This n8n workflow automates the process of importing, indexing, and querying documents received via WhatsApp, leveraging OpenAI’s language models and vector search for intelligent responses. It starts by listening for messages on WhatsApp, then processes various media types such as text, images, audio, and documents. The workflow downloads media, extracts content from PDFs and spreadsheets, and imports Google Docs for comprehensive data ingestion. Text and image contents are fed into AI nodes for generating prompts and embeddings, which are then stored in a MongoDB vector database for fast similarity searches.
When users send a message or upload files through WhatsApp, the workflow identifies the media type, retrieves the content URL, and processes it accordingly. Texts are split into chunks for better analysis, and embeddings are created using OpenAI. The system can handle multiple file formats, including CSV, HTML, calendar files, PDFs, XLS, and XLSX, routing them through appropriate processing nodes. An embedded knowledge base using GPT-4o-mini provides context-aware answers, while conversation memory maintains context over sessions.
This setup is ideal for customer support, knowledge base querying, or internal documentation assistance, enabling real-time, AI-driven responses to user inquiries about uploaded documents or external content. The use case is particularly valuable for organizations needing efficient document access and intelligent query resolution via WhatsApp.
Reviews
There are no reviews yet.