Back to projects
AI Agents
AI WhatsApp RAG Chatbot
An AI-powered WhatsApp assistant that answers questions from company documents using RAG, vector search, automated document ingestion, and conversational memory.
Business Problem
Businesses repeatedly answer customer questions using information stored across documents, schedules, policies, and internal knowledge files.
Solution
Built two connected automation workflows: one automatically ingests and embeds company documents into a vector database, while the second handles WhatsApp conversations and retrieves relevant knowledge before generating grounded responses.
Workflow
- 01Google Drive Documents
- 02Automatic Document Ingestion
- 03Gemini Embeddings
- 04Supabase Vector Store
- 05WhatsApp Message
- 06AI Agent
- 07RAG Retrieval
- 08Gemini Response
- 09WhatsApp Reply
Key Features
- Automated document ingestion from Google Drive
- Vector search over company knowledge
- WhatsApp conversational responses with RAG
- Grounded replies based on retrieved documents
Tech Stack
n8nWhatsApp Business APIGoogle GeminiSupabaseRAGVector SearchGoogle DriveLangChainAI Agents
Architecture / System Flow
One n8n workflow ingests and embeds documents into Supabase, while a second workflow handles WhatsApp messages with RAG retrieval and Gemini responses.
Resources