knowledge source — new files are automatically embedded with Cohere and indexed in Pinecone. Users interact via a chat trigger, with Groq LLM generating context-aware answers backed by simple conversation memory.
Problem
Teams often store critical knowledge in Google Drive documents that are hard to search or query. Manual lookup is slow, and important context gets buried across hundreds of files.
Solution
Built a fully automated RAG (Retrieval-Augmented Generation) pipeline using n8n. When a new document is added to Google Drive, it is automatically chunked, embedded via Cohere, and stored in Pinecone. A chat trigger then allows users to ask questions — the pipeline retrieves relevant chunks, maintains conversation memory, and generates accurate answers using Groq LLM.
How It Works
- 01Google Drive trigger fires when a new document is added.
- 02Document is chunked and sent to Cohere for vector embeddings.
- 03Embeddings are stored and indexed in Pinecone vector database.
- 04User sends a query via the chat trigger.
- 05Query is embedded and matched against Pinecone for top-k relevant chunks.
- 06Simple memory appends recent conversation context.
- 07Groq LLM generates a grounded, context-aware response.
Key Features
- 📂 Google Drive trigger (auto-ingest)
- 🧠 Cohere embedding model
- 📌 Pinecone vector database
- 💬 Chat trigger interface
- 🗂️ Simple memory (conversation context)
- ⚡ Groq LLM for fast responses
- 🔄 n8n workflow orchestration
- 🔍 Semantic document search
Project Status
Prototype / Production-Ready MVP
Designed as a reusable knowledge-base automation for teams, support systems, and internal documentation workflows.